Jump to ContentJump to Main Navigation
Show Summary Details
More options …


Official Journal of the Society to Improve Diagnosis in Medicine (SIDM)

Editor-in-Chief: Graber, Mark L. / Plebani, Mario

Ed. by Argy, Nicolas / Epner, Paul L. / Lippi, Giuseppe / Singhal, Geeta / McDonald, Kathryn / Singh, Hardeep / Newman-Toker, David

Editorial Board: Basso , Daniela / Crock, Carmel / Croskerry, Pat / Dhaliwal, Gurpreet / Ely, John / Giannitsis, Evangelos / Katus, Hugo A. / Laposata, Michael / Lyratzopoulos, Yoryos / Maude, Jason / Sittig, Dean F. / Sonntag, Oswald / Zwaan, Laura

CiteScore 2018: 0.69

SCImago Journal Rank (SJR) 2018: 0.359
Source Normalized Impact per Paper (SNIP) 2018: 0.424

See all formats and pricing
More options …

Diagnostic Error in Medicine 7th International Conference

Merging Policy, Practice and Technology: Paths to Improve Diagnosis

Published Online: 2015-01-24 | DOI: https://doi.org/10.1515/dx-2015-0001

These abstracts have been reproduced directly from the material supplied by the authors, without editorial alteration by the staff of this Journal. Insufficiencies of preparation, grammar, spelling, style, syntax, and usage are the authors’.

14–17 September 2014, Atlanta, Georgia

Planning Committee Members

Paul L. Epner, MBA, MEd (Chair)

Hardeep Singh, MD, MPH (Co-Chair)

Karen Cosby, MD; Robert El-Kareh, MD, MPH, MS; Devery Howerton, PhD, MS;

Kathryn McDonald, MM, MBA; David L. Meyers, MD, FACEP; Geeta Singhal, MD, Med;

Robert A. Swerlick, MD, FAAD; Lorri Zipperer, MA, Laura Zwaan, PhD


Failure to recognize red flags and delays in diagnosis of spinal epidural abscesses

A Al-Mutairi1, V Bhise1, A Meyer1, D Murphy1, L Wei1, E Russo1, H Singh1.

1 Baylor College of Medicine, Houston, TX

Background: Spinal epidural abscesses (SEA) are uncommon, but if missed can lead to major harm. Most patients with a SEA have back pain at some point during disease evolution. However, because back pain is very common and SEAs are infrequently the reason, providers rely on associated red-flags to avoid missing the diagnosis of SEA. In this study, we determined the frequency and types of diagnostic errors associated with SEA.

Methods: Retrospective chart reviews for all patients who had primary care visits for back pain and had subsequent diagnosis of SEA (based on ICD-9 code) between 2009 and 2010 at a large tertiary-care VA facility. Two physicians independently reviewed each record to identify the presence or absence of “red flags” (based on American College of Radiology guidelines, such as unexplained weight loss, neurological deficits, and fever) associated with back pain. Reviewers also assessed whether action was taken in response to these red flags, such as ordering imaging studies. We defined diagnostic errors as missed opportunities to make a correct or timely diagnosis of SEA that occurred when no additional action or evaluation was undertaken despite the presence of SEA-related red flag. Reviewers also collected the dates of each patient’s first provider encounter for back pain where red-flags were present. T-tests were used to compare the number of red flags and the number of comorbidities for patients with and without diagnostic errors.

Results: Of 43 cases of SEA reviewed so far, 25 cases were either diagnosed elsewhere and transferred only for treatment, miscoded as having SEA or did not have a new SEA diagnosis. In the 18 remaining SEA cases, all had primary care visits with back pain prior to diagnosis. Of these, both reviewers identified diagnostic errors in 10 records (kappa 1.0). The average number of red-flags for the two groups was comparable (2.7 in error group vs. 2.0 in non-error group; p =.18) and while the error group had more comorbidities, it was not statistically significant. (2.9 vs 1.5, p=.09). Fever and neurological deficit were among the most commonly missed red flags.

Conclusions: Many patients with SEA had red flags that were missed leading to delays in diagnosis and treatment. Using an objective data collection instrument, we achieved a high reliability for diagnostic error determination. Rigorous definitions, such as the one used here, are needed to determine presence or absence of diagnostic errors to improve measurement in this area.

Electronic triggers to detect missed opportunities in test result follow-up

A Al-Mutairi1, A Meyer1, D Murphy1, D Sittig2, L Wei1, E Russo1, H Singh1.

1 Michael E. DeBakey Veterans Affairs Medical Center and Baylor College of Medicine, Houston, TX

2 University of Texas, Houston, TX

Background: Delays in follow-up of abnormal test results (missed results) can lead to patient harm. EHR-based triggers (electronic signals for detecting likely adverse events in an electronic health record [EHR]) have the potential to identify patients whose results may have been missed. We sought to develop EHR-based triggers to identify missed opportunities in follow-up of abnormal thyroid stimulating hormone (TSH) tests in outpatients with hypothyroidism on active thyroid replacement therapy.

Methods: We used current diagnosis and treatment guidelines for hypothyroidism to develop electronic triggers to identify cases where abnormal TSH was missed in patients with hypothyroidism. Triggers were developed and applied on Veteran Affairs Informatics and Computing Infrastructure (VINCI), a large repository of national VA data. The trigger algorithm extracted records lacking pre-defined appropriate and timely follow-up actions within 4 weeks of an abnormal TSH result (defined as >10mU/mL) in patients actively taking thyroid replacement therapy. The trigger excluded patients who had computer-based evidence of timely follow-up actions (e.g., new medication orders or repeat TSH testing) and TSH ordered on inpatients. A physician reviewed 20 random records to verify the accuracy of our exclusion criteria. Records that were trigger positive were reviewed to calculate the trigger’s positive predictive value (PPV) as defined by the number of true positive cases identified by chart review divided by total number of records identified by the trigger. We conducted preliminary local testing of the trigger and then applied it to a network of 7 VA facilities.

Results: We applied the trigger algorithm to the records of all patients in the VA network between January 1 and December 31, 2011. Of 85 trigger positive records identified and reviewed by a physician-reviewer, 56.4% (48/85) were found to have missed opportunities in follow-up. In the remaining records, 37 had documentation of follow-up efforts in addressing abnormal TSH levels (e.g., provider documented non-adherence or failure to take medication as prescribed). We are currently applying the trigger nationally to all VA facilities.

Conclusions: We developed an EHR-based trigger to identify missed opportunities in follow-up of outpatient TSH tests. Our trigger program can be run nationally and is potentially exportable to other EHR systems for measurement and monitoring of missed opportunities in follow-up of several types of diagnostic tests. Organizations should consider developing robust quality assurance programs to ensure that the triggered information is acted upon and patients are prevented from falling through the cracks of the health system.

Diagnostrial plus: a prospective study of diagnostic accuracy and concordance

L Alonso1, J Lora1, A Cuenca1, M Costalago1.

1 Hospital Clinico Universitario de Málaga, Málaga, Spain

Background: Diagnostic accuracy (DA) is a basic condition for planning a correct treatment in the Internal Medicine field. No study or clinical trial have shown a practical influence to improve DA, but an Electronic Diagnostic Reminder Tool (EDRT), a good differential and a reflective period seems to be logical tools for improvement.

Methods: We evaluated in an Internal Medicine Department the initial DA for clinicians in a first cohort of 24 patients, using as gold standard the final diagnosis obtained after active follow-up. At the same time we used an EDRT, Isabelhealthcare®, and we considered a positive result if the final diagnosis was included in the print-out list of the EDRT. In a second phase we inform to the clinicians about their results of DA in the first cohort and they were more confident using Isabelhealthcare®, then we implemented a second phase of the study collecting 26 patients, with a total of 50 patients. Eleven patients were excluded previously. The cohort included patients of both sex with constitutional and general symptoms not related with previous diagnoses. Percentages were used to describe DA and the Kappa index was used as an index of agreement between the clinician and the EDRT.

Results: In the first cohort of 24 patient the DA was 62% and 75% for clinicians and the EDRT respectively; kappa index was very low, 0.090, showing a bad concordance and a different operational behaviour. In the second cohort of 26 patients the DA was similar to the first, 61% and 73%, but kappa value was 0.56, showing a better operational agreement between clinicians and the EDRT. For the complete series of 50 patients the DA was 60% and 74% respectively with a kappa index of 0.33.

Conclusions: Our DA of 60% is similar to values described in the literature. Kappa value improved between the first and the second cohort due in our opinion to a “learning curve” for clinicians. This improvement was not followed by a better DA but there was a reduction in the percentage of cases no diagnosed by the clinicians but suggested by the EDRT from 29% to 15%. Our main conclusions are that EDRT have a learning curve and that overconfidence is an important bias because clinicians are reluctant to change their initial clinical differential, even if they see a suggested diagnosis in a printed list.

Can administrative data-based triggers provide insights on missed opportunities in diagnosis of outpatient deep vein thrombosis?

V Bhise1,2, A Wright3, D Sittig4, H Singh1.

1 Michael E. DeBakey Veterans Affairs Medical Center and Baylor College of Medicine, Houston, TX

2 University of Texas School of Public Health, Houston, TX

3 Brigham & Women’s Hospital, Boston, MA

4 University of Texas, Houston, TX

Background: Deep Vein Thrombosis (DVT) is a relatively common condition affecting nearly 1 in 1000 adults, and its timely diagnosis and treatment is essential to avoid harm from complications such as pulmonary embolism. We previously used “trigger” methodology in electronic health records to identify diagnostic errors but such record reviews are fairly resource intensive. In this study, we explored the use of an administrative claims database to determine if it can suggest presence of potential missed opportunities in diagnosis in outpatient settings, using DVTs as a case example.

Methods: We conducted a preliminary retrospective cross-sectional analysis of administrative claims data of patients insured by a large private carrier in Texas. We identified patients admitted with a new diagnosis of DVT (defined as no previous billing code for DVT and not on warfarin) in years 2008-2012 who also had an office visit or Emergency Room (ER) visit in the previous 30 day period prior to the diagnosis. We defined the presence of “potential missed opportunities” when any lower limb morbidity such as swelling or pain was present at the earlier visit. We used ICD-9 CM and CPT codes to identify visits, lower limb medical conditions and DVT and conducted descriptive analysis to determine the presence of potential missed opportunities. STATA 12 was used for analysis.

Results: There were 9,569 inpatient admissions with DVT diagnosis in the study period. Of these, 5,670 (59.25%) and 2,079 (21.73%) were associated with one or more outpatient visits and ER visits, respectively, in the 30 day period prior to diagnosis. Lower limb swelling was present in 454 (8.01%) of previous outpatient visits and 193 (9.29%) of ER visits, respectively, and lower limb pain was present in 443 (7.81%) of previous outpatient visits and 244 (11.74%) of ER visits, respectively.

Conclusions: Preliminary retrospective analysis of administrative claims data suggests presence of potential missed opportunities to diagnose DVT in outpatient settings. However, our pilot work also suggests the need for further exploration on how administrative data can be effectively used to evaluate for real missed opportunities in diagnosis. Use of “smarter” algorithms with more clinical data and clinical validation with medical records review could improve validity. Because integrated EHRs with complete longitudinal patient data across the care continuum are still years away for most US populations, once refined, this type of methodology could be used to screen for high-risk records for further outcomes review.

Using a publicly available dataset to understand uncertainty in outpatient diagnosis

V Bhise1,2, H Singh1.

1 Michael E. DeBakey Veterans Affairs Medical Center and Baylor College of Medicine, Houston, TX

2 University of Texas School of Public Health, Houston, TX

Background: Undifferentiated symptoms are common in outpatients, and often lead to diagnostic uncertainty. However, little is known about diagnostic uncertainty in clinical practice. We hypothesized that certain types of outpatient visits are associated with greater degrees of diagnostic uncertainty and hence warrant additional measures (e.g. close follow-up) to prevent possible harm from errors or delays. We used publicly available data to estimate uncertainty and determined potentially associated patient and visit-related factors.

Methods: We conducted a retrospective cross-sectional analysis of publicly available 2010 National Ambulatory Medical Care Survey(NAMCS) outpatient physician visit data. We defined two visit cohorts to test our hypothesis (“potentially high risk” and “potentially low risk” for diagnostic uncertainty) based on a scoring system we developed using certain patient characteristics (new patient; non-white race) and visit variables (new problem during visit; abnormal temperature). Uncertainty was defined to occur when providers self-reported probable or questionable diagnosis. We compared the probability of being uncertain in high vs. low risk cohorts and explored differences in visit dispositions (ER referrals; other physician referrals, inpatient admission and return appointment). We also conducted multivariate logistic regression to identify factors associated with uncertainty.

Results: Of the 31,229 outpatient visits, 569 (1.82%) were reported as uncertain diagnosis. More than half (17,813; 57.04%) were classified as high-risk. Of these, 5,037 (28.28%) were new patients, 9,813 (55.09%) presented with a new problem, 9, 052 (50.82%) were non-white and 1,941(10.90%) presented with an abnormal temperature. Probability of uncertain diagnosis was higher in the high-risk cohort (OR, 1.87; 95% CI, 1.55-2.25). Presenting as a new patient (OR, 1.23; 95% CI, 1.01-1.51) or with a new problem (OR, 2.56; 95% CI, 2.15-3.05) was more likely to be associated with uncertainty, while being non-white was less likely (OR, 0.65; 95% CI, 0.53-0.80). High-risk patients were also more likely to be referred to another physician (OR, 1.48; 95% CI, 1.36-1.60), or admitted to ER or hospital (OR, 1.58; 95% CI, 1.18-2.10) but less likely to receive a follow-up appointment (OR, 0.50; 95% CI, 0.48-0.53).

Conclusions: Although preliminary analysis suggests that diagnostic uncertainty is likely more frequent with new patients or those with new problems, it might be grossly under-reported due to lack of an appropriate ICD-9 code. Data also suggests further need to evaluate follow-up patterns for certain high-risk patients. Publicly available datasets might be useful to explore possible relationships between patient/visit characteristics and uncertainty in diagnosis, and generate further hypothesis for future research on diagnostic decision-making.

Race and gender disparities in ocular fundus examination

B Bruce1, P Thulasi1, D Wright1, V Biousse1, N Newman1.

1 Emory University, Atlanta, GA

Background: The Institute of Medicine’s report Unequal Treatment: Confronting Racial and Ethnic Disparities in Health Care found that blacks and Hispanics tend to receive lower quality healthcare even after controlling for access-related factors, such as insurance status. Beyond disparities that may result from systematic issues and health care seeking behavior, they reported evidence of racial disparities at the level of the clinical encounter. The report suggested that the typical uncertainty of any clinical encounter, especially when compounded by significant time constraints and high-pressure decision making (such as in the emergency department [ED]) combined with patients from different racial and ethnic backgrounds may unconsciously release physicians’ biases and stereotypes in a way that affects patient encounters.

Methods: The 704 patients enrolled in the Fundus photography vs. Ophthalmoscopy Trial Outcomes in the Emergency Department (FOTO-ED) study were included. In the FOTO-ED study, adult patients who presented to our university emergency department with headache, focal neurologic deficit, diastolic blood pressure ≥120 mmHg, or acute visual change were enrolled. Non-mydriatic fundus photographs were obtained and provided to ED physicians during clinical care. Direct ophthalmoscopy was also available. Fundus photographs were examined for optic disc edema, isolated intraocular hemorrhage, grade III/IV hypertensive retinopathy, retinal vascular occlusion, and optic disc pallor. Logistic regression modeling controlling for presenting complaint/condition, examination technique, age, and mean arterial blood pressure were employed to evaluate for evidence of race and gender differences in examination frequency.

Results: Men with focal neurologic deficits were half as likely as women to have their ocular fundus examined (OR 0.49; 95%CI: 0.25-0.99; p=0.047) controlling for age, body mass index, mean arterial pressure, and race, even though they had a similar frequency of fundus abnormalities (6/84=7.1%) to women (12/140=8.6%). Black patients with visual complaints were less than half as likely as patients of other races to have their ocular fundus examined (OR=0.39; 95%CI: 0.15-1.08; p=0.056) despite having twice the risk of an abnormality (27% vs. 13%, p=0.03).

Conclusions: Disparities in ocular fundus examination frequency were seen in this high-risk population based upon complaint, gender, and race. Our findings suggest that an exploration of gender and racial disparities in the examination of the ocular fundus is warranted and that additional education of ED physicians about the epidemiology of ocular fundus abnormalities may be needed to heighten awareness about which patients are at greatest risk of ocular abnormalities and to avoid diagnostic errors.

Differential diagnosis generators: a systematic review of their efficacy and utility

S Cheraghi-Sohi1, N Riches1, M Panagioti1, R Alam1, S Campbell1.

1 University of Manchester, Manchester, United Kingdom

Background: Rates of diagnostic error have been estimated at 10-15% in most areas of clinical medicine and it has been estimated that such errors are responsible for 63-66% of medicolegal claims against primary care doctors. Consequently diagnostic error has been identified as a priority area for the NIHR Manchester Primary Care Patient Safety Centre. Differential diagnosis (DDX) generators are electronic tools which allow users to input two or more clinical features to generate a differential diagnosis list in cases of diagnostic uncertainty. Although originally validated in paediatric and hospital medicine, their efficacy in primary care is now being explored. No systematic review of the literature has yet been performed in this area. We have conceptualised the two most important characteristics of DDX generators to be accuracy and utility. Utility describes features which will predict uptake amongst clinicians, such as time taken to use and number of diagnoses generated.

Methods: We conducted a systematic review to determine the accuracy and utility of DDX generators in clinical medicine. We searched Medline, Embase and CINAHL from inception, in addition to some specialist databases. 4016 articles were retrieved, 3929 were excluded by title and abstract screening resulting in 87 articles eligible for full text screening. Data is currently being extracted from the 47included studies for the main accuracy and utility outcomes.

Results: A scoping study suggested that there is significant heterogeneity of published articles and so it is anticipated that a narrative synthesis of the evidence will be performed which will be ready for presentation at the time of the conference.

Conclusions: DDX tools have the potential to reduce diagnostic error in primary care. This review will provide evidence about the effectiveness of DDX tools and will help justify whether there is a need to conduct further exploratory work on the characteristics of DDX tools in in clinical medicine including primary care.

Diagnostic errors in a pediatric intensive care unit: insights from the morbidity and mortality conference

C Cifra1, K Jones2, J Ascenzi2, U Bhalala2, M Bembea2, D Newman-Toker2, J Fackler2, M Miller2.

1 University of Iowa, Iowa City, IA

2 Johns Hopkins University, Baltimore, MD

Background: Diagnostic errors are an important yet understudied cause of health care-associated harm caused by both cognitive- and system-related factors. The Pediatric Intensive Care Unit (PICU) presents a unique environment to study diagnostic errors, as the complex and high-acuity nature of critical care combines with the particular diagnostic challenges of pediatrics. We offer a first look at diagnostic errors in the PICU primarily through the morbidity and mortality conference (MMC), as we aim to describe the characteristics of these errors (modified Goldman classification, medical category, and severity), identify possible causes (contributing factors and preventability), and pinpoint where problems arise in the diagnostic process.

Methods: We performed a retrospective record review of MMC agendas, patient charts, and autopsy reports in a single tertiary referral PICU, including 96 patients discussed at the PICU MMC over a 14-month period (November 2011 to December 2012). Categories within diagnostic error characteristics were compared using one-sample test of proportions for binomial variables and multinomial goodness-of-fit for multinomial variables.

Results: Eighty-nine of 96 patients (93%) discussed at the PICU MMC had at least one identified safety event. A total of 377 safety events were identified. Twenty (21%) of patients had identified diagnostic errors, comprising 5.3% of all safety events. Out of 20 total diagnostic errors identified, 35% were discovered at autopsy while 55% were reported primarily through the MMC. Almost all diagnostic errors (95%) had an impact or could have had a potential impact on patient survival or safety (p = 0.001). Forty percent of errors did not cause actual patient harm, but 25% were severe enough to have potentially contributed to death (40% no harm vs. 35% some harm vs. 25% contributed to death, p = 0.189). Half of all diagnostic errors were preventable (50%, p = 0.062). There were slightly more system-related factors (40%) that solely contributed to diagnostic errors compared to cognitive factors (20%), however 35% had both system and cognitive factors that played a role (40% vs. 20% vs. 35%, p = 0.238). Most errors involved vascular (35%) followed by neurologic (30%) events.

Conclusions: Diagnostic errors in the PICU are not uncommon and cause patient harm. The MMC helps to identify diagnostic errors outside of the traditional autopsy. Most errors are preventable by targeting both cognitive- and system-related contributing factors. Prospective studies are needed to further determine how and why diagnostic errors occur in the PICU, and what interventions would likely be effective for prevention.

Application of a “diagnostic error evaluation tool” to reliably detect diagnostic errors in the pediatric intensive care unit

M Davalos1, K Samuels1, S Thammasitboon1, M Sur1, K Roy1, A Meyer2, H Singh2.

1 Texas Children’s Hospital, Houston, TX

2 Michael E. DeBakey Veterans Affairs Medical Center and Baylor College of Medicine, Houston, TX

Background: A structured and reliable approach to detect the occurrence of diagnostic errors is essential for their measurement and reduction. However, methods to help identify and reach consensus about the presence of a diagnostic error are underdeveloped. We used a recently developed data collection instrument, the “Diagnostic Error Evaluation Tool”, to evaluate if it can reliably detect the occurrence of diagnostic errors in patients admitted to a Pediatric Intensive Care Unit (PICU).

Methods: Using expert input, a multidisciplinary team iteratively developed and refined a structured tool that included twelve questions to evaluate various aspects of the diagnostic process and a final question determining if a diagnostic error occurred. After piloting the tool in a random selection of 90 PICU cases, we used it to assess for diagnostic errors in a purposeful sample of PICU medical records. The sample included two patient cohorts of PICU patients admitted between 6/2013 and 12/2013 that were selected because of the likelihood of finding error cases. Cohort 1 consisted of patients who had died and had autopsies. Cohort 2 consisted of patients who were evaluated at any outpatient location within 2 weeks prior to PICU admission. Two clinicians used the tool to independently review records. Raw percentages of agreement and weighted Kappas were calculated to measure inter-rater agreement.

Results: In cohort 1 (n=16) both investigators independently agreed on the presence of 2 diagnostic errors, whereas in cohort 2 (n=41), they agreed on one. In cohort 1, agreement was 70.3% (weighted kappa=0.75) for the 12 questions about various aspects of the diagnostic process involved and 93.8% for determination of presence/absence of error (weighted Kappa=0.76). In cohort 2, agreement was 86.2% (weighted Kappa=0.89) for the 12 diagnostic process questions and 100% (weighted Kappa=1.0) for presence/absence of error. Combining both cohorts (n=57), agreement was 81.7% (weighted Kappa=0.85) for the diagnostic process questions and 98.3% (weighted Kappa=0.85) for presence/absence of diagnostic error.

Conclusions: The preliminary use of a structured “Diagnostic Error Evaluation Tool” led to high record-review agreement for presence/absence of diagnostic error. Further validation could identify which questions on the tool are most valuable and lead to more reliable and efficient methods for measuring diagnostic errors in PICU and possibly elsewhere. Adoption and implementation of practical tools could also inform which diagnostic process breakdowns should be targeted in quality initiatives.

Feedback of patients’ outcomes for physicians in training: are those who need it most getting it the least?

S Feupe1, R El-Kareh1.

1 University of California, San Diego, La Jolla, CA

Background: Physicians need to learn the outcomes of their diagnostic and therapeutic decisions in order to properly calibrate their decision-making. However, this outcome feedback is often inadequate, leading to many missed learning opportunities and potentially dangerous diagnostic errors that may be repeated rather than avoided in the future. We sought to use electronic health record (EHR) chart access logs to determine how often emergency medicine (EM) and internal medicine (IM) residents follow up on their hospitalized patients after handing their care off to other providers and to describe factors related to this follow-up.

Methods: We studied the chart access within a comprehensive commercial EHR of 26 EM and 54 IM residents involving patients hospitalized at UC San Diego Heath System between December 16, 2012 and April 30, 2013. We identified handoffs where the primary care responsibilities were transferred from one physician to another. We generated descriptive statistics regarding the earliest post-handoff chart access relative to physician, patient and encounter characteristics. We performed logistic regression analysis to model resident re-access within 14 days of handoff controlling for patient age and gender, resident department, year of training, rotation assignment, gender, whether laboratory or radiology tests were pending at the time of handoff and whether handoffs occurred on a weekend or off hours.

Results: We analyzed records for 7750 handoffs involving 2715 patients during the study period. Overall, residents re-accessed patient charts within 14 days following handoffs in 129/4390 (3%) cases for EM residents and 1798/3360 (54%) cases for IM residents. Compared with residents on daytime IM ward rotations, re-access of charts occurred significantly less frequently for EM residents (OR 0.04, 95% CI 0.02-0.07) and IM residents on night rotations (OR 0.02, 95% CI 0.01-0.03) with p<0.001 for overall differences among rotations. First-year EM residents (interns) were significantly more likely to re-access charts compared with higher-year EM residents (OR 6.39 95%CI 4.19-9.74, p<0.001). However, IM interns were significantly less likely to re-access charts than higher-year IM residents (OR 0.67 95%CI 0.56-0.79, p<0.001).

Conclusions: We found that EM residents and IM residents who admitted patients overnight were least likely to re-access patient charts within 14 days following handoffs of patient care. Because a large proportion of initial medical decision-making is performed by EM and night IM residents, they would benefit from feedback of the results of their decisions. However, at our institution, these residents do not use the EHR to obtain this feedback frequently.

Using interactive voice response technology coupled with a pharmacist intervention to detect adverse drug reactions

J Figueroa1, A Salazar1, E Klinger1, J Haas1, D Bates1, G Schiff1.

1 Brigham & Women’s Hospital, Boston, MA

Background: Delayed or failed recognition of adverse drug reactions (ADRs) is a common problem and a major contributor to patient harm. As part of an AHRQ-funded Centers for Education and Research on Therapeutics (CERT) study (grant # 5U19HS021094), we conducted a clustered-randomized control trial (RCT) to evaluate the effectiveness of a strategy to decrease delays and missed detections of ADRs.

Methods: The study deployed interactive voice response (IVR) technology to survey primary care patients newly started on oral agents for four common conditions (hypertension, diabetes, depression, and insomnia) to screen for common side effects. Patients were recruited from 14 primary care clinics over a 12-month period. Patients reporting symptoms on IVR calls were contacted by a pharmacist who performed an evaluation of their symptoms to determine the likelihood of an ADR. An initial analysis of the first 146 completed IVR calls was made and will be reported here.

Results: In summary, 23/146 (15.8%) of reported symptoms were thought to be unlikely related to newly prescribed medication, 71/146 (48.6%) were possibly related to new medication, and 52/146 (35.6%) were most probably related to new medication. The severity of the reported symptoms was categorized as mild in 80/146 (54.8%), significant in 63/146 (43.2%), serious in 1/146 (0.7%), and life threatening in 2/146 (1.3%). A temporal relationship of symptoms with initiation of prescribed medications was established in 95/146 (65%), was not present in 21/146 (14.4%), and was difficult to establish in 28/146 (19.2%). Furthermore, 6/146 (4.1%) of patients were taking a different dose than the initially prescribed dose and 34/146 (23.3%) of patients were no longer taking the prescribed medication; 16/34 (47.1%) were stopped by a physician while 18/34 (52.9%) were self-discontinued by patients, of which 0/18 (0%) of the prescribing physicians were aware at time of IVR call. Out of the 112/146 patients still taking the prescribed medication, a separate physician evaluation of IVR calls was performed and recommend the following: consideration of stopping the medication in 18/112 (16.1%) of cases, reducing the dose in 8/112 (7.1%) of cases, continuing medication in 39/112 (34.8%) of cases, while more information would be needed prior to taking action in 47/112 (42%) of cases.

Conclusions: This initial analysis shows how a pharmacist intervention coupled with IVR technology can be beneficial. This study aims to be one of the first RCT to show the effects of such an intervention on reducing missed and delayed diagnosis of ADRs.

A year in review: early diagnostic error & delay data from one intern’s year

J Liao1, G Schiff1.

1 Brigham & Women’s Hospital, Boston, MA

Diagnostic Error (DE) is common, but compared to other kinds of medical error (e.g., medication or process error) its prevalence and epidemiology is understudied. DE can occur at a number of points within the diagnostic process, and developing a better understanding of how these errors occur can improve patient safety and quality of care. To understand patterns/trends in DE, we created criteria for reviewing a consecutive series of all inpatients seen by PGY-1 physician during internship year. We tracked diagnosis across the following “time nodes” related to their internal medicine admission: a.) outpatient clinic visits within three months prior to that admission; b.) ED visit leading to admission; c.) inpatient stay d.) discharge from the admission; e.) clinic visits for the next 3 months following discharge. Patients were excluded if they had a short length of stay (<48 years); over half of patients (55% of 121 patients) were found to meet this exclusion criteria. In the eligible patients, there was evidence of DE for significant percentage of patients (20/55; 36%). Using the Diagnostic Error Evaluation and Research (DEER) taxonomy, we identified an average of 4.1 steps/instances of error or delay per case. The majority were in the “Assessment” (33%, 27/82) and “Lab and Radiologic Testing” (20%, 16/82) domains, with evidence of both medical underuse and overuse of testing. This data suggests that although DE are common and distributed throughout the diagnostic process, it may be possible to target interventions to specific areas or repeated failure Another important finding is the large numbers of patients excluded due to short stays on the intern’s service raising implications for diagnosis related to work hour restrictions.

Delay in the diagnosis of tuberculosis in health services in the city of Vitoria-es-Brazil

R Loureiro1.

1 Rio of Janeiro State University, Rio of Janeiro, Brazil

Background: In Brazil, since 2004 is adopted primary health care actions for TB control. The purpose of this study was to evaluate the delay in the diagnosis of tuberculosis (TB) in the municipality Vitoria-Espirito Santo-Brazil.

Methods: Cross-sectional study using the instrument of the Primary Care Assessment Tool (PCAT). The sample for convenience consisted of patients being treated for TB residing in Victoria, diagnosed between January and December 2009. Data were collected through semi-structured interviews and secondary sources of information (medical records).

Results: There was a predominance of males (59.4%), incomplete primary education (55.4%) and those living in their own house (93%), while 38.4% had no gainful activity, including 32.6% were unemployed, 4.9% and 0.9% of home student. Regarding the gateway was observed that the median time found, related to the patient, between the onset of signs and symptoms was 15 days, with 25% of patients took up to 5 days to seek health services and 75% of patients, this delay was 30 days. Observed even patients who reported consuming more than 60 days to seek health services. Furthermore, only 40.5% of health professionals, patients who met the first time they sought health services, diagnosed tuberculosis.

Conclusions: We concluded that there was a delay in diagnosis of TB due to delays in seeking care and errors in diagnosis by health professionals, suggesting greater attention to service these facts, which involve the implementation of strategies to reduce misdiagnosis or improve detection of TB in the city, and for the development of public health policies.

Diagnostic errors related to acute abdominal pain in the emergency department

L Medford-Davis1, E Park1, G Shlamovitz1, J Suliburk1, A Meyer1,2, H Singh1,2.

1 Baylor College of Medicine, Houston, TX

2 Michael E. DeBakey Veterans Affairs Medical Center, Houston, TX

Background: Unscheduled return visits to the Emergency Department (ED) within 72 hours could result from diagnostic errors during the initial visit. Abdominal pain is a frequent complaint associated with unscheduled return visits. We determined the frequency of diagnostic errors related to acute abdominal pain presentations to the ED and described the process breakdowns involved.

Methods: We conducted a retrospective chart review study of selected at-risk patients seen in the ED of an urban academic hospital. An electronic algorithm identified patients >18 years who met two criteria: 1) had chief complaint of abdominal pain recorded in the ED provider note followed by discharge home from the ED, and 2) return ED visit within 10 days of the initial visit leading to hospitalization. We manually excluded patients who presented with either chronic (>6 weeks) or traumatic abdominal pain. Two emergency physician “primary” reviewers independently determined presence/absence of error using a standardized data collection form. Disagreements were independently reviewed by a surgeon and an additional emergency physician. If secondary reviewers disagreed, the case was discussed among the primary and secondary reviewers and two diagnostic error experts to reach a team consensus. We used an existing 5-level taxonomy to categorize diagnostic process breakdowns.

Results: Of 99 cases reviewed, 33 had a diagnostic error (33.3%). Of the 33 error cases, over half had issues with follow-up and tracking (57.6%; n=19), most frequently related to follow-up of abnormal diagnostic test results (54.5%; n=18). Additionally, over half (51.5%; n=17) had errors in the patient-provider encounter such as failure to order sufficient tests for further work-up (33.3%; n=11), or problems gathering the history (21.2%; n=7). Twelve (36.4%) cases had issues related to diagnostic test performance/interpretation, the most frequent of which was misinterpretation of test results (33.3%; n=11). Additional breakdowns included failure to arrange proper follow-up (n=5, 15.2%) and patient-related issues such as failure to mention key symptoms or delay in seeking care (9.1%; n=3). The most frequently missed diagnoses were acute gallbladder pathology (n=10), urinary system infections (n=5), diverticulitis (n=2), small bowel obstruction (n=2), and ectopic pregnancy (n=2).

Conclusions: Using an electronic algorithm, we identified at-risk cases of abdominal pain in the ED and found that nearly a third had confirmed diagnostic errors. Process breakdowns most commonly involved follow-up and tracking of test results and issues with ordering additional tests in the patient-provider encounter. Our methodology might be useful for others to identify abdominal pain-related diagnostic errors in the ED.

Exploring the context of electronic health record (ehr)-based safe test result follow-up

S Menon1, M Smith1, D Sittig2, N Petersen1, S Hysong1, H Singh1.

1 Michael E. DeBakey Veterans Affairs Medical Center and Baylor College of Medicine, Houston, TX

2 University of Texas, Houston, TX

Background: Electronic health record (EHR)-based alerts can facilitate transmission of test results to healthcare providers, helping ensure timely and appropriate follow-up. However, failure to follow up on abnormal test results (missed test results) persists in EHR-enabled healthcare settings. We aimed to identify contextual factors associated with facility-level variation in missed test results within the Veterans Affairs (VA) health system. Thus far, organization- or facility-level information about test results management practices is poorly documented or understood, but this knowledge of local organizational context may be useful in understanding organization-wide vulnerabilities and may explain why some healthcare settings have fewer missed test results than other healthcare settings.

Methods: We used a mixed-methods approach to compare VA facilities deemed at higher and lower risk for missed test results on a variety of socio-technical variables. Based on a previous survey, we categorized VA facilities according to primary care providers’ (PCPs’) perceptions of low (n = 20) versus high (n = 20) risk of missed test results. We interviewed facility representatives to collect data on several contextual factors derived from a sociotechnical conceptual model of safe and effective EHR use. We compared these factors between facilities categorized as low and high perceived risk, adjusting for structural characteristics. We compared high and low perceived risk facilities on categorical variables using Fisher’s exact test statistics. To analyze open-ended responses, we used qualitative content analysis.

Results: A vast majority of facilities in both groups customized alert settings locally and required unread alerts to remain in the ordering provider’s inbox for at least 14 days. However, only about 70% of facilities had some mechanism to prevent alerts from remaining unread (unacknowledged). Facilities with low perceived risk were significantly more likely to use specific strategies to prevent alerts from being lost to follow-up. Qualitative analysis identified three high-risk scenarios for missed test results: alerts on tests ordered by trainees, alerts “handed off” to clinicians’ designees, and alerts on patients not assigned in the EHR to a PCP. Test result management policies and procedures to address these high-risk situations varied considerably across facilities.

Conclusions: In addition to implementing provider-level strategies to prevent missed test results, healthcare organizations would need to address several contextual factors that affect test result management. Several contextual sociotechnical issues we identified are generalizable to many healthcare institutions and pose a higher-risk for missed test results.

Workarounds indicate the need to optimize electronic health records for improving test results follow-up

S Menon1, D Murphy1, A Meyer1, H Singh1, D Sittig2.

1 Michael E. DeBakey Veterans Affairs Medical Center and Baylor College of Medicine, Houston, TX

2 University of Texas, Houston, TX

Background: Failure to follow-up test results leads to delays in diagnosis/treatment, and this continues to persist in electronic health records (EHRs)-based settings. Our previous qualitative work suggested that providers use alternate methods (or system workarounds) to bypass perceived or real barriers in their “electronic” workflow. In general, workarounds often lead to new inefficiencies and inaccuracies and their presence suggests that system-related enhancements are needed for ensuring safety. We determined the frequency and types of workarounds used by providers in EHR-enabled follow-up of test results.

Methods: We previously conducted a web-based survey of 5001 Veterans Affairs primary care providers to solicit input on several technical and non-technical issues related to test-result management. In this sub-analysis, we defined our outcome variable “use of workarounds” based on participants’ responses to a question asking whether they used alternate methods in addition to existing EHR to follow-up test results. Responses to a survey item about what methods, other than existing EHR, do providers use to help follow test results were coded to generate frequencies for types of workarounds. Predictor variables were derived from survey items related to provider perceptions about workload, volume of notifications (alerts) received in the EHR, organizational support, perceived effectiveness of the alert system, and personal history of missed test results. We used factor analysis to reduce relevant survey items into a set of 6 dimensions, which were entered in a multivariate logistic regression model to look for predictors of workarounds.

Results: 2554 (51%) respondents answered the question regarding use of alternate methods of which 43% (n=1104) reported using them. Most (92%) indicated that they use some form of paper-based method; paper lists/logs and sticky notes were the two most common methods, used by almost half (49%). 20% of respondents used a combination of paper- and computer-based workaround methods such as electronic calendars and tickler files. Multivariate logistic regression analysis revealed that providers who report having limited support to notify patients of test results, and providers who reported personal history of missing results or were aware of colleagues missing results, were more likely to use workarounds (p=.02 and p<.001, respectively).

Conclusions: Use of both paper and electronic workarounds to manage test results is common in EHR systems, suggesting the need to optimize EHRs for supporting fail-safe results follow-up. Because the EHR we studied is considered quite sophisticated and state-of-the-art, our findings also call for usability and functionality enhancements for other commercial EHRs.

* Patient-initiated second opinions for diagnosis and treatment: an evaluation of outcomes of a national program

A Meyer1, H Singh1, M Graber2.

1 Michael E. DeBakey Veterans Affairs Medical Center and Baylor College of Medicine, Houston, TX

2 RTI International, Research Triangle Park, NC

Background: One in five patients seek second opinions, an intervention that might prevent diagnostic/treatment errors. However, unlike in radiology and pathology, the impact of formal second opinions in general clinical practice has not been evaluated. We examined outcomes of a patient-initiated second-opinion program, specifically evaluating changes in diagnosis and treatment, estimated clinical impact, and patient satisfaction.

Methods: The nationally-administered second-opinion program, an employment benefit, allows employees to request second opinions at no additional cost. Program staff perform clinical intakes and obtain patients’ entire medical records, including notes, tests (laboratory, pathology, and imaging) and procedures performed. Trained physicians summarize the cases and identify key, unresolved clinical questions. For each case, all diagnostic data and key questions are forwarded to expert specialists. Experts’ recommendations are synthesized and given to patients to review with their physicians. The outcomes of these second opinions are graded by trained nurses who determine whether initial diagnoses and treatment recommendations were confirmed, clarified, or changed; and trained physicians who assess whether estimated clinical impact is none, minor, moderate, or major. One to two weeks after discussing recommendations with program experts, patients are invited to complete satisfaction surveys. A research team aggregated and independently analyzed data from all patient-initiated second opinions from 2011 and 2012.

Results: In the evaluation period, 6,791 patient-initiated second opinions were completed. These resulted in changes in diagnosis in 15%; confirmation in 57%, and clarification in 17% of cases. Impact on diagnosis was moderate or major in 22% of cases. Second opinions resulted in changes in treatment in 37%; confirmation in 27%, and clarification in 27% of cases. Impact on treatment was moderate or major in 32% of cases. 41% of cases had recommended changes in either diagnosis or treatment; 11% had changes in both. 93% of patients were satisfied with the experience, 90% said their questions were answered, 84% discussed results with their doctors, and 88% were more confident in their diagnoses/treatments. However, only 61% planned to follow the given recommendations.

Conclusions: A national program providing patient-initiated second opinions recommended moderate or major changes in diagnosis or treatment for over one-third of participants. This suggests significant diagnostic and treatment variability in real-world clinical practice, and a potential role for second opinions in preventing diagnostic and treatment errors. Although patient satisfaction with the program was high, further evaluation is needed to determine whether second opinions impact clinical outcomes, including the reduction of diagnostic errors.


*Presented in an oral session.

Calibration of diagnostic accuracy and confidence in physicians working in academic and non-academic settings

A Meyer1,2, P Krishnamurthy3, M Sur2,4, V Payne1,2, D Meeks1,2, R Rao1,2, H Singh1.

1 Michael E. DeBakey Veterans Affairs Medical Center, Houston, TX

2 Baylor College of Medicine, Houston, TX

3 University of Houston, Houston, TX

4 Texas Children’s Hospital, Houston, TX

Background: We previously found a surprisingly high degree of misalignment between diagnostic accuracy and confidence (i.e. poor diagnostic calibration) among physicians, such that physicians were found to be overconfident about their diagnostic accuracy. We sought to determine if there was a relationship between physician characteristics and diagnostic calibration.

Methods: We conducted a secondary analysis of data collected in a recent vignette-based study to evaluate physicians’ diagnostic calibration. 118 physicians solved cases through an online educational community. Calibration was examined using two measures: the “calibration statistic” and an index of miscalibration direction (O-U Index). We regressed the following physician characteristics onto both measures using separate multivariate regressions: experience since residency (years), sex, race, work status (full- vs part-time), country of birth (US vs International), country of medical education (US vs International), work institution type (private vs government), and academic affiliation of work institution (nonacademic vs academic).

Results: Overall, physicians diagnosed 31% of cases correctly with a confidence of 6.8 (on a scale of 0-lowest confidence to 10-highest confidence). Physicians’ mean diagnostic calibration (using the calibration statistic) was 0.38 (SD=0.16) (on a scale of 0-best possible alignment between accuracy and confidence to 1-worst possible alignment). Calibration significantly differed based on academic affiliation of the physicians’ work institution: calibration was better for those with an academic affiliation than for those without one (p=.002). Calibration did not differ, however, for other physician characteristics. Physicians’ mean direction of miscalibration (O-U Index) was in the overconfident direction (mean=0.38; SD=0.24) (on a scale of?1 for highest level of underconfidence to +1 for highest level of overconfidence). The extent of overconfidence also differed by academic affiliation (p=.004): academicians were less overconfident than their nonacademic peers. The extent of overconfidence did not differ by any other physician characteristic.

Conclusions: Our study suggests physicians’ diagnostic calibration may differ based on the academic nature of their work environment. One possible explanation is that academicians are likely to be involved in more continuous learning and feedback given their training work environment. If confirmed in real-world clinical settings, our findings could have implications for development and implementation of interventions to better align diagnostic accuracy and confidence to reduce diagnostic error. For example, the American Board of Internal Medicine’s Maintenance of Certification program, that most practicing physicians are now required to enroll in, could provide a possible venue through which interventions to promote continuous learning and feedback could be implemented in non-academic settings.

Providers’ responsiveness to messages about their patients’ delayed diagnostic evaluation

A Meyer1, D Murphy1, L Wu1, E Thomas2, S Forjuoh3, H Singh1.

1 Michael E. DeBakey Veterans Affairs Medical Center and Baylor College of Medicine, Houston, TX

2 University of Texas Medical School and the University of Texas-Memorial Hermann, Houston, TX

3 Scott & White Healthcare and Texas A&M Health Science Center, Temple, TX

Background: We previously developed and used electronic health record (EHR)-based “triggers” to identify patients at risk for delayed diagnostic evaluation for cancer. We then performed a cluster randomized controlled trial (RCT) to test the impact on time to subsequent evaluation when communicating information about these delays to providers. The communication strategy involved escalating steps, progressing from secure email to phone calls to informing clinic leadership. In this sub-analysis, we examined the effectiveness of these methods of information delivery in terms of provider responsiveness.

Methods: The RCT included 72 primary care providers at two sites (with 36 intervention providers). Prior to the RCT, providers specified communication preferences including preferred email addresses/phone numbers, times of day to call, and whether phone calls should be made to providers or their nurses. After triggers identified patients in need of further diagnostic evaluation, this information was communicated to providers in the intervention group via secure email. If action was not documented within a week, up to 3 telephone calls were made to providers or their nurses. Lastly, if providers or their nurses could not be reached, clinic leadership was informed. In these sub-analyses, we descriptively examined provider responsiveness to each method of information delivery. Additionally,?2 analyses compared responsiveness resulting from provider versus nurse contact for telephone calls and for each method of information delivery by site.

Results: In the intervention group, 369 patients required further diagnostic evaluation for cancer. Although we found reduced times to diagnostic evaluation for patients in the intervention group, informing providers about possible delays in their patients’ care did not always lead to their responsiveness. Specifically, providers responded to secure emails in only 11.1% of cases (n=41 of 369), subsequent phone calls in 68.6% of cases (n=225 of 328), and leadership contact in 45.5% of cases (n=5 of 11). Whether providers versus nurses were the designated phone call recipients made no difference in responsiveness (p=.82). However, providers at one site were more responsive to phone calls (73.6% vs. 52.5%; p=.006), but responsiveness for secure emails and leadership contact was comparable across sites (p=.61 and.55, respectively).

Conclusions: While triggered information has potential to be used as a safety net for diagnostic delays, communicating this information back to frontline providers does not necessarily lead to action. Because providers might be experiencing information overload, alternative strategies such as delivering this information to other care team members or quality assurance personnel warrant further evaluation.

* Video-oculography to reduce stroke misdiagnosis among acutely dizzy patients: a cost effectiveness analysis of the ‘eye ecg’

D Newman-Toker1, G Butchy2, H Lehmann1, E Aldrich3, A Chanmugam1, K Frick4.

1 Johns Hopkins University School of Medicine, Baltimore, MD

2 UMDNJ-Robert Wood Johnson Medical School, Stratford, NJ

3 Howard County General Hospital, Columbia, MD

4 Johns Hopkins Bloomberg School of Public Health, Baltimore, MD

Background: Dizziness and vertigo account for over 4 million annual US emergency department (ED) visits at a societal cost of more than $9 billion for ED assessment and post-admission inpatient care. Most have benign vestibular or cardiovascular causes, but ∼4% have stroke or transient ischemic attack (TIA). Roughly 35% of these cerebrovascular events are missed, and misdiagnoses often result in disability or death. Recent studies suggest bedside eye exam (‘HINTS’) is more accurate than MRI. Portable video-oculography (VOG) devices measuring these eye movements have been tested for preliminary accuracy. Before conducting large-scale clinical trials of VOG-based diagnosis, we sought to model the cost effectiveness of this approach relative to current practice and other alternatives for stroke diagnosis in ED patients with acute dizziness.

Methods: Cost-effectiveness analysis from the societal perspective. Combining literature and expert-derived estimates of probabilities and utilities with local hospital variable cost estimates, we constructed a decision model to compare current national ED diagnostic practice with four other hypothetical diagnostic strategies. Our base case was a 65 year-old in average health without disability presenting acute, continuous dizziness (>24 hrs, at high risk for stroke). We assessed current practice versus VOG and three non-selective diagnostic alternatives (“CT all”, “MRI all”, “admit all”). Outcome measures were cost, quality-adjusted life-years (QALYs), and incremental cost-effectiveness ratios ($/QALY). We also estimated potential national cost savings from reduced overtesting in patients with benign vestibular disorders.

Results: Applying VOG operating at 99% sensitivity and 97% specificity ($7,735/QALY) or “MRI all” ($12,200/QALY) would improve stroke outcomes and be highly cost effective. “CT all” would be less effective and “admit all” would not be cost effective ($190,000/QALY beyond “MRI all”). Results were sensitive to VOG accuracy, baseline prevalence of cerebrovascular events, and post-event life expectancy. Potential national cost savings from reduced overtesting in benign vestibular disorders would be ∼$1 billion per year (half from reducing unnecessary CT, half from reducing unnecessary inpatient admissions).

Conclusions: Bedside VOG (‘eye ECG’) could reduce stroke misdiagnosis and improve clinical outcomes among acutely dizzy patients at minimal societal cost. Improved diagnosis is cost effective but not cost saving with respect to stroke because correct stroke care (hospitalization) to deliver better patient outcomes increases costs, and more strokes are identified by VOG than current care. Implementing such an approach as part of broader diagnostic strategy in acute dizziness, could, however, save approximately $1 billion per year.


*Presented in an oral session.

Diagnosing error in diagnosis: a theory-driven approach to the diagnostic process

D Nystrom1, L Williams1, D Paull1, M Graber2.

1 VA National Center for Patient Safety, Ann Arbor, MI

2 RTI International, Research Triangle Park, NC

Background: Previous theoretic frameworks that describe diagnosis have failed to provide an adequate description of the diagnostic process or produce valid explanations for the occurrence of diagnostic error (Patel, Kaufman, & Kannampallil, 2013). I propose that assessing diagnosis with a theory other than System 1 System 2 thinking (Kahneman, 2011) will provide new insights into the diagnostic process. The theory of mental models (Johnson-Laird, 1983) suggests people reason by constructing representative working models of the world within their mind that allows them to understand information they receive from the environment and anticipate future states of the present situation. Given the theory of mental models description of the interplay between cognition and information from the environment, this theory seems appropriate to investigate the cognitive features of diagnosis and the determinants of diagnostic error.

Methods: To incorporate the theory of mental models into the diagnostic process, a representation that describes the development of a clinician’s mental model of a patient’s state (i.e., diagnosis) was created. To create this model of diagnosis, a cognitive engineering technique known as work domain analysis (WDA) (Vicente, 1999; Lintern, 2011) was used to elicit the important functional components of diagnosis from diagnostic training literature (LeBlond, DeGowin, & Brown, 2009).

Results: Figure 1 presents the WDA model of diagnosis. A legend has been provided within the figure to facilitate comprehension of this model of diagnosis. (Figure). The WDA Model of Diagnosis that was developed from diagnosis training literature and the application of the theory of mental models.

Conclusions: The WDA model of diagnosis provides a unique, holistic account of the diagnostic process. This model describes where information is collected from the clinician’s environment (e.g.: electronic health records, physical examination, medical literature, etc.) and the way this information influences their reasoning to arrive at a diagnosis. Under this model of diagnosis, diagnostic error is the result of a clinician committing to an incorrect model of the patient state based on their selection and comprehension of the information available to them. To mitigate the probability of committing a diagnostic error, this model suggests research efforts should focus on the development of clinicians’ information collection, elicitation and comprehension skills; enhance technological systems to ensure information is correct and easy to comprehend for clinicians; introduce variability of disease presentations to enrich clinician’s mental model of that particular disease; and encourage the deliberate practice of diagnosis via simulation or other means.

Using voluntary physician reporting to learn from diagnostic errors in emergency medicine

N Okafor1, V Payne2, Y Chathampally1, S Miller1, P Doshi1, H Singh2.

1 University of Texas-Health Science Center at Houston, Houston, TX

2 Michael E. DeBakey Veterans Affairs Medical Center and Baylor College of Medicine, Houston, TX

Background: Diagnostic errors are common in the emergency department (ED) but few studies have evaluated their types and origins. Reporting systems offer an opportunity to identify and learn from diagnostic errors. We sought to analyze incidents reported by ED physicians to determine the disease conditions, contributory factors and patient harm associated with ED-related diagnostic errors.

Methods: Between 2009 and 2013, we analyzed 209 diagnostic error incidents reported at two academic hospital-affiliated EDs in a large metropolitan area. ED physicians used a department-specific voluntary physician incident reporting system. Each incident was reviewed by an ED physician champion-led multidisciplinary quality assurance team. We generated descriptive statistics quantifying the disease conditions, contributory factors and patient harm associated with each incident.

Results: We identified 214 diagnostic errors associated with 65 unique diseases/conditions, most common being sepsis (9.6%) and acute coronary syndrome (9.1%), followed by fractures (8.6%) and vascular injuries (8.6%). We categorized contributory factors into three categories: cognitive, system-related and non-remedial. Across the209 incidents, we identified 317 cognitive factors, 192 system-related factors and 106 non-remedial factors. Faulty information verification (41.3%) and faulty information processing (30.6%) were the major cognitive contributing factors. High workload (34.4%) and inefficient processes (40.1%) such as diagnostic test processing delays or delays in critical finding notification by radiology accounted for the majority of the system-related contributing factors. Atypical presentation (31.3%) and the patients’ limited ability to provide a medical history (31.3%) were the major non-remediable contributing factors. More than one contributory category was represented in 157/209 (75%) incidents. Both cognitive and system-related factors were identified in 83 (40%) incidents, whereas all 3 categories (cognitive, system-related and non-remedial) were identified in 53 (25%) incidents. Major harm defined as death, permanent disability or a life or limb threatening event was associated with 34/209 (16.3%) of reported incidents. The remaining incidents were associated with moderate (31.6%), minor (34.4%), no (9.1%) or unknown harm (8.6%). Results of this project have led to several quality improvement projects such as targeted clinician education modules, creation/revision of diagnostic ED protocols and the redesign of the ED sign-out process.

Conclusions: A physician reporting system was useful in identifying diagnostic errors in the ED. Majority of reported diagnostic errors were associated with patient harm and involved a combination of cognitive and system-related errors that were amenable to improvement strategies. Our project was valuable in making diagnostic errors a focus of quality improvement strategies in the ED.

Evidence in support of a role for disease prototypes in a system 1 based approach to differential diagnosis

F Papa1, K Kalinowski2, D Aldrich3, J Ruffin3.

1 University of North Texas Health Science Center, Fort Worth, TX

2 National Board of Osteopathic Medical Examiners, Chicago, IL

3 ACDET, Fort Worth, TX

Background: According to Dual Processing Theory, System 1 enables differential diagnosis (DDX) via the use of disease knowledge stored in the form of both: 1) a record of the S/S associated with each case portrayal (‘instance’) of every disease previously encountered, and 2) a single, probabilistically portrayed summarization of the S/S associated with each known disease (‘prototype’). System 1 arrives at a diagnosis via a pattern recognition mechanism which compares and thereby determines the particular disease instance or disease prototype which best matches the S/S associated with a new case encounter. Evidence of the predominance of disease instances or prototypes during DDX could lead to improved approaches to DDX instruction. Such evidence could be determined if subject performance better correlates with the ‘frequency-based’ typicality gradient associated with disease instances or the ‘probabilistic’ typicality gradient associated with disease prototypes.

Methods: Following IRB approval, 117 year one medical students were provided 54 Acute Chest Pain training cases (six case portrayals of each of nine diseases such as myocardial infarction, pulmonary embolus, and pneumonia). Each training case was assigned both a frequency-based and probabilistic typicality gradient estimate. Following training, a different set of 36 test cases was used to assess student performance (four cases portrayals of each of the nine diseases, with each case assigned both a frequency and probabilistic typicality gradient estimate). Each of four test cases for all nine diseases was further assigned to one of four typicality quartiles based upon their respective frequency and probabilistic gradient estimates. Student performance against the 36 test cases was determined and each student was assigned to one of four performance groupings (lowest thru highest performing group). The assignment of test cases into typicality quartiles, and students into performance quartiles, enabled the correlation of student performance with case typicality. A frequency-based typicality gradient would predict no correlation (flat line), while a probabilistic typicality gradient would predict a positive (ascending) correlation across the case typicality quartiles.

Results: Each of the four student performance groupings demonstrated performance that correlated with a probabilistic, typicality/performance gradient (see Figure 1 below).

Conclusions: The finding suggests that: 1) disease prototypes are likely responsible for the observed performance, 2) performance differences across each student grouping are likely due to either differences in the students’ capacity to form ‘robust’ disease prototypes and/or their inherent pattern recognition mechanisms, and 3) educators emphasize training to disease prototypes in their efforts to optimize DDX performance.

* Diagnostic errors in medicine: view from the front lines-the patients’ perspective

V Payne1, S Menon1, V Modi1, T Davis-Giardina1, H Singh1.

1 Michael E. DeBakey Veterans Affairs Medical Center and Baylor College of Medicine, Houston, TX

Background: Individuals at the front lines of care can potentially provide valuable insights on diagnostic errors. We sought to ascertain patients’ perspectives on diagnostic error-related contributory factors and prevention strategies.

Methods: We conducted in-depth interviews with 35 participants who either personally experienced (n=22) or witnessed a diagnostic error of a family member (n=13). Participants described their experiences and identified diagnostic error-related contributing factors and prevention strategies. Using grounded theory methodology, two reviewers independently coded transcripts and identified recurring themes. To facilitate analysis, we created graphical timelines for certain interviews to depict diagnostic process breakdowns and illustrate patient and provider actions and key events.

Results: Some of the predominant contributory factors for errors according to patients included providers anchoring on an incorrect diagnosis early in the process; not listening to patients’ complete medical history; lack of investigation despite the presence of symptoms; not viewing the patient holistically; lack of care coordination across multiple providers; disease demographic biases; unfair patient stereotyping and health insurance restrictions. Some patients indicated not seeking timely healthcare might have contributed to their delayed diagnosis. Many patients researched their symptoms on the Internet and reviewed medical literature and manuals, arriving at a diagnosis determined to be the definitive diagnosis. When presenting their findings, they often felt providers quickly dismissed their views. Patients wanted to participate in the diagnostic process, but often felt unheard and disrespected when asking questions or offering suggestions. Patients reported that diagnostic errors could be prevented if providers “listen to the patient” while taking the medical history and be open to their suggestions regarding tests and potential diagnoses; think outside the box and consider a wider range of possibilities for their symptoms; investigate rather than automatically assume an undiagnosed mental health issue; follow-up on test results promptly and shift to a team-based care paradigm. After the event several patients provided feedback to providers who missed the diagnosis; however, many indicated they were too emotionally laden to do so. Most patients and family members suffered significant and long-lasting emotional and/or financial effects. Patients and their families want accountability and found an apology helped them cope.

Conclusions: Patients and family members who have experienced a diagnostic error can offer valuable insights on the diagnostic process and identify specific contributory factors and preventive strategies. Our findings should encourage health care institutions and providers to engage with patients as a source of learning and prevention of diagnostic errors.


*Presented in an oral session.

A grounded provider feedback framework-the relationship between providers’ feedback acceptance and performance improvement

V Payne1, S Hysong1.

1 Michael E. DeBakey Veterans Affairs Medical Center and Baylor College of Medicine, Houston, TX

Background: Audit and feedback has been used widely with mixed success for performance and quality improvement across various disciplines. However, giving providers feedback on their diagnostic performance has not been widely implemented. There has been extensive research on characteristics of feedback associated with improving outcomes. However, there is limited research regarding clinicians’ acceptance of performance feedback and if feedback alters clinical practice behavior specifically. We investigated physicians’ acceptance of clinical performance feedback and identified factors that induce or deter clinician behavior change.

Methods: We analyzed interviews with primary care clinicians practicing in 16 geographically dispersed Veterans Affairs medical centers, selected purposefully by their clinical performance profile. Interview questions focused on perceptions of clinical performance feedback regarding outpatient measures designed to improve quality of care, with particular emphasis on VA’s clinical performance measurement system. Using grounded theory, transcripts were deductively analyzed searching for statements referring to feedback acceptance and factors impacting practice changes.

Results: A physician feedback framework emerged indicating feedback acceptance and clinician behavior changes are interconnected. We found that providers had mixed reactions to clinical performance feedback. Factors facilitating physician acceptance included having their performance assessed by someone that understands the environment, feedback providing patient-level rather than facility-level data exclusively. Feedback was commonly ignored when it was untimely, performance measures were unrealistic, and performance was based on a small sample size. A sense of competition among providers and financial incentives facilitated clinician behavior change. Additionally we found that systemic and environmental factors impact implementation of performance enhancement strategies into the clinical setting. Providers also indicated implementing changes into their practice was difficult due to patient quota expectations and time restrictions. A quality clinical team to remind patients of appointments and tests provided a means to enhance clinical outcomes and performance.

Conclusions: Our physician feedback framework postulates specific factors associated with physicians’ acceptance of performance feedback and factors inducing clinician behavior change designed to improve performance. This framework serves as a foundation to be used when developing physician diagnostic performance feedback systems intended to reduce diagnostic errors. Giving providers feedback regarding diagnoses commonly missed will provide specific areas for improvement. Delivering timely and specific information on patients misdiagnosed by providing the definitive diagnosis and tests/procedures used to arrive at the correct diagnosis will be valuable when diagnosing patients with similar symptoms. We conclude that feedback has the potential to enhance the diagnostic process and reduce the diagnostic error rate.

Missed opportunity to diagnose and treat sleep breathing disorders

T Power1, Y Chi1, N Munns1.

1 AIM Specialty Health, Chicago, IL

Background: Sleep breathing disorders (SBD) are common and underdiagnosed. Untreated SBD leads to poor outcomes and increased health care costs. Diagnosis of SBD requires either a home sleep test or polysomnography. Positive airway pressure therapy and oral appliances are first-line approaches to the management of SBD. Many payers require preauthorization for diagnostic sleep testing and subsequent therapy. When testing is requested and approved but not performed, an important diagnostic and therapeutic opportunity is missed. We sought to quantify this missed opportunity.

Methods: Using the preauthorization database for a sleep management program covering approximately 4 million lives across 6 states, we determined the number of patients approved for sleep studies between 1/1/2013 and 6/30/2013. Interrogation of insurance claims data from 1/1/2013 to 9/30/2013 established the number of approved studies actually performed. The number of members approved for treatment following testing between 1/1/2013 and 12/31/2013 was then compared to claims for testing between 1/1/2013 and 3/31/2014.

Results: 18,359 members were approved for testing for SBD. Of these, 11,907 (65%) were linked to a payment claim. Therefore the missed diagnosis opportunity (MDO) is 6,452 (18,359-11,907) or 35%. Therapy for SBD was authorized in 6,984 patients tested, of which 6,356 were followed by a payment claim. Prevalence of SBD in patients undergoing testing was 58.7% (6,984/11,907). Assuming the same prevalence of treatment-requiring SBD in the cohort who did not undergo testing, the missed treatment opportunity (MTO) can be calculated as (MDO X 58.7%) + (treatments approved – treatment claims for payment). Thus, MTO = (6,452 X 58.7%) + (6,984 – 6,356) = 4,415 (24% of total members approved).

Conclusions: Our data demonstrate that clinically appropriate testing for SBD, even when authorized for payment on the basis of clinical appropriateness, is often not followed by performance of the test. Failure to have the approved test performed represents MDO which occurred in 35% of cases. MTO was seen in 26% of approved studies primarily driven by failure to have the diagnostic test performed. Patient care is not limited to selection of the most appropriate diagnostic test but also to ensuring that the test is performed and the treatment initiated. When the diagnosis under consideration has high pretest probability and significant clinical and cost implications when left untreated, the impact of MDO and MTO is not insignificant. Physicians managing patients with SBD need to identify those who fail to follow testing and treatment recommendations.

Use of alternative diagnostic strategies for coronary artery disease

T Power1, Y Chi1.

1 AIM Specialty Health, Chicago, IL

Background: Myocardial Perfusion Imaging (MPI) and Stress Echocardiography (SE) are diagnostic tests used in the evaluation of coronary artery disease (CAD), the leading cause of mortality in the United States. Clinicians ordering these tests should anticipate that test results will direct subsequent care. When testing is requested, but cannot be performed, it behooves the clinician to develop an alternative diagnostic strategy (ADS). Radiology Benefit Management (RBM) entities deny requests for diagnostic testing (DT) when such requests are considered to be clinically inappropriate. We evaluated the frequency of use of ADS following denial of MPI or SE.

Methods: Insurance claims for 90 days following denial of requests for MPI or SE were queried. ADS was defined as exercise treadmill test (ETT), SE, MPI, cardiac PET, coronary CTA, or coronary arteriography (CA). ADS included re-request and approval of the originally denied study. Site of service of the ADS and the frequency of revascularization procedures were evaluated.

Results: Between 1/1/2011 and 12/31/2012, 9689 requests (5506 SE and 4183 MPI) did not meet medical necessity guidelines and were denied. In the subsequent 90 days, an insurance claim for an ADS was seen in 4297 (44%) of denied cases {2292 (42%) SE and 2005 (48%) MPI}. ETT was the most frequently selected ADS (44% of ADS). 1.8% of ADS services were rendered in the inpatient setting. 12 (0.11%) of denied patients underwent PCI/CABG.

Conclusions: The premise for use of DT is that results should direct subsequent patient management. Denial of requests for DT should therefore present the ordering clinician with a diagnostic dilemma, thereby triggering an ADS. We found that following denial of MPI and SE, ADS was pursued in less than 50% of cases suggesting that the clinicians’ perception of the value of the DT was altered by the denial. Alternatively, ordering of DT may be driven by factors other than the need for pivotal clinical information including, medico legal concerns, lack of familiarity with appropriate use criteria, patient generated requests for imaging, financial remuneration etc. Low rates of hospitalization, CA and revascularization suggest that denial of MPI and SE occurs in low risk cohorts where withholding services does not materially change outcomes. In an era of constrained healthcare resources, clinicians need to consider how DT will impact subsequent management and, in the event of denial, what, if any alternative strategies should be employed.

The paucity of diagnostic findings in geriatric patients admitted for community acquired pneumonia

J Rubenstein1.

1 North Shore LIJ Health System, Roslyn Heighs, NY

Background: Community Acquired Pneumonia (CAP) is a common admitting diagnosis in the geriatric (65yo or >) population. The inpatient treatment of CAP consumes substantial assets; the patients suffer high morbidity and mortality. CAP however remains a syndrome rather than a distinct disease and it is difficult to make a specific diagnosis related to nebulous clinical symptoms, signs, and diagnostic studies. CAP early treatment is highly incentivized. Undifferentiated symptoms trigger an easy diagnostic hypothesis on which to treat. We hypothesize that the admitting diagnosis of CAP takes place early in the diagnostic process based on its prevalence rather than on the basis of definitive findings. Diagnostic errors due to heuristics may occur perhaps compounding the expenditures, morbidity, mortality. We present data on the findings in a group of geriatric patients hospitalized for the treatment of CAP.

Methods: A retrospective chart review was performed of the ED evaluations culminating in hospitalization for the primary diagnosis of CAP in patients 65 yo or greater. Patients with specifically immunosuppressive states and secondary diagnosis of CAP were excluded. The charts were assessed for Fever > 100F, Tachycardia >100bpm, Tachypnea>24bpm, pulmonary exam findings other than bilat wheezes, cachexia, and confusion. Chest x-ray findings were denoted as No Acute Disease (NAD), non-specific(NS), possibly(poss), probably(prob), or consistent with (c/w) pneumonia. The patient received one point for each of the clinical findings present and the reading of the x-ray was recorded. The distributions and averages are presented.

Results: Patients meeting the study criteria were reviewed (n=115). The mean number of clinical points/patient in the study group was 1.4. The most common x-ray reading was non-specific. No correlation was noted between the number of clinical parameters present and the CXR findings.

Conclusions: CAP is a common early diagnostic hypothesis in geriatric patients presenting with symptoms such as SOB, fever, cough, sputum, or confusion. It is triggered in the diagnostic process because of its prevalence and not definitive findings. Cognitive biases play a role in early termination of the diagnostic process without appropriate diagnostic verification. The resultant diagnostic error is likely amplified by the early treatment algorithms for CAP. Diagnostic momentum ensues perpetuating the early error. Diagnostic accuracy suffers and leads to an excess in costs, which might be diminished by maintaining skepticism when definitive clinical findings are absent.

Attention to Troponin I thresholds, rather than instability, leads to missed cases of acute coronary syndrome

R Schulteis1, J Boggan1.

1 Durham VA Medical Center, Durham, NC

Background: For patients presenting with chest pain, early cardiac biomarker testing can assist in the diagnosis of acute coronary syndrome (ACS). Cardiac biomarkers such as Troponin I are often interpreted as positive if exceeding a threshold. As the sensitivity of Troponin I testing has improved, however, subthreshold changes in Troponin I levels have become detectable. This subthreshold instability could indicate underlying myocardial necrosis that might be predictive of cardiovascular events. Furthermore, at-risk individuals would not have been identified with the conventional method of Troponin I interpretation.

Methods: Using the Veterans Affairs (VA) electronic health record, we identified 376 individuals presenting to the Emergency Department at a single VA hospital from January 2012 to August 2013, with 2 or more Troponin I values drawn between 3 and 12 hours apart. We excluded 43 patients whose Troponin I values were indicative of myocardial infarction (MI) leaving 333 patients. We defined a patient as having an unstable Troponin I if there was a greater than 20% change between the first and second values. We defined ACS as a combined endpoint composed of MI, death, and revascularization. Using cox proportional hazards, we estimated the effect of Troponin I instability on 30-day ACS.

Results: Of the 333 included patients without MI, 240 had a stable Troponin I while 93 had an unstable Troponin I. Figure 1 demonstrates that, after excluding those patients presenting with a Troponin I above threshold, there are higher rates of ACS amongst patients with subthreshold but unstable levels of Troponin I (hazard ratio 1.53 with 95% CI 1.01 to 2.32, p = 0.05).

Conclusions: By interpreting subthreshold Troponin I levels as “negative”, providers miss the opportunity to identify individuals that remain at an increased risk for early cardiovascular events. The subset of individuals with “negative” but unstable Troponin I values have a 50% increased risk of ACS relative to those with stable values. By focusing on the stability of Troponin I, we could potentially improve outcomes amongst this group via earlier recognition of risk.

Figure 1 ACS-Free Survival By Group.

* Internal medicine and emergency medicine residents’ perceptions of outcome feedback after handoffs

E Shenvi1, R El-Kareh1.

1 University of California, San Diego, La Jolla, CA

Background: Learning patient outcomes after forming assessments and plans is recognized as being crucial for calibrating diagnostic and therapeutic decision-making, but handing off patients to other care teams frequently disrupts this process. This lack of feedback may be especially problematic to physicians in training, although resident perspectives on post-handoff outcome feedback are not well understood.

Methods: We conducted a web-based 9-item survey of internal medicine and emergency medicine residents at our institution to study how often they find out about patients they have handed off, methods that they use to obtain feedback, their perceptions on the value in learning such outcomes, and the barriers to doing so. The first 8 questions were scored on 5-point Likert scales and the final item allowed for comments.

Results: Seventy-eight of 149 residents responded to our survey, for a 52.3% response rate. 73% responded that they only “sometimes” or “rarely” find out the outcomes of patients they have handed off, although 95% said learning outcomes was “moderately” or “very important” to both their education and job satisfaction. 84.6% were not satisfied with current systems of learning outcomes of patients after handoffs. The main barriers cited were too little time and lack of a reliable system to track prior patients.

Conclusions: Despite perceived importance of learning outcomes after handoffs, the majority of residents cite difficulty with obtaining such information. This study demonstrates that providing feedback on patient outcomes, which would help with future diagnostic decision-making, would meet a recognized need among physicians in training.


*Presented in an oral session.

Detectable outcomes of inpatient diagnostic error: a conceptual framework and scoping review

E Shenvi1, R El-Kareh1.

1 University of California, San Diego, La Jolla, CA

Background: Diagnostic error has been difficult to detect and study, and may benefit from automated screening using “trigger” tools, which have been used successfully in other error detection endeavors. We sought to create a conceptual framework of potentially detectable outcomes of inpatient diagnostic error.

Methods: We created our initial framework of potential outcomes of inpatient diagnostic error based on related published literature. We refined the framework after reviewing published adverse event screening criteria and analyzing eight appropriate clinical cases of diagnostic error. For the scoping review, we searched three databases and reviewed 8568 citations to include 33 studies that developed or validated record screening criteria for detection of adverse events that can include diagnostic error.

Results: Outcomes of diagnostic error, as seen in clinical cases and detection studies, were of four categories: clinical deterioration, correction of error or modifying plan, diagnostic uncertainty, and unexpected time course of illness. We found that these outcomes could be detected as changes in patient status, clinical assessments, and/or clinical management. Almost all studies use death, cardiorespiratory arrest or code, and transfer to intensive care as screening criteria for adverse events. Thirteen studies reported detecting diagnostic error, although definitions varied. In these studies, the proportions of detected errors that were diagnostic in nature ranged from 5.1% to 67.5%. We also identified criteria, such as changes in diagnoses on patient problem lists, that might have potential for diagnostic error screening but have not been used in published systems.

Conclusions: We developed an initial conceptual framework of potentially detectable outcomes of inpatient diagnostic error. Such a framework may be useful for developing screening and detection tools. Clinical cases demonstrate that changes in diagnosis and treatment plans are common outcomes of error, in addition to other common hospital screening events such as death and arrest. We anticipate that this set of criteria will continue to be revised and warrants evaluation for automated screening of inpatient diagnostic errors.

Changing post-analytic test reporting informatics formats to avert diagnostic error in high risk patients

E Travers1.

1 Private Medical Practice, Philadelphia, PA

Background: Reducing the incidence of diagnostic adverse medical events requires physician awareness of critical patient risks at the time of patient encounter, not after the visit. In clinical medical applications, Pattern Recognition Analysis is used to expedite anomaly detection in the patient’s data, obtained from both clinical in vivo and in vitro tests. A real world problem contributing to diagnostic mishaps, is that raw test data is delivered to the physician, and the physician is expected to mentally perform data reduction and pattern analysis at the site of encounter without technologic aids to organize significant abnormal values into organ-specific patterns. In vitro biomarker tests on blood and body fluids “cluster” into characteristic patterns for each diagnostic entity. Multiple laboratory biomarker tests, their correlations and relationships identify an illness. As with a fingerprint one line does not identify the individual, and neither does a single biomarker test identify a disease or illness.

Methods: Systematic literature review and metaanalysis of peer-reviewed journal articles, that evaluated pathophysiologic biomarkers signaling deterioration of patients in any clinical setting using informatics metric methods, and with any study design, provided that they addressed a patient-related outcome based on grading and risk stratification of biomarker abnormality.

Results: Patterns of significantly abnormal clinical and laboratory test results have predictive value for early detection of impending physiologic deterioration. Laboratory-directed interventions to optimize the reporting of significant patterns of abnormal biomarkers using diagnostic interpretation are available and can be effective in reducing errors.

Conclusions: Post analytic patterns of significantly abnormal blood test results are important “Trigger Tools” to signal high risk states in complex care patients. The unfocused, conventional reporting process for results of abnormal biomarker tests is a significant contributor to diagnostic error, since reports are scientifically formatted & not patient-specific. The simultaneous evaluation of multiple biomarkers and their interrelationships, using combinations of significantly abnormal clinical subjective and scientific objective laboratory test results improves diagnostic accuracy.

Using discriminant pattern analysis to stratify patient risk susceptibility to diagnostic adverse events

E Travers1.

1 Private Medical Practice, Philadelphia, PA

Background: Reducing the incidence of diagnostic adverse medical events requires physician awareness of critical patient risks at the time of patient encounter, not after the visit. In clinical medical applications, Pattern Recognition Analysis is used to expedite anomaly detection in the patient’s data, obtained from both clinical in vivo and in vitro tests. A real world problem contributing to diagnostic mishaps, is that raw test data is delivered to the physician, and the physician is expected to mentally perform data reduction and pattern analysis at the site of encounter without technologic aids to organize significant abnormal values into organ-specific patterns. In vitro biomarker tests on blood and body fluids “cluster” into characteristic patterns for each diagnostic entity. Multiple laboratory biomarker tests, their correlations and relationships identify an illness. As with a fingerprint one line does not identify the individual, and neither does a single biomarker test identify a disease or illness.

Methods: Systematic literature review and metaanalysis of peer-reviewed journal articles, that evaluated pathophysiologic biomarkers signaling deterioration of patients in any clinical setting using informatics metric methods, and with any study design, provided that they addressed a patient-related outcome based on grading and risk stratification of biomarker abnormality.

Results: Patterns of significantly abnormal clinical and laboratory test results have predictive value for early detection of impending physiologic deterioration. Laboratory-directed interventions to optimize the reporting of significant patterns of abnormal biomarkers using diagnostic interpretation are available and can be effective in reducing errors.

Conclusions: Post analytic patterns of significantly abnormal blood test results are important “Trigger Tools” to signal high risk states in complex care patients. The unfocused, conventional reporting process for results of abnormal biomarker tests is a significant contributor to diagnostic error, since reports are scientifically formatted & not patient-specific. The simultaneous evaluation of multiple biomarkers and their interrelationships, using combinations of significantly abnormal clinical subjective and scientific objective laboratory test results improves diagnostic accuracy.

Physician reactions to uncertainty and decision-making in primary care

E Tsiga1, E Panagopoulou1, A Benos1.

1 Aristotle University of Thessaloniki, Thessaloniki, Greece

Background: Physicians routinely face uncertainty while making diagnostic and treatment decisions. How physicians deal with this uncertainty might affect their decisions and affect patient outcomes, i.e. suboptimal decisions can lead to errors and harm. However, little is known about how general practitioners (GPs) deal with uncertainty and how this affects their decisions. Focusing on affective dimensions, we investigated how GPs’ reactions to uncertainty related to their diagnostic and treatment decisions in primary care.

Methods: We administered an on-line questionnaire to GPs practicing across all (both public and private) primary care settings in Greece. The questionnaire used twenty brief clinical scenarios to assess GPs’ decisions to refer a patient to a hospital for further diagnostic evaluation/treatment for the given clinical presentation (e.g., additional tests, procedures, subspecialty consultation, and possible admission). Clinical scenarios were developed and selected through two-stage expert focus groups with GPs. The first focus group identified the most common cases in primary care and the second selected 10 scenarios that would require referral to the hospital for further diagnostic evaluation/treatment per clinical guidelines and an additional 10 where referral was not required. We measured GPs’ affective reactions to uncertainty through four dimensions, based on available psychology literature: GPs’ stress induced from facing uncertainty, fear of malpractice, and general and professional self-esteem. Outcome indicators included 1) total number of decisions to refer, 2) number of decisions to refer when not clinically indicated, and 3) number of decisions not to refer when clinically indicated.

Results: We recruited 135 GPs (66 male), with 11.01 average years of clinical experience (SD;8.25). When controlling for clinical experience and gender, GPs with less general self-esteem (RPearson:-0.258;p:0.008), increased stress of uncertainty (RPearson:0.226;p:0.002) and increased fear of malpractice (RPearson:0.306;p:0.001) made more total hospital referrals. However, GPs with less general self-esteem made more referrals even when not indicated (RPearson:-0.259;p:0.005), while physicians with decreased stress of uncertainty and decreased fear of malpractice decided not to refer patients even when clinically indicated (RPearson:-0.220,p:0.002 and RPearson:0.228,p:0.001,respectively). No effects were found for professional self-esteem.

Conclusions: GPs affective reactions to uncertainty could play an important role in their diagnostic and treatment decisions. While more comprehensive research on the types, origins and effects of these reactions is needed in simulated/clinical settings, our preliminary research suggests that future strategies that can modulate or help manage certain affective reactions to uncertainty could produce better decisions. These in turn could lead to safer diagnosis and treatment.

Persistence of diagnostic delay in year 2013: a bibliometic study

J Wofford1, C Campos1.

1 Wake Forest University School of Medicine, Clemmons, NC

Background: Previous studies of diagnostic delay (DD) have been few, inconsistent in methodology, and tentative in defining the amount of delay for specific diagnoses. Diagnostic delay should have decreased over time with more sophisticated diagnostic tools and greater awareness of diagnostic error, but data are lacking.

Methods: We conducted a bibliometric search of the MEDLINE database using the free text phrase “diagnostic delay” and restricting our search to publications in year 2013. Abstract review allowed categorization of abstracts by country of origin, study type, and disease category. The degree of diagnostic delay was determined for most common diseases and with disease registries as the study type.

Results: Of the 135 citations, most citations came from institutions in the USA (9) and UK (19), but 35 countries were represented in the search results. 12% (17/135) of citations were from developing countries. Study type was not discernible from the abstract in the case of 7 citations. Study types included case-reports or case series (37), discussions/literature reviews (22), registries/clinical networks (17), physician interviews (2). There were 3 prospective studies with the remaining studies retrospective or cross-sectional in design. Disease categories were most often infection (25/135, 19%), cancer (20), neurologic (15), rheumatologic (10), or vascular (8). The most common specific diagnoses were tuberculosis (13), ALS (4), endometriosis (3), oral cancer (3), ocular cancer (3) but other diagnoses were represented no more than twice among the citations. Disease registries offered the largest number of patients (mean 6564, range 100-56462), and specific duration of DD was offered in 16 of 17 studies. No studies of systematic approaches to diagnostic error were evident with this search strategy.

Conclusions: Diagnostic delay is alive and well in the medical literature. Compared with previous studies of diagnostic delay, data from disease registries offers larger sample sizes and more confidence in defining duration of delay. The literature on DD does not intersect with systematic or clinical approaches to reducing diagnostic error.


Skill degradation evaluation toolkit for eliminating competency-loss (skill-detect): a practice management framework

E Baker1, R Champney1, K Stanney1, R Long1, S Duff1, J Jacko2, F Sainfort2, J Chan2, A Nelson2.

1 Design Interactive, Inc., Oviedo, FL

2 Biomedical Metrics LLC, Edina, MN

Statement of problem: Diagnostic error in medicine is oftentimes attributed to health practitioners, as medicine has become increasingly complex, creating more opportunities for human error. Achieving proficiency is essential to avoiding such error, however, maintaining proficiency may be just as important for reduced complications rates and improved outcomes. In order for health practitioners to maintain proficiency, a practice management framework is needed that can do more than monitor errors or specify procedure practice frequency, it must be able to detect areas of skill decay and specify appropriate retraining regimens.

Description of the intervention or program: To meet this need in practice management, the Skill Degradation Evaluation Toolkit for Eliminating Competency-loss Trends (Skill-DETECT) framework is being developed to (1) identify when degradation of cognitive clinical knowledge, skills, and abilities (KSAs) occurs, (2) predict probable and confirm onset of cognitive skills degradation through the analysis of electronic health records (EHR) data on task frequency and complexity (based on procedure codes) and associated skill decay modeling, and (3) prescribe retraining regimens which enable physicians to preemptively refresh knowledge and maintain proficiency. A hybrid cognitive model is used to assess skill decay, wherein a power law model is utilized as a first pass filter. Additional filters utilize rule-based and statistical modeling methods to further refine decay prediction based on patterns of performance compared to standards of care. The Skill-DETECT framework is designed to scan EHR data associated with a physician, flag KSAs that show potential for skill decay, evaluate through a model-based approach the level of predicted decay for flagged KSAs, and then specify retraining that will maintain proficiency.

Findings to date: Through literature review and physician interviews with a panel of six civilian physicians, a taxonomy was developed that outlines the types of cognitive KSAs necessary to model skill decay, with the initial focus on internal medicine. The taxonomy was then cross-validated through a panel of six military physicians. The resultant taxonomy will serve as the foundation for the Skill-DETECT practice management architecture by identifying requisite KSAs, associated norms for each KSA (i.e., frequency, complexity, measures of performance) that can be extracted from EHR data sources, and parameters for modeling KSA decay.

Lessons learned: EHR data may be a necessary but not sufficient data source for modeling skill decay and thus these data may need to be complemented with objective measures of cognitive skill decay that interactively assess the particular skills and knowledge areas that have degraded for a physician.

Can we teach internal medicine residents to recognize and reduce diagnostic errors using structured debiasing methods and mindfulness?

A Bullen1, M Lopez2, J Hanley1.

1 University of Texas Health Science Center, Harlingen, TX

2 University of Texas Health Science Center, San Antonio, TX

Statement of problem: One of the principle causes of diagnostic errors is thought to be cognitive failure, often because of cognitive biases. These errors could potentially be reduced by creating an environment of awareness (mindfulness) and utilizing specific debiasing strategies, described by Croskerry.

Description of the intervention or program: Observational prospective study with following goals: determine if Residents can identify potential diagnostic errors, determine if Residents can classify potential diagnostic errors based on cognitive biases, provide Residents with methods of debiasing techniques and to add mindfulness to their diagnostic process and finally, determine if adding these methods would change their diagnostic index. During their 15-day night float rotation where residents admit patients from the emergency room, they will be asked to complete the following steps: prior to the beginning of their rotation, to perform a diagnostic thinking inventory, after each admission resident, the Resident will do a survey which contains five cognitive debiasing techniques (such as structured data acquisition and more skeptical strategy). Every morning, facilitated check-in rounds will be held where each admission will be discussed. Two weeks after the end of the rotation, the Resident will be asked to perform a diagnostic thinking inventory once more.

Findings to date: In 27% of the cases residents did not agree with the ED diagnosis because of different history and physical exam findings combined. Most of them spent between 15 and 30 minutes doing their evaluation. In almost half of the cases, Residents had doubts regarding the diagnosis and identified findings that did not fit the initial ED physician diagnosis. Residents identified potential diagnostic errors committed by ED physicians in half of the cases. The most common subtypes of cognitive biases included: premature closure and anchoring. Results of the pre and post rotation diagnostic thinking inventory are at this time very preliminary but show a trend in improvement after the rotations.

Lessons learned: Data from the intervention tool survey suggest that it serves not only as a method to teach Residents to detect diagnostic errors but also forces Residents to be mindful. We believe that having a facilitated check-in round is essential to process, requiring the Resident and Faculty to be engaged in an effort to learn about and to detect diagnostic errors. With the acknowledgement of diagnostic errors as critical issues in patient safety, it is imperative that training programs develop methods designed to teach and recognize these cognitive biases, closely monitoring to ensure faithful and strict implementation, and subsequent reduction of diagnostic errors.

Implementation of the “final check” to reduce mis-labeled lab specimens

G Ferguson Jr1.

1 The Regional Medical Center of Orangeburg and Calhoun Counties, Orangeburg, SC

Statement of problem: Thousands of laboratory specimens are submitted for diagnostic testing with erroneous label information causing risk of mis-diagnosis. The causes of mis-labeling are varied. A common location where specimen mis-labeling is prevalent is in the Emergency Department.

Description of the intervention or program: In the ED a computer generated, pre-printed label is taken to, or located in the patient’s room. Patient identification is performed by verifying armband demographics and comparing elements between the armband and the pre-printed label like name, date of birth, Financial Number and Medical Record Number. After specimens are collected the labels are applied to containers. Numerous opportunities exist for the wrong label to be present in the patient’s room or for containers leaving the room before labeling, both leading to mis-labeling. “The Final Check” is a process of specimen labeling which was piloted in the ED by nursing staff. The process builds on the practice of 2 identifiers used for patient identification. The Final Check occurs after specimen labeling has been completed; the collector reads out loud the last three digits of the Medical Record number from each specimen and the last three digits from the Medical Record number on the armband. Three features of this process improve the accuracy of labeling; the process must be carried out at the patient bedside, a common cause of mis-labeling when specimens are removed from the collection location prior to labeling; the “out loud” reading of the digits makes the process easily monitored for compliance; finally by reducing the comparison of demographic elements to just three digits greatly enhances compliance.

Findings to date: In the three month baseline period prior to training and implementation of the Final Check the blood specimen mis-labeling rate in the EMD was 4.5 per month. In the fisrt month post implementation the rate dropped to three mis-labels; the second month the rate dropped to two mis-labels; the third month the rate dropped to one and after 90 days the rate was zero. A zero rate of mis-labels was maintained for six months.

Lessons learned: This intervention can work in any patient area where laboratory specimens are collected. Compliance to the intervention is dependent on frequent auditing and monitoring of staff practice. Successful implementation is also dependent on a culture that supports self-reporting of errors and near misses. The evolution to a learning environment helps the adoption of this intervention.

Reducing diagnostic error through a synoptically based electronic health record [EHR]

M Gusack1.

1 Veterans Affairs Medical Center of Huntington WV, Huntington, WV

Statement of problem: Safe affordable health care requires accurate diagnosis. Today’s EHR’s merely recapitulate conventional paper-based charts and do not make effective use of capabilities provided by advanced relational database management systems [RDMS]. How do we utilize these capabilities most effectively to solve this problem?

Description of the intervention or program: I have developed a synoptic approach to structuring EHR’s that offers an effective means of organizing patient information as predefined, Synoptic Clinical Elements [SCE] that optimize data entry, retrieval, presentation, and analytical methods to reduce errors in diagnosis.

Findings to date: A model for Synoptically Based EHR has been developed as a means of testing this concept. At this time I have successfully developed and tested:

* A configurable clinical heading Extensible Markup Language [XML] structure that provides a means of establishing a systematic approach to gathering, organizing, retrieving, and presenting clinical information to the provider in a clear and unambiguous manner.

* A configurable linear synoptic data model that can be applied across the entire EHR that provides the means for defining and implementing standards of practice and diagnostic criteria without requiring a reprogramming of the underlying application or restructuring of the underlying database.

* A model configurable set of Synoptic Clinical Elements that can be used to automatically build a context sensitive set of required or recommended clinical tasks and means of automating data entry of task results for diagnostic purposes.

* A model configurable set of standardized predefined values that can be linked in a context sensitive manner to each individual Synoptic Clinical Element to provide an extremely efficient and open ended means of completing the EHR with searchable findings that are amenable to the application of standard diagnostic protocols as well as differential diagnosis engines.

Lessons learned:

* First, review of presently available EHRs indicates that the usefulness of these for diagnostic purposes is primarily dependent on the accuracy of entry, degree of organization and clarity of presentation of clinical data which is poor.

* Second, use of standardized configurable synoptic headings provides a means of assuring that critical clinical elements are grouped together for more effective compliance with necessary tasks and more accurate interpretation thereby guiding the clinical process through diagnostic work up, therapy, and management.

* Third, the implementation of a structure of standardized Synoptic Clinical Elements provides a simple linear database model that allows for closer integration of all clinical data enhancing the diagnostic process.

Reducing diagnostic error through a synoptic anatomic pathology reporting system

M Gusack1.

1 Veterans Affairs Medical Center of Huntington WV, Huntington, WV

Statement of problem: Safe health care requires accurate diagnosis. Today’s Anatomic Pathology Reporting Systems merely recapitulate conventional paper-based reporting systems and do not make effective use of advanced relational database management systems [RDMS]. How do we utilize these capabilities effectively to solve this problem?

Description of the intervention or program: I have developed a working RDMS application that implements a synoptic approach to entering, storing, retrieving, and reporting Anatomic Pathology diagnostic information that offers an effective means of applying diagnostic criteria and producing reports optimized to reduce errors in diagnosis and interpretation.

Findings to date: A working application implementing a Synoptic Anatomic Pathology Reporting System has been developed providing the means of testing this concept operationally. At this point I have successfully implemented:

* A configurable set of record based synoptic headings that provide a means of defining a systematic, process oriented approach to data entry, diagnosis, and reporting.

* A configurable synoptic data structure has been successfully applied across the entire application proving that the diagnostic process can be redefined as needed without having to reprogram the application or restructure the database.

* A configurable set of predefined Synoptic Pathology Elements [SPE] have been successfully defined for a large number of specimen types and diagnostic categories allowing for automated context sensitive linking to and population of the appropriate report heading for each accessioned case.

* A configurable set of predefined standardized values or states that can be linked in a context sensitive manner to each individual Synoptic Pathology Element that provides an extremely efficient and open ended means of completing the Anatomic Pathology Report generating findings that are optimized for the application of SQL, diagnostic protocols and differential diagnosis engines.

Lessons learned:

* First, the organization and presentation of anatomic pathology data is at least as important as its collection regarding the prevention of diagnostic error.

* Second, use of standardized configurable synoptic headings provides a means of guiding the pathologist through the diagnostic process to assure that all critical information is obtained, evaluated and reported.

* Third, implementation of a configurable synoptic structure organizing each Pathology Element provides a means of structuring the Anatomic Pathology reporting provides the means of highly automating the capture of data via predefined result entry elements that simultaneously standardize terminology and naming conventions for application of diagnostic criteria and more accurate communication of the diagnosis to providers.

Critical journal article review can reduce systemic diagnostic and therapeutic error in medicine

M Gusack1.

1 Veterans Affairs Medical Center of Huntington WV, Huntington, WV

Statement of problem: Rapid advancement in diagnostic methods combined with more effective therapeutic modalities has placed increasing pressure upon the medical community to make accurate and timely diagnoses. Yet our practice is based upon conclusions published in a growing and profligate journal literature. The complexity and obscurity of statistics used to validate journal articles have amplified the potential for establishing inherently flawed diagnostic criteria as standards of care leading to extended periods of systematic error in diagnosis. How do we avoid this serious trap?

Description of the intervention or program: I have developed a presentation in critical journal reading for our students and residents that provides a tool set for discerning if conclusions published in the professional literature are justified and what bias’ might be at work that could distort the design of research protocols, reporting of results, and formulation of conclusions.

Findings to date: A critical review of the journal literature has shown how:

* Published conclusions going back over decades are often inherently false or misleading yet accepted in the absence of critical review or successful replication by other authors.

* Despite “statistical significance” there is often no “clinical significance” to the diagnostic value of a test or procedure heralded as an improvement over prior tests or procedures.

* Many terms used by authors subtly lead the reader into a position of agreement with their conclusions regardless of their validity.

* Graphic presentations can lead the reader to draw conclusions unstated in the article and unsupported by the underlying data.

* Small scientifically unsubstantiated changes in clinical criteria advocated in the literature can lead to widespread implemented greatly affecting the number of patients diagnosed with a disease they most likely do not have.

Lessons learned:

* First, that there is a significant failing of our professional journals to effectively peer review articles submitted for publication to assure they present scientifically based content and apply sound logic.

* Second, there is a significant failing of the medical profession to subject published research protocols, statistical analyses, and conclusions to rigorous review and confirmatory replication. Instead there appears to be almost passive acceptance at face value of much of the professional journal publications.

* Third, these two failings result in the introduction of systematic diagnostic error that could be easily and inexpensively avoided all together through prospective elimination using The Scientific Method at the time of peer review and by subsequent professional challenge.

How to get it right and how to talk about it when you don’t: a longitudinal curriculum in medical decision making and cognitive error

A Olson1, E Ruedinger1.

1 University of Minnesota, Minneapolis, MN

Statement of problem: Cognitive error is common and associated with high patient morbidity. Graduate medical education programs and healthcare systems devote significantly more time and resources to discussion and analysis of system error and medical knowledge than to analysis of medical decision-making and cognitive error. Senior clinicians rarely model open discussion of cognitive error with trainees. Trainees experience negative feelings following involvement in error, but lack avenues for shared reflection upon and learning from these incidents.

Description of the intervention or program: To address this need, we designed a novel, longitudinal, integrated curriculum on medical decision-making and cognitive error with special attention to emotional safety and constructive discussion. This curriculum was piloted in early 2014 with the Pediatrics Residency Program. It included three afternoon block education sessions over six months. The first session consisted of a brief didactic presentation on the Two System Model of decision-making and cognitive error and a moderated panel discussion by respected faculty about their personal experiences with error. The second session included a writing exercise about residents’ own experience with error and small group discussion (adapted from Myers et al), and simulated patient and colleague encounters discussing error. The last session involved critical analysis of a recent case and a large group wrap up discussion. Intersession activities included the use of resident and faculty “champions” to promote dialogue about cognitive error on rounds and in teaching conferences, and readings about medical decision-making and error. An evaluation tool was administered at the beginning and end of the curriculum to assess effectiveness as well as residents’ attitudes about error.

Findings to date: Nearly all residents were able to describe personal involvement in at least one case involving cognitive error(s). Many residents reported experiencing negative emotions following involvement in these cases but also identified them as having great learning potential. Residents reported institutional responses ranging from supportive to demoralizing. At the end of the curriculum, learners stated they were more likely to reconsider diagnoses and to share uncertainty with patients and families and were more aware of cognitive biases that may influence medical decision-making. They also identified barriers present to the discussion of error, including the hierarchical nature of medical education.

Lessons learned: A longitudinal curriculum focused on medical decision-making and cognitive error is effective at engaging residents in open, constructive discussions. Integration into residents’ busy clinical workflow and modeling by senior clinicians are important for maintaining positive changes.

* Validation of an instrument to assess clinical reasoning during oral case presentations on bedside rounds

M Ottolini1, C Tuluca1, N Akar-Ghibril1, J Sestokas1.

1 Childrens National Health System, Washington, DC

Statement of problem: Diagnostic errors are a major threat to patient safety, but effective strategies to develop clinical reasoning expertise among trainees are lacking. Oral case presentations provide trainees an opportunity to engage in deliberate practice to build expertise in clinical reasoning. There are insufficient tools currently available to allow faculty to provide structured feedback to enhance clinical reasoning.

Description of the intervention or program: We established content and response validity for a tool to provide feedback to trainees on their communication and clinical reasoning skills during oral case presentations during bedside rounds. Our diagnostic reasoning framework for oral case presentations, “PBEAR” (Problem representation, Background Evidence, Analysis, Recommendation), analogous to “SBAR” (Haig, 2006), is based upon “Elaborated knowledge” using “Problem Representation” (Bordage, 1994) and “Illness Scripts”(Bowen, 2006). The “PBEAR Oral Case Presentation Tool” encourages case presentations that differ from the standard framework in that learners begin the presentation with a “Problem representation” to share their mental model at the outset. The “Background Evidence” encourages learners to filter data from EMR documentation for relevance based upon their mental model. The “Analysis” section encourages comparing and contrasting of the patient’s findings with 2-3 known illness scripts. Learners are directed to identify and explain data that doesn’t fit well. The “Recommendation” is a problem-based plan.

Findings to date: We used feedback from clinical reasoning workshops at national meetings from >100 hospitalists to establish construct validity for a draft of our tool. We then sent it to 10 pediatric hospitalists who had publications regarding clinical reasoning to assess content validity. After recommended additions, deletions or changes to the scoring rubric we piloted the tool with 10 medical students who agreed to have their presentations recorded and received feedback using the tool. After endorsement of the tool’s usefulness by students we established response validity with 10 different hospitalist educators by having them rate an oral case presentation video and then discuss their scoring rationale aloud. We further modified the tool to improve ease of use. We are in the process of establishing internal reliability using faculty ratings of 3 videos designed to depict below average, good and ideal presentations embedded in an e-Learning module to teach the “PBEAR” framework.

Lessons learned: Pilot group students appreciated the structured, specific feedback from faculty using the tool to allow them to deliberately practice improving their clinical reasoning during case presentations, while faculty found it easy to use.


*Presented in an oral session.

Inaccurate or delayed diagnosis of hollow organ perforation by emergency physicians

C Peng1, C Feng1, C Lin1.

1 Hsinchu Cathay General Hospital, Hsinchu, Taiwan

Statement of problem: Mistakes or delays in the diagnosis of hollow organ perforation may be detrimental to prognosis. Nonetheless, emergency physicians (EPs) are prone to misdiagnose this condition in specific scenarios. The factors that lead to errors in their cognitive processes, however, have received little attention.

Description of the intervention or program: Using a qualitative study approach, we conducted in-depth, semi-structured interviews with EPs in 2014. The interviews took place in the emergency departments (EDs) of 3 hospitals in Taiwan. We selected participants using a purposive sampling technique, generating a sample that would be most likely to contribute essential information about the diagnostic process. Sampling continued until novel information was no longer being gathered. All interview audiotapes were transcribed verbatim and then analyzed by 2 investigators according to the grounded theory.

Findings to date and Lessons learned: Based on 20 cases from 10 EPs, 10 themes regarding the reasons for diagnostic errors emerged from analysis: (1) EPs lacked an awareness of abdominal pain severity; (2) patients presenting with fevers were mistakenly diagnosed with other infectious diseases, such as pneumonia, acute cholecystitis, or urinary tract infections; (3) symptoms presented atypically; (4) EPs overlooked small amounts of free air in X-rays or CT films; (5) images were taken but not checked; (6) EPs arranged too many workups for abdominal pain that delayed diagnosis; (7) the investigation of underlying etiologies was neglected after patients in critical condition were initially resuscitated; (8) physicians were overloaded with too much work in a short period of time; (9) inexperienced physicians evaluated patients first without the supervision of attending physicians; and (10) multiple factors confounded diagnosis. These findings provide valuable insight into the cognitive processes of EPs. Additionally, the results offer a basis on which to build a framework for teaching physicians how to avoid misdiagnosing hollow organ perforation in the ED setting.

Diagnostic discordance in surgical pathology

M Priebe1.

1 QualityStar, Stoughton, WI

Statement of problem: Over the past 20 years, multiple studies have shown that the rate of major discrepancies identified for cancer patients referred to another institution, range from 0.6 to 15-18% (Dahl, 2006; Raab 2010). Yet published data indicates the current intralab Quality Assurance (QA) ability to detect these discrepancies is only 0.8-1.7% (Raab, 2010). This clearly identifies a gap in our current quality practice and an opportunity to improve quality assurance initiatives.

Description of the intervention or program: A new AP/QA program that is built around external case review potentially offering a new level of granularity with error detection of 1.0 to 45% (Dahl J. 2006) was introduced by abstract at the 2013 SIDM meeting. Two characteristics stand out when comparing the sensitivity of error detection between intra- and inter- laboratory case review: 1) the difference in the ability to gain incremental case scrutiny by using subspecialists for review, and 2) the difference in moving the review outside the institution to reduce on-site bias and feedback confrontation. With the current scarcity of pathologists, having subspecialists on-site is rare in the average hospital setting (3-4 pathologists), and having multiple sub-specialists to provide quality assurance peer review is extremely rare. Cases are de-identified prior to submission and cases with digital images are uploaded to a secure cloud. NCI, Academic Medical Centers provide blinded sub-specialist case review. The benefit is a standardized program that allows benchmarking at an increased level of granularity without adding to pathologist’s workload. The program is also ABP approved for MOC Level IV.

Findings to date: Over 200 slides and 100 cases from more than 20 pathologists have been reviewed by external subspecialist. Quality metrics for Diagnostic accuracy, Clerical Accuracy, Turn around Time, Diagnostic Consult and Image quality were collected. To date, 6% of cases had clinical comments but were concordant with the final diagnosis. Three percent, (3%) had significant discordance. Studies are on-going and areas of comments and discordance will be highlighted in the poster/presentation.

Lessons learned: External peer review for quality assurance in anatomic pathology appears to offer an additional level of granularity to quality assurance case review that cannot be obtained in most stand-alone laboratories. Such programs need to be confidential, automated and timely in their structure to also provide near real-time feedback and contribute to an accurate diagnosis. Additional cases and studies need to be collected and compared to expected in-house QA programs to confirm that these cases would not have been identified with current programs.

Recommendations for secondary review methods in anatomic pathology interpretive error detection and prevention

S Raab1.

1 Memorial University of Newfoundland, St. John’s, NF, Canada

Statement of problem: Although anatomic pathology practices have long used secondary review methods to detect diagnostic interpretive error, these methods are not fully standardized to detect and/or prevent error. These methods are a form of redundancy and are used at various points in the diagnostic pathway.

Description of the intervention or program: The College of American Pathologists and the Association of Directors of Anatomic and Surgical Pathology examined United States anatomic pathologist secondary review processes as an effort to improve quality practices through guideline development. A panel of 10 expert surgical pathologists performed a systematic evidence review of medical literature to address key questions. Electronic searches identified 828 articles with final data extraction performed on 152 articles of secondary review practices. The panel used Institute of Medicine methods to develop recommendations based on the quality of study evidence (i.e., strong recommendation, recommendation, and expert consensus opinion). Recommendations were derived from strength of evidence, open comment feedback, and expert panel consensus.

Findings to date: The mean secondary review discrepancy (i.e., crude agreement) frequency was 18.2% (8.2%-34.0%, 25th-75th%) and higher discrepancy frequencies were found in single organ, compared to multi-organ, and in external, compared to internal, reviews. The frequency of discrepancy that had a potential or actual impact of harm was 4.8% (1.4%-10.2%, 25th-75th%). Interobserver variability studies showed that the level of precision varied considerably by organ type, diagnostic decision (e.g., tumor type, margin status, tumor grade, etc.), level of expertise, and procedure type with kappa values ranging from.25-.87. The panel developed five recommendations and recommended (i.e., with low evidence) that: 1) pathologists should develop procedures for secondary review to detect disagreements and 2) these reviews should be performed in a timely manner. The other three recommendations were based on expert consensus opinion.

Lessons learned: The findings indicate that secondary review methods detect diagnostic interpretive error and that these reviews produce a useful measure of internal pathology practice quality. Best practices of performing secondary review as a quality improvement initiative are undetermined. Study heterogeneity limited the panel’s ability to determine sources of error, clinical significance of error detection, optimal methods of review, sensitivity of review method, and an expected range of performance.

Identification of diagnostic pathways for undifferentiated abdominal pain

G Rao1,2, K Kirley1,2, P Epner3, A Solomonides1, J Silverstein1, Y Zhou1.

1 NorthShore University Health System, Evanston, IL

2 University of Chicago Pritzker School of Medicine, Evanston, IL

3 Paul Epner LLC, Evanston, IL

Statement of problem: Undifferentiated complaints in primary care such as abdominal pain, headache, and chest pain present a significant challenge and are frequently associated with diagnostic error. This is at least partly due to the lack of formal guidance on diagnostic approaches to such complaints. Analysis of EHR data from a large number of patients has the potential to reveal patterns of diagnostic evaluation associated with more timely diagnosis, lower costs, lower morbidity and mortality, and greater patient satisfaction. Our study is designed to develop and apply a methodology to describe diagnostic pathways (i.e. steps taken to establish a diagnosis) for patients with undifferentiated abdominal pain.

Description of the intervention or program: The 2 aims of this AHRQ supported project include: (1) Identification of diagnostic pathways for adults presenting for the first time with undifferentiated abdominal pain. Pathways include diagnostic testing, prescriptions, procedures, referrals, and follow-up recommendations. We created a detailed template for describing diagnostic pathways and extracted and inserted EHR data into the template from an enterprise data warehouse (EDW) for 500 patients. Manual chart reviews will be used to extract some data from free text notes by physicians and to validate some data elements from the EDW. We will analyze the diagnostic pathways we generate in various ways, including for example, which tests are ordered in which sequence in which types of patients, and which patients are likely to follow specific pathways (e.g. empiric treatment with medication prior to referral to a gastroenterologist); (2) Patients and diagnostic pathways attributable to individual physicians will be available. We will interview roughly 20 physicians about the reasons for specific diagnostic patterns. Prompts will include, for example,

“You seemed to order test Y frequently for your patients. What do you see as the value of this test?”

Findings to date: To date, as noted, we have created our structured template and successfully extracted all available and necessary EDW data for 500 adult patients. Furthermore, our template has been used to store the data in a structured RedCap database. We are in the process of completing manual chart reviews.

Lessons learned: Creating a template for undifferentiated abdominal pain which captures the diversity of diagnostic approaches to the problem proved extremely challenging. EDW data elements needed significant translation to fit our structured template. Nevertheless, our template data extraction procedures represent a key initial step in establishing a methodology for addressing the problem of identifying best approaches to undifferentiated complaints.

Display of diagnostic test results in electronic health records (ehr): implications for diagnostic decision-making

D Sittig1, D Murphy2, M Smith2, E Russo2, A Wright3, H Singh2.

1 University of Texas, Houston, TX

2 Michael E. DeBakey Veterans Affairs Medical Center and Baylor College of Medicine, Houston, TX

3 Brigham & Women’s Hospital, General Boston, MA

Statement of problem: Interpretation of laboratory results is an important component of diagnosis. With wider implementation of electronic health records (EHRs), computerized displays allow data to be aggregated, organized and visually trended over time with relative ease. Graphing patterns could provide important diagnostic clues, such as a downward hemoglobin trend with gastrointestinal bleeding, increasing creatinine with renal failure on NSAIDs, or a rising PSA level suggestive of prostatic disease. However, little is known about whether graphical displays of laboratory data in EHRs are optimized to enable diagnostic decision-making. We evaluated several current EHR user interfaces on how they displayed laboratory test results graphically and if this had any implications for diagnostic decision-making.

Description of the intervention or program: We used a convenience sample of similar laboratory results graphs from 8 EHRs; 7 certified (i.e., by Office of the National Coordinator for Health Information Technology authorized certification and testing bodies (ONC-ACTB)) and one prototype EHR. We developed criteria for optimal graphs based on literature and expert input. Criteria included: Title – succinct description of the graph’s contents, including analyte(s) being graphed; Patient’s name/age/gender; X and Y axis labels to explain meaning of the units; Evenly-spaced tick marks with labels indicating their numerical value; Normal range(s) shown for the analyte(s); Legend-if more than one analyte depicted; and Details of data points (e.g., x-y coordinates).

Findings to date: None of the EHRs reviewed satisfied all criteria. Two-thirds of EHRs did not have x- or y-axis labels and several did not even have tick marks let alone labels for intermediate data points (i.e., only first and last values were displayed). One system graphed results in reverse chronological order. Several systems plotted data collected at unequally-spaced points in time using equally-spaced data points, which had the effect of significantly changing the slope perception between data points. Only one system allowed the user to see or “drill-down,” hover-over, or click on a data point to see the actual x-y coordinates. Several EHRs did not display patient’s ID on the graph, although it was available in the background.

Lessons learned: Current EHRs have significant limitations in their graphing capabilities of laboratory results and display results in non-standardized haphazard fashions. Some graphs, such as those using reverse chronological order, could be particularly confusing to users, and have serious implications for diagnostic decision-making related to test interpretation. Future ONC-ATCB EHR certification tests should include a review of capabilities for graphing different types of numeric, clinical data.

* Video review produces insight into diagnostic errors

L Su1, S Kaplan2, M Waller3.

1 Children’s National Medical Center, Washington, DC

2 George Mason University, Fairfax, VA

3 York University, Toronto, ON, Canada

Statement of problem: Current standards in event review and performance improvement utilize retrospective analysis and participant memory to reconstruct events and identify missed opportunities. We know from research in other domains that eyewitness accounts are flawed.

Description of the intervention or program: In December 2011, for quality improvement purposes, video cameras were placed in every room in our 26-bed pediatric Cardiac Intensive Care Unit. Only 1 room (which also functions as a simulation room) has audio capability. These videos are stored on a password-protected server for 20-30 days (based on available server space) and then automatically deleted. The videos have a privacy mode which family members and staff are empowered to activate when desired. Our video review program has evolved to our current program which involves one on one “coaching sessions” where any staff member can review any event with a “coach” of his/her choosing. Coaches are made up of Critical Care Faculty and Nursing Educators. In addition, all serious events identified by staff for our weekly critical event review are reviewed by a core group of 2 critical care faculty and 2 nurses (1 nurse practitioner and 1 nurse educator). This information is used to clarify event timelines. After explicit consent is obtained from all parties in videos, selected events thought to have especially revealing or insightful behaviors are shown to staff.

Findings to date: Since its inception, we have discovered several trends in human behavior patterns and have changed our educational initiatives to address these: 1. The power of expectation- Unexpected cardiac arrest leads to a delay in chest compressions of up to 7 minutes even when video and monitor evidence of cardiac arrest is clear. Repetitive behaviors such as listening to breath sounds and checking for a pulse are done while bag mask ventilation is started but there seems to be a psychological hurdle difficult to overcome when cardiac arrest is unexpected. 2. Bystander effect- When other staff enter the room for a non patient related purpose, the primary caregiver is falsely reassured if those staff do not express concern. Staff entering the room do not realize they are providing reassurance. 3. Distinguishing an abnormal rhythm from a baseline abnormal rhythm is difficult for the non-expert. Malignant arrhythmias including ventricular tachycardia are difficult to diagnose when the baseline arrhythmias are not familiar for the staff.

Lessons learned: Video review allows us to gain insights into the behaviors of medical personnel in the midst of a medical crisis. This deserves further investigation.


*Presented in an oral session.

Multiple presentations to the northern emergency department: are we missing something?

K Zarour1.

1 The University of Melbourne, Melbourne, Australia

Statement of problem: The Northern Hospital Emergency Department (ED) is one of the busiest in Victoria with a rapidly expanding catchment area along with increasing numbers of multiple presenters (MPs). Australian studies estimate MPs to account for 1.4-4% of all ED presentations which places a huge burden on the public healthcare. MPs are also at a higher risk of missed, delayed and misdiagnoses.

Description of the intervention or program: To estimate the frequency of MPs and to explore how emergency and inpatient data can be used to identify diagnostic error or delay in this patient group. Methodology This is an exploratory retrospective cohort study of all adult (>18) patients who presented to the Northern ED 4 or more times over the 12 month period June 2011 to July 2012. Patients younger than 18 or adult patients who required a mental health consultation were excluded. Patients who were admitted to the Northern Hospital at least once were selected. Patients with at least 3 ED presentations prior to their first admission were selected for further investigation. A misdiagnosis was established if the ED diagnosis (ICD 10 –AM) code did not match the inpatient diagnosis code. The data used is the routinely collected Victorian Emergency Minimum Dataset (VEMD) with the approval of the Northern Health Low Risk Ethics Committee. SPSS version 22 and Microsoft Excel 2007 were used for data analysis.

Findings to date: A total of 38,293 adult patients resulting in 53,818 ED visits to the Northern ED were observed during the 12 month investigated period. 1,224 patients (3.2%) were identified as MPs with a range of 4 to 40 ED presentations per patient, and 1,019 patients were admitted at least once. Of those, 141 fulfilled the selection criteria of at least 3 ED visits preceding the first admission. 19 (13.5%) had a misdiagnosis and 59 (41.8%) had a delay in diagnosis. The most common ED diagnoses in this group were abdominal pain (12.1%), chest pain (8.5%) and dyspnoea (5%).

Lessons learned: Patients use EDs for a range of health emergencies. Multiple presentations may identify gaps in the care system. In a proportion of cases, repeat presentations may signal acute conditions that remain unresolved. By focusing on patients with little evidence of mental health conditions, we hope to estimate the proportion of MPs with a delayed or missed diagnosis subsequently treated in hospital.


Radiologists are the new clinicians

J Akhtar1, J Fei1, J Roth1.

1 University of Pittsburgh Medical Center, Pittsburgh, PA

Learning objectives: Patients history and the physical have traditionally been the pillars of medical diagnosis. However, in modern day medicine, greater emphasis is being placed on early imaging. We report a case where it was the radiologist who suggested the diagnosis.

Case information: A 19-year-old man with no significant past medical history presented with a few days history of a pleuritic lower right chest/right upper quadrant pain and fever. His vitals were unremarkable and the initial physical examination was normal except for tenderness in the right upper quadrant/epigastric region. The initial workup showed an abnormal liver function test and elevated D-dimer. The total bilirubin was 1.1 mg/dL with an ALT of 296 IU/L, AST of 155 IU/L and alkaline phosphatase of 252 IU/L. The D-dimer was high at 1.19. Based on these tests an ultrasound of the right upper quadrant was done which showed no evidence of a cholelithiasis and a CTA of the chest showed no evidence of a pulmonary embolism. However, on the CTA the radiologist noted axillary lymphadenopathy and splenomegaly. The radiologist raised the concern for infectious mononucleosis. The monospot test subsequently sent was negative, but the EBV IgM antibody was positive confirming the diagnosis of infectious mononucleosis induced hepatitis.

Discussion: Clinical history and examination has been the cornerstone of medical diagnosis. Clinical examination is becoming less important as the number and quality of diagnostic tests improves. Use of imaging will lead to better care, more prompt diagnosis and improved outcomes. History and physical is dying. Radiologists are assuming a greater role and will be the new clinicians.

Miliary tuberculosis mimicking peritoneal metastasis

N Ando1, J Ushio1, T Kenzaka1, M Matsumura1.

1 Jichi Medical University Hospital, Tochigiken, Japan

Learning objectives: Tuberculosis is still great masquerader.

Case information: An 85 year-old man with postherpetic neuralgia presented with anorexia and weight loss. Two months prior, at regular visit to a pain clinic, blood test showed white cell count 5,700/?L, hemoglobin 12.3 g/dL, and platelet count 67,000/?L. Fecal occult blood test was positive but abdominal ultrasonography and gastroscopy revealed no abnormalities. Four weeks prior, anorexia, and weight loss developed. He had 5 kilogram of weight loss for 3 weeks. On admission, the patient seemed chronically ill with temperature of 38.5C. His abdomen showed no tender or guarding. Blood tests showed white cell count 2,000/?L, hemoglobin 11.0 g/dL, platelet count 26,000/?L, and erythrocyte sedimentation rate 47 mm per in one hour. Chest X-ray showed mild cardiomegaly. Enhanced computed tomography (CT) revealed ascites, mediastinal and abdominal lymph node enlargement, nodular wall thickening of peritoneum which were highly suggestive of peritoneal metastasis. However, adenosine deaminase of ascites was 139 IU/L and Tb-PCR in sputum was positive. We diagnosed military tuberculosis. Also, bone marrow aspiration led to the diagnosis of myelodysplastic syndrome (MDS) as the cause of pancytopenia. He was started on anti-tuberculous agents. Follow-up CT revealed improvement of nodular wall thickening and no ascites.

Discussion: Japan is still intermediate prevalence country of tuberculosis. Diagnosis of tuberculosis requires a high index of suspicion. In this era, CT brings substantial information of patient. Although CT findings strongly suggested peritoneal metastasis from unknown primary site, we have to consider the cause of weight loss other than neoplasm. In this case, cellular immunodeficiency is a risk factor for tuberculosis which resulted from age and MDS. One case series reported the percentage of extrapulmonary tuberculosis was 50% in patients with MDS1). In conclusion, tuberculosis is still great masquerader.

Pulmonary embolism mimicking anteroseptal acute myocardial infarction

P Charoenpong1, K Shahwan1, N Poowanawittayakom1, C Clarke1.

1 Advocate Illinois Masonic Medical Center, Chicago, IL

Learning objectives: To recognize that pulmonary embolism can mimic acute coronary syndrome in term of clinical presentations and EKG findings. Echocardiography is a helpful tool to differentiate these conditions.

Case information: A 55-year-old male was brought to the hospital after he was found unconscious at home. In the field, he was tachypneic, tachycardic, and hypotensive with a systolic blood pressure in the 60’s. On arrival, he was alert but was diaphoretic and visibly dyspneic. He then became bradycardic and went into cardiac arrest with pulseless electrical activity. The ACLS was implemented; he eventually regained a pulse and consciousness following 2 doses of epinephrine. His EKG showed ST-segment elevation(STE) in leads V1-V4, RBBB. The patient then underwent coronary angiography for a presumptive diagnosis of acute myocardial infarction, however this revealed normal coronary arteries without significant blockages. Subsequently, an echocardiogram was performed which revealed a severe dilatation of the RV with an elevated RV systolic pressure of 50-60 mmHg, severe dilatation of the RA, and a hyperdynamic LV. These findings were suggestive of an alternative diagnosis of APE. A CT chest with contrast confirmed the diagnosis of a massive APE. He was started on alteplase and a heparin drip and was transferred to ICU.

Discussion: STE on EKG can be seen in several conditions including APE and ACS. Studies suggest that STE in the anterior leads in APE is the result of transmural ischemia in the RV and subendocardial ischemia in LV as a result of hypotension, hypoxemia, RV strain, and surge of catecholamines. This case demonstrates the difficulty to differentiate between APE and ACS. In circumstances which the diagnosis is equivocal, echocardiography is an important tool to demonstrate changes in pulmonary artery pressures or RV function, or the presence of a thrombus in the pulmonary artery in cases of APE.

* Thinking outside the ‘orbit’: a long painful ordeal

J Fei1, J Akhtar1.

1 University of Pittsburgh Medical Center, Pittsburgh, PA

Learning objectives: To identify difficult and unconventional presentations in order to prevent misdiagnosis.

Case information: A 51 year old male with no significant past ocular history presents to the emergency department for evaluation of right eye pain, blurry vision and headaches for 5 months. He had been diagnosed with conjunctivitis, migraines, sinusitis and biopsied for temporal arthritis. Several courses of NSAIDs, prednisone and Augmentin were provided to the patient with minimal clinical improvement. MRI orbit and CT head were unremarkable in the previous month. Physical exam was significant for periorbital erythema and trace injection in the right eye with normal fundoscopy. The physicians pivoted on the periorbital pain and eye redness to suspect a cavernous sinus syndrome. MRA/MRV was significant for superior ophthalmic vein dilation and enhancement of the right cavernous sinus, which was concerning for carotid cavernous fistula. This was confirmed by cerebral angiogram, and transvenous embolization of the fistula was performed with resolution of his symptoms.

Discussion: Uncommon presentations of diseases are difficult to detect and often are not identified in differential diagnoses. Carotid-cavernous fistulas can present with a variety of nonspecific symptoms including orbital erythema, headaches, and vision loss. The ability to think outside the box is critical to avoid missing this diagnosis particularly in patients whose symptoms can easily be attributed to other causes resulting in misdiagnosis. Several ophthalmologists, neurologists and surgeons had already evaluated this patient with no definite diagnosis established. In patients who have returned multiple times, keeping an open mind and recognizing our biases can help reduce these errors and improve diagnosis.


*Presented in an oral session.

Anchoring...do not get stuck, let it go

D Ford1, J Akhtar1.

1 University of Pittsburgh Medical center, Pittsburgh, PA

Learning objectives: Anchoring is the tendency to lock onto salient features in the patient’s initial presentation too early in the diagnostic process and failing to adjust this initial impression in the light of later information.

Case information: A 84-year-old man with a past medical history of asthma, prior pneumonia, MRSA empyema status post VATS in 2008 presented to the emergency department with shortness of breath over the course of the last 2 weeks. He had a progressively worsening dyspnea on exertion. He did not have any chest pain. Physical examination revealed him to be tachypneic with a respiratory rate of 22 per minute and a room air saturation of 92%. The pulmonary and cardiovascular examination was normal. A chest x-ray showed left basilar atelectasis. The EKG examination showed nonspecific ST-T wave changes in the inferolateral leads which were new. His initial troponin was elevated at 0.31ng/ml. The physicians anchored on the non-specific EKG changes and did not adjust when the chest xray was negative for pulmonary edema. The non-specific troponin confirmed their initial bias. The patient was admitted to the hospital with a diagnosis of coronary artery disease. The floor team was concerned for pulmonary embolism because of the the pronounced shortness of breath. A d-dimer was sent and came back elevated at 18.50 mcg/ mL. A V/Q scan (and not a CTA of the chest because of acute kidney injury) was read as high probability for pulmonary embolism.

Discussion: Anchoring bias is often considered the Achilles’ heel of diagnostic reasoning. It’s as though our brains close ranks around our first impression, then refuse to consider anything else. Greater self-awareness of the process of your own cognition can make us better clinicians.

A cause of sudden onset drenching

S Gomi1, K Matsumoto2, M Kumagai1, K Sekiguchi1, Y Tokuda3.

1 Shinshu University Hospital, Matsumoto, Japan

2 Osaka National Hospital, Osaka, Japan

3 Japan Community Healthcare Organization, Minatoku, Japan

Learning objectives: To consider spontaneous hepatoma rupture for a cause of acute blood loss without evidence of gastrointestinal bleeding

Case information: A 70-year-old man with a known history of hepatitis C presented with sudden onset drenching and mild back pain. He had been regularly checked sonographically but lost to clinical follow-up for 2 years. No ascites, cirrhosis, or masses had been documented. Vital signs on admission were afebrile, BP 88/58, and HR 101 but no tachypnea. He was anemic and soft systemic murmur was heard at the apex, but other physical examinations were normal including negative fecal occult blood test with digital rectal examination. Laboratory tests, ECG, chest-X ray and cardiac and abdominal ultrasound were conducted, showing un-contributory results. For investigating the cause of shock, abdominal CT without contrast showed hemoperitoneum and intraabdominal hemorrhage from a spontaneous rupture of undiagnosed hepatocellular carcinoma (hepatoma: Image). Vital signs normalized soon after rigorous intravenous fluids. Interventional radiology team was consulted and trans-arterial embolization successfully performed for controlling acute bleeding, leading to clinical stabilization. AFP and PIVKA II were elevated but HCV-RNA was not detectable.

Discussion: Spontaneous rupture of hepatoma has been occasionally recognized in HCC patients in Japan and approximately 10% of these patients have died due to this complication. This can be the first manifestation as a sudden onset shock state in an otherwise asymptomatic individual. However, its diagnosis is frequently missed or delayed because it is not considered by physicians without prior knowledge of this complication. Vergara et al reported that an accurate pre-operative diagnosis was obtained in only 25% of cases, despite shock present in 33-90% of the patients. Hepatoma rupture should be considered in patients with chronic hepatitis or cirrhosis for a cause of acute blood loss despite no evidence of gastrointestinal bleeding.

Sutton’s Slip: all that flashes is not floaters

L Graham1, J Akhtar1.

1 University of Pittsburgh Medical Center, Pittsburgh, PA

Learning objectives: Most times the obvious answer is the correct answer. We report a case where focusing on the obvious led to a misdiagnosis.

Case information: A 44 year-old woman presented to the Emergency Department (ED) with acute onset of intermittent gray spotty vision and bright central flashes of light that made it difficult for her to read her computer screen. She was unable to localize the eye involved. She also noted 4 days of right-sided neck stiffness as well as a mild right temporal headache. Her symptoms had resolved at the time of evaluation in the ED. She had a normal physical exam, and a dilated eye examination by ophthalmology showed no evidence of retinal or vitreous pathology. She was diagnosed with floaters and discharged home. She represented nine days later to the ED complaining of feeling off-balance and dizzy. She had no visual symptoms, but continued to endorse mild right-sided head and neck pain. Her neurologic exam was normal without any evidence of cerebellar dysfunction. Her cluster of symptoms was concerning for posterior circulation stroke, and MRA of the neck confirmed right vertebral artery dissection without MRI evidence of infarct. A diagnostic angiography confirmed the diagnosis.

Discussion: The bank robber Willie Sutton is alleged to have told the judge that he robbed banks, “because that’s where the money is!”. Sutton’s Law suggests that ‘obvious’ diagnoses more often explain symptoms than ‘non-obvious’ ones. The Sutton’s Slip occurs when possibilities other than the obvious are given inadequate consideration. Perhaps the best way to summarize the simplicity of this diagnostic error is to quote Sherlock Holmes: “Nothing is more deceptive than the obvious.”

A case of delayed diagnosis of Hodgkin’s lymphoma

B He1, T DeJulio1, Y Chen1, A Petrich1.

1 Northwestern University, Chicago, IL

Learning objectives: 1. Inadequate biopsies may delay lymphoma diagnosis. 2. Repeat biopsies should be strongly considered when unconfirmed lymphoma remains in the differential diagnosis.

Case information: A 19 year old previously healthy male presented in October 2012 with flu-like symptoms and joint pains in his hands. Evaluation revealed leukocytosis, a mediastinal mass, and cervical lymphadenopathy. Lymph node biopsy showed reactive changes. He was diagnosed with adult onset Still’s disease (AOSD) in December 2012 and treated with high dose steroids and serial immunomodulators through July 2013. His symptoms waxed and waned but never fully resolved. Transthoracic mediastinal mass biopsy in September 2013 showed fibrous tissue with atypical lymphoid infiltrate, but definitive diagnosis of lymphoma was not possible given inadequate sampling/crush artifact. He presented to our medical center in November 2013. Video assisted thorascopic surgical biopsy yielded adequate tissue with well preserved architecture. A diagnosis of classical Hodgkin lymphoma (cHL) was made based on microscopic appearance and immunohistochemistry. He consented to participation in a randomized trial of doxorubicin/bleomycin/vinblastine/dacarbazine (ABVD; Arm A) versus doxorubicin/vinblastine/dacarbazine plus brentuximab vedotin (AVD plus BV; Arm B). He was randomized to Arm B, completed six cycles, and achieved rapid symptom resolution.

Discussion: This case represents a delayed diagnosis of cHL from misdiagnosed AOSD. Malignancy can manifest with rheumatologic features such as arthropathy, skin rash, and elevated inflammatory markers. Lymphoma can overlap with rheumatologic disorders. In retrospect, the presence of a large mediastinal mass, along with the poor response to AOSD-directed therapies, suggested a different underlying process. Importantly, this case highlights the problem of inadequate tissue sampling. Non-definitive diagnoses with fine needle aspiration can occur. Open biopsy may also be nondiagnostic from inadequate sampling and/or crush artifact, which can distort cells. Given inadequate sampling, repeat open sampling may significantly increase diagnostic yield. Clinicians should consider frequent re-evaluation of underlying pathology in cases where response is suboptimal.

Figure 1 Chest x-ray shows large anterior mediastinal mass and moderate left pleural effusion.

Subacute bacterial endocarditis associated with proteinase-3-antineutrophil cytoplasmic antibody positive serology

H Inada1, S Nishiguchi1, Y Tokuda2, J Branch1.

1 Shonan Kamakura General Hospital, Kamakura, Japan

2 Japan Community Healthcare Organization, Minatoku, Japan

Learning objectives: To appreciate that subacute bacterial endocarditis (SBE) can cause glomerulonephritis with positive proteinase-3-antineutrophil cytoplasmic antibody (C-ANCA).

Case information: A 62-year-old Japanese male presented to the nephrology department of our hospital with a one month history of albuminuria and hematuria. C-ANCA was found to be elevated at 64.6 IU/l, and myeloperoxidase-antineutrophil cytoplasmic antibody (P-ANCA) was negative. On the fifth hospital admission day immunosuppressive treatment was planned after a renal biopsy due to suspected ANCA-associated vasculitis. However, after his admission he developed a fever, a diastolic murmur and splinter hemorrhages in the nails of his hands. Urgent transthoracic echocardiography revealed a vegetation on the aortic valve prior to proposed corticosteroid treatment. As a result of the physical examination and associated tests, SBE was considered as the likely diagnosis, and antibiotic therapy was commenced. Streptococcus mutans was eventually grown from the blood cultures thereby confirming the diagnosis. The patient underwent aortic valve replacement 11 days into his admission which proved successful. He continued intravenous antibiotics for a total of six weeks with a good clinical outcome. The patient’s albuminuria and hematuria also improved and follow-up C-ANCA testing was negative.

Discussion: Positive immunofluorescence findings such as C-ANCA in a systemic disease suggesting vasculitis can lead to a diagnostic error in which potentially fatal treatment, such as corticosteroids, could have been administered to a patient who was later reassessed to have SBE. It is therefore important to consider other diagnoses that must not be missed when such serology is found to be positive, and to explore the cause of glomerulonephritis from a detailed history and physical examination, in addition to appropriate testing. Immunosuppressive agents should only be commenced in cases strongly suggesting vasculitis in which a delay in such therapy can be detrimental to the patient, but only after infection has been rigorously excluded.

Radiology misreading followed by overconfidence bias

K Kamata1, Y Tokuda2.

1 Tsukuba Medical Center Hospital, Ibaraki, Japan

2 Japan Community Healthcare Organization, Minatoku, Japan

Learning objectives:

Careful double checkup for prior clinical results is recommended for avoiding overconfidence bias.

Case information: A 64-year-old Japanese man was transferred to our hospital for a 3-week history of fever of unknown origin (FUO). He had been in his usual state of health until 3 weeks ago, when he developed a sudden-onset chest and back pain and he was admitted to another hospital by ambulance. He underwent chest X-ray, ECG, abdominal and cardiac ultrasound exam, and whole-body CT scan without contrast, for which radiologist and cardiologist reported these were normal. However, he had persistent chest and back pain and developed fever of 38 Celsius degrees, which did not respond to multiple antibiotics and steroids. He received cardiac and abdominal ultrasonography, another whole body CT scan without contrast, thoracic MRI scan without contrast, but radiologists reported that these were all normal.

On exam he was alert and appeared well with blood pressure of 178/110, the conjunctivae were anemic and peripheral pulses were equally palpable. Because the reports of prior radiologists suggested normal results for all imaging exams, we did not reexamine the imaging results carefully. However, in the second week of admission, we rechecked laboratory exam, which revealed renal severe acute kidney injury and thus, the reexamination of the previous CT scan was finally conducted, revealing a type B acute aortic dissection (AAD) with intramural hematoma and he underwent successful vascular surgery.

Discussion: Causes of FUO are broad and include vascular events such as AAD. Diagnosis of AAD is usually made by radiologic tests such as CT or MRI scans. However, these scans can lead to false negative results if they are not conducted with contrast enhancement. Physicians who over-rely on radiologist reports may make misdiagnosis by overconfidence bias, which could be safely avoided by careful re-checkup.

Anchored to an uncomfortable diagnosis-pivot and cluster strategy revisited

I Kitagawa1, T Shimizu2, S Sato1, S Tomiyama1, T Watari1, S Nishiguchi1, K Matsumoto3, Y Tokuda4.

1 Shonan Kamakura General Hospital, Kamakura, Japan

2 University of Hawaii, Honolulu, HI

3 Osaka National Hospital, Osaka, Japan

4 Japan Community Healthcare Organization, Minatoku, Japan

Learning objectives: Application of the Pivot and Cluster Strategy(PCS) will enable to diagnose shwannoma as a differential diagnosis of axillary nodes, mimicking lymph nodes.

Case information: A 39 year-old man with no significant history presented with two-week history of a non-tender node in his left axillae. He also admitted that he had had an left inguinal node without change in size for 3 years. He denied any recent sexual contact, animal contact or sick contact. On exam of the axillae, a soft, non-tender node with approximately 3 cm in size was palpated. Ultrasonography revealed a well-defined, 3 cm node suggesting a lymph node as a most likely possibility. The physician who saw this patient suspected a possibility of malignancy and planned biopsy of the node. After the biopsy, the patient developed radial nerve palsy. The biopsy result showed shwannoma, not lymph node.

Discussion: Cognitive bias should be avoided in every diagnostic case. Especially in cases with diagnostically-challenging case, the cognitive bias is one of the key factors of diagnostic errors. This case exhibits an example of misdiagnosis causing irreversible conditions, also showing the process of a typical cognitive error-anchoring. The physician in this case was anchored to an intuitive diagnosis of lymph node, however the true diagnosis was not what he expected. In such a case, de-biasing measures will play a key role to fill the gap between correct and incorrect diagnosis. Here we emphasize the usefulness of a cognitive de-biasing strategy, Pivot and Cluster Strategy, which can be universally applied to avoid unexpected misdiagnosis, improving diagnostic accuracy and quality of patient care.

A delayed diagnosis of type II necrotizing fasciitis in a physician

R Manesh1, L Tierney Jr2.

1 University of California, San Francisco, San Francisco, CA

2 UCSF-VA, San Francisco, CA

Learning objectives: To recognize type II necrotizing fasciitis.

Case information: A 50-year-old physician presented with nine days of waxing and waning fevers, chills, and sweats. Nine days earlier he developed sore throat which resolved spontaneously. Four days prior to presentation, he suddenly developed right thigh pain while playing tennis. Later that night he developed fever, night sweats, and chills. On presentation his only complaints were fever and right thigh pain. Physical exam revealed a drenched hospital gown, temperature 39.5°C, heart rate 115 beats per minute, and point tenderness of the anterior-mid-right-thigh without overlying crepitus, erythema, or edema. Labs revealed sodium of 129 mmol per liter, bicarbonate 20 mmol per liter, white-cell count 12,900 per cubic millimeter, and normal creatine kinase. An ultrasonogram of the thigh showed fluid between the right rectus femoris and vastus lateralis muscles. He was admitted to the hospital for observation. Twelve hours later his blood pressure decreased to 81/53 mmHg. Lactate was 3.3 mmol per liter, white-cell count 33,000 per cubic millimeter (96% neutrophils), and C-reactive protein 174 mg/L. A few hours later the right thigh became erythematous and edematous. Broad spectrum antibiotics were administered and the patient was taken to the operating room where the fascia was easily lifted off the right thigh muscles with expression of turbid fluid. Necrotizing fasciitis (NF) was diagnosed. The tissue was debrided and the patient recovered. The fascia cultures grew group A streptococcus.

Discussion: The patient likely first acquired group A streptococcal pharyngitis leading to bacteremia, which then seeded the injured thigh resulting in NF. The delay in diagnosis was from the juxtaposition of necrotizing fasciitis against two far more common conditions: viral syndrome and mechanical injury. The “eyeball” test was the tie-breaker, and the reassurance of the interactive patient was given more weight (perhaps errantly) than profound diaphoresis.

Mitral stenosis without opening snap

K Matsumoto1, Y Kurose2, Y Tokuda3, K Mitsunami4.

1 Osaka National Hospital, Osaka, Japan

2 Shiga University Of Medical Science Hospital, Otsu, Japan

3 Japan Community Healthcare Organization, Minatoku, Japan

4 Vories Memorial Hospital, Omihachiman, Japan

Learning objectives: Another important differential diagnosis of mitral stenosis.

Case information: A 46-year-old obese woman presented with a 3-day history of intermittent left anterior chest pain, which is exacerbated by deep inspiration. She had palpitation and dyspnea on exertion for the last 3 months but denied fever, cough, wheezing, hemoptysis, or edema. Two days ago she went to a local ER due to chest pain and sent home based on normal ECG and Chest X ray results. Vitals signs on admission were afebrile, BP 125/97, HR 68, and pulse oximetry oxygen saturation of 98% while breathing ambient air. There were loud S1 and a presystolic accentuated murmur of Levine II/VI with rumbling which was loudest at the apex without opening snap. The remainder of other physical examinations was normal.

Mitral stenosis or pulmonary embolism was suspected. However, transthoracic echocardiography showed a huge (3.7 x 5.1 cm) isoechoic mass attached at the left atrial septum. A chest CT showed a consolidation at left subpleural lingular segment. Chest surgery team was consulted and successfully performed left upper lobectomy and left atrial myxomectomy of massive myxoma. Pathological exam of the consolidation revealed organizing pneumonia. She completely recovered without any symptoms.

Discussion: Opening snap is a sharp, short and high pitched sound best heard at apex and produced by mitral cusps downing into left ventricle at the end of isometric relaxation period. Mitral valve stenosis is usually suspected if diastolic rumbling murmur coexists with the opening snap, but otherwise the obstructing thrombus or left atrial myxoma should be considered. Manifestations of cardiac myxoma are typically related to obstructive valvular or embolic events and 10-15% of patients are asymptomatic, and thus diagnosis would be often made incidentally or delayed until severe symptoms occur or the tumor occupies much of the left atrium.

Bloody diarrhea post-antibiotics: holding on to certainty despite an uncertain diagnosis

L Medford-Davis1, H Singh1,2.

1 Baylor College of Medicine, Houston, TX

2 Michael E. DeBakey Veterans Affairs Medical Center, Houston, TX

Learning objectives: Analyze a failure to recognize diagnostic uncertainty causing diagnostic delay.

Case information: A 38 year-old previously healthy female presented to the Emergency Department complaining of 10 days of bloody diarrhea, 5-10 episodes per day with diffuse abdominal pain but no fevers, nausea, or vomiting. An outside clinic prescribed her macrobid and ciprofloxacin for a UTI 2 weeks prior, and the diarrhea began 1 day later. She had then returned to the same clinic, and was given metronidazole, anaspaz, and imodium. However, the diarrhea had persisted. Significant ED vitals: T100.1 HR 120 BP 114/71. Labs showed Na 130, iCal 0.94 and WBC 12.5. C.Difficile toxin was negative with other stool studies pending. She was discharged home from the ED with the presumptive diagnosis of Clostridium Difficile infection resistant to metronidazole, and was prescribed oral vancomycin. Three days later she returned to the ED with no improvement, unable to afford vancomycin. Labs showed Na 133, K 2.9, WBC 10.6. She was admitted for IV hydration and further workup. An inpatient colonoscopy revealed the correct diagnosis of Ulcerative Colitis with no C.Difficile infection.

Discussion: A return visit for similar or worsening complaints should prompt clinicians to reconsider the original diagnosis and treatment plan. In this case, the ED physician presumed the patient had been diagnosed with C.Difficile based on her report of receiving a prescription for metronidazole soon after antibiotic use. This was likely due to framing bias. Despite evidence against it, physicians continued to anchor on a C.Diff diagnosis. If uncertainty of this patients’ diagnosis had been recognized, reflective strategies such as documenting a differential diagnosis and seeking help, as well as mitigating strategies such as a close follow-up plan, would have been beneficial. Further research should investigate how frontline physicians recognize and deal with diagnostic uncertainty and if this recognition could mitigate diagnostic error.

Inattentional blindness: case of a missing tooth

K Moretti1, J Akhtar1, M Hsieh1.

1 University of Pittsburgh Medical Center, Pittsburgh, PA

Learning objectives: Inattentional blindness is an inability to perceive something that is within one’s direct perceptual field because one is attending to something else. This phenomenon suggests that what we look for determines what we see.

Case information: A 36-year-old man presented to a trauma center after he crashed his car onto the side of a bridge, and then jumped over the bridge to avoid police. He was found approximately 30 feet below with facial trauma and bilateral wrist deformities. He was intubated in the field. His oromaxillofacial examination showed evidence of significant facial injury with bilateral periorbital ecchymosis and mobility of the maxilla and mandible with multiple lacerations and gingival tear. There was active bleeding from the mouth, with loose and missing teeth. His CT scan of the face showed him to have extensive complex facial fractures. His chest x-ray showed no evidence of acute traumatic injury. The CT scan of the chest showed a tooth in the left main bronchus. The chest x-ray was reviewed, and clearly showed the tooth missed on the initial interpretation.

Discussion: Inattentional blindness is defined as the event in which an individual fails to recognize an unexpected stimulus that is in plain sight and can occur independent of cognitive deficits. Radiologists are especially vulnerable to inattentional blindness. They receive limited information and may do their job better if they have a fuller understanding of a patient’s clinical history. Awareness of pitfalls around diagnostic errors will help us identify those teeth!

Double-edged Occam’s Razor

M Ono1, Y Tokuda2.

1 Social Welfare Corporation Tokyoyurinkai Yurin Hospital, Setagaya-ku, Japan

2 Japan Community Healthcare Organization, Minatoku, Japan

Learning objectives: To understand diagnostic difficulty for a multi-organ disease in patients with other comorbidity.

Case information: A 74 year-old man presented with a history of bilateral eyelid swelling for one year, cough for 8 months, and multiple episodes of syncope for 2 weeks. His past history included hypertension, paroxysmal atrial fibrillation and hyperuricemia. His medications were warfarin, allopurinol, and calcium channel blocker. On exam, orthostatic hypotension was identified with 190/100 mmHg on supine and 147/100 mmHg on sitting. The bilateral eyelids were swollen. He had mild cognitive dysfunction, parkinsonism and visual hallucination of turtles.

In laboratory exam, he had mild macrocytic anemia, hypergammaglobulinemia (IgG, 3602 mg/dL) and mild renal failure with proteinuria (urine protein, 0.5g/g Cr). Serum and urine immunoelectrophoresis revealed only polyclonal proteins. Rapid ACTH stimulation test was normal. After discontinuing calcium channel blocker, his episodes of syncope ceased, but eyelid swelling only partially improved. His dementia, visual hallucination and parkinsonism were attributed to probable diffuse Lewy body disease. Because of his delirium, he was discharged on day 9.

However, worsening of his eyelid swelling was noted at outpatient and CT scan revealed swellings of bilateral lacrimal glands and mediastinal lymphnode and interstitial pneumonitis. Ophthalmologist was consulted and possibility of IgG4-related disease was suggested. Serum IgG4 was elevated and biopsy of salivary gland confirmed the diagnosis. Steroid treatment improved eyelid swelling and renal function.

Discussion: IgG4-related disease is a multi-organ disease, causing dacryoadenitis, sialoadenitis, interstitial pneumonitis, pancreatitis, or interstitial nephritis. Steroid treatment is usually effective and thus its diagnosis is important. It is difficult to make a diagnosis of multi-organ disease such as IgG4-related disease, especially in patients with other comorbidity, as in our patient with diffuse Lewy body disease with multiple complaints. Double-edged Occam’s razor (two hypotheses rather than single) may be needed for explaining two complex diseases.

Sometimes when you hear hoof beats, it could be a unicorn

A Peterson1, J Akhtar1, Z Schultz1.

1 University of Pittsburgh Medical Center, Pittsburgh, PA

Learning objectives: Common diseases are what physicians should expect to encounter. Unfortunately, in a world of horses, the hoofbeats of zebras often go unrecognized or misdiagnosed.

Case information: A 46-year-old man had bilateral endoscopic sphenoethmoidectomies in 2011 for bilateral nasal polyposis and allergic fungal frontal and sphenoid sinusitis. He on followup examination had mild recurrence of polyps in the sphenoid cavity. The patient developed pain 6 days ago prior to presentation behind his left eye and was started on amoxicillin. He continued to have pain associated with mild swelling and felt his vision being “off” with diplopia on left gaze. A CT orbits was similar to the patient’s prior sinus CT. The eye examination confirmed a partial sixth nerve palsy, and was thought to be from the previous fungal infection. The patient was discharged home on Augmentin. The patient came back to the emergency room 2 days later with progressive loss of vision and left periorbital pain. The eye examination exam showed decreased central vision and a new partial third nerve palsy. The MRI showed abnormal soft tissue within the patient’s left anterior clinoid and anterior portion of the left cavernous sinus extending into the left orbital apex concerning for granulation tissue. A diagnosis of Tolosa Hunt syndrome was made, and was started on IV antibiotics and steroids. The patient underwent endoscopic retro orbital decompression with findings of a left orbital apex and optic strut abscess which grew pseudomonas aeruginosa. The final diagnosis was pseudo Tolosa Hunt syndrome secondary to bacterial infection.

Discussion: The common axiom about hoof beats serves the diagnostic process well… except sometimes it might be something uncommon than a zebra. These situations challenge clinicians to think and be aware of the less-common possibilities.

* Whipple’s disease: a fatal disease that is often missed without gastrointestinal symptoms

A Rezk1, A Gunnerson1, M Komar1.

1 Geisinger Medical Center, Danville, PA

Learning objectives: Whipple’s disease is a potentially fatal condition that is usually missed due to its rarity and nonspecific symptoms. Obscure disease without gastrointestinal manifestations can be a diagnostic challenge. Patients could be misdiagnosed and treated with biological therapy that could aggravate preceding Whipple’s disease.

Case information: A 58 year old female presented with years of migratory polyarthralgias in her knees, ankles, elbows and wrists. She had a right knee effusion and skin hyperpigmentation. Labs were negative for rheumatoid factor, anti-cyclic citrullinated peptide antibody and anti-nuclear antibody. She was diagnosed with seronegative rheumatoid arthritis and was treated with hydroxychloroquine. Ten months later, prednisone and methotrexate were added for refractory symptoms. She started etanercept six months later with little improvement in symptoms. Six months later she developed mild cognitive dysfunction. One year later she started having abdominal pain, diarrhea and weight loss. CT scan demonstrated mesenteric lymphadenopathy. Mesenteric lymph node biopsy revealed histiocytes with periodic acid-Schiff positive globules and positive Tropheryma whipplei immunolabeling. Her symptoms were unified in Whipple’s disease. She was treated with ceftriaxone followed by trimethoprim-sulfamethoxazole. Her symptoms started to improve.

Discussion: Whipple’s disease is a chronic systemic infection due to Tropheryma whipplei that commonly manifests itself through gastrointestinal symptoms of malabsorption. The disease has a prodromal stage noticeable for arthralgia and arthritis followed by a steady state stage remarkable for diarrhea and weight loss. The average time between the two stages is six years but immunosuppressive therapy can expedite clinical progression. We believe this occurred in our patient. Articular symptoms can represent obscure Whipple’s disease leading to misdiagnosis and administration of biological therapy that could aggravate existing Whipple’s disease. Diagnosis is established by periodic acid-Schiff staining of small bowel biopsies. Medical providers should consider Whipple’s disease in patients with abdominal and joint symptoms after omitting common diseases especially if symptoms are refractory to treatment.


*Presented in an oral session.

These labs are totally out of whack! What am I missing?

L Rutherford1, H Singh1,2.

1 Baylor College of Medicine, Houston, TX

2 Michael E. DeBakey Veterans Affairs Medical Center, Houston, TX

Learning objectives:

* Describe a missed opportunity to make correct diagnosis for a patient presenting with significant lab abnormalities

* Describe how safeguards could help overcome diagnostic errors in busy inpatient settings.

Case information: A 67 year-old male with distant h/o thrombosis leading to left AKA and partial small-bowel resection presented with 2 weeks of abdominal pain. He’d had “intermittent diarrhea” for years, and received stable long-term anticoagulation with 5mg warfarin. On admission, exam revealed a large rectus diastasis. His INR was >17 with K-3.0, Cl-94, Ca-5.5, Phos-1.9, Mg-0.8. Electrolytes were attributed to decreased PO intake and were replaced. CT abdomen showed fluid in the herniated sac and bowel wall edema concerning for ischemia. Patient declined exploratory laparotomy and left when the pain resolved. He was sent home on 5mg Warfarin with outpatient antiphospholipid syndrome work-up. One month later, he presented with similar abdominal pain and serosanguinous drainage from ulcers beneath his diastasis. Electrolyte abnormalities had worsened: K-2.1, Cl-99, Ca-4.9, Mg-0.5, Phos-2.3, Vit-B12-143, Vit-D-10.7 and INR>17. CT-scan showed no fluid or wall edema. Based on these findings, the new admitting team performed a “diagnostic time-out” and suspected short bowel syndrome with malabsorption-related Vitamin K deficiency. On consultation, Gastroenterology additionally suspected bacterial overgrowth and started Rifamixin. After brief IV supplementation, he maintained electrolytes with oral replacement. He was discharged home on 1mg Coumadin with outpatient follow-up. Abdominal pain was attributed to diastasis and resolved with mild analgesics. Homocysteine levels were elevated.

Discussion: The initial admission largely focused on possible ischemia-related abdominal pain but missed the opportunity to work-up major lab abnormalities. Strategies to better “calibrate” physicians so they know when to seek help or pause to reassess complex situations, are largely unknown. Our case illustrates how safeguards such as appropriate use of consultants and diagnostic “time-outs” are essential to overcome diagnostic errors related to missed red flags and premature closure in busy inpatient settings.

Testing for glycemic control in stroke patients

E Ryan1, J Ritchie1.

1 Emory University, Atlanta, GA

Learning objectives: To describe the instances when HbA1c testing is appropriate and inappropriate in stroke patients.

Case information: A 42 year old female with history of stroke and sickle cell disease presented to the emergency department complaining of worsening headache and left-sided numbness for the last 4 ½ hours. She also complained of pain in the chest when lying on her side, left-sided weakness but denied any trouble speaking or forming words, fever, chills, productive cough, or swelling in the legs. The ED physician activated the stroke order set which included both Stat and Routine testing: Complete Blood Count, Basic Chemistry Panel with Magnesium, Troponin, PT INR, Lipid panel, and HbA1c. The patient was admitted to the ICU for a transfusion and monitoring. Shortly, the HbA1c results were reported as: 7.9%--S, C, or F variant hemoglobin noted. Physicians treating the patient disagreed in how to interpret these results.

Discussion: There is strong medical evidence that diabetes and dyslipidemia are associated with an increase in short-term mortality and overall poor outcomes in stroke patients. Because of this the American Heart Association and the American Stroke Association recommend testing patients for LDL-cholesterol and HbA1c within 24 hours of admission with ischemic stroke symptoms. These guidelines are for the general population and in order to improve internal compliance with the guidelines lab order sets have been created. Unfortunately, in hospital setting there are many reasons HbA1c cannot be used as a valid measurement of glycemic control. In individuals who do not have endogenous HbA, are not hemodynamically stable, and/or have received transfusions in the last 3 months HbA1c is no longer an appropriate surrogate for long-term glycemic control. As HbA1c is used throughout medicine more education is needed to help the medical community to understand the analyte, the method of measurement, and aid in interpreting the result.

Unremitting vomiting as a principal symptom of neuromyelitis optica

S Sato1, S Nishiguchi1, J Branch1, T Watari1, S Tomiyama1, I Kitagawa1, Y Tokuda2.

1 Shonan Kamakura General Hospital, Kamakura, Japan

2 Japan Community Healthcare Organization, Minatoku, Japan

Learning objectives: Consider possibility of neuromyelitis optica presenting with unremitting vomiting.

Case information: A 69-year-old female with a prior history of Miller-Fisher syndrome presented with unremitting vomiting. The patient had been diagnosed on three occasions. On the fourth occasion, she was admitted after which she developed hiccups. Gastrointestinal investigations were undertaken including a gastroscopy, which was within normal limits. As a result, she was discharged and her symptoms subsided on the 11 day. However, the patient had noticed some right sided visual loss during the admission. She sought an ophthalmological opinion on the next day after her discharge, at which point she was diagnosed with complete visual loss of her right eye. She was referred back because of new onset lower extremity weakness on the subsequent morning. A neurological examination showed an afferent pupillary defect of her right eye, bilateral lower extremity manual muscle testing of 3/5, a sensory loss with a level at Th4, and weak anal tone. Twenty three days after her initially symptom, she was definitively diagnosed with neuromyelitis optica (NMO) due to results of spinal and optic nerve magnetic resonance imaging abnormalities. There was significant improvement in her lower extremity motor symptoms except for her visual loss after received plasma exchange therapy.

Discussion: Vomiting is a common symptom usually suggesting gastrointestinal disease rather than a rare and severe neurological disorder such an NMO. Moreover, because vomiting is regarded as a low-yield symptom for purposes of diagnoses, it can often be overlooked as relatively benign. However, it is important to consider central nervous system disease as a cause of unremitting vomiting so that early diagnosis and treatment can be undertaken with the aim to reduce the disability associated with NMO.

“She gets like this:” a missed diagnosis of stroke due to underlying dementia

S Shah1.

1 Hospital of the University of Pennsylvania, Philadelphia, PA

Learning objectives: 1. Recognize possible bias of preexisting dementia during acute presentations. 2. Review the importance of thorough physical examination to facilitate timely diagnosis.

Case information: A 78 year-old woman with dementia was brought to the ED for acute mental status change and facial droop. On arrival, she was newly aphasic and lethargic. Head CT was negative for acute changes; no neurologic exam was documented. The patient’s daughter, via telephone, remarked to the ED clinician, “She gets like this.” Five hours after triage, during handoff, the admitting team posited that the presentation was consistent with acute stroke; the patient received an aspirin suppository. She became progressively obtunded, unable to protect her airway. The daughter had arrived and confirmed the acuity of the facial droop and level of unresponsiveness. The family pursued comfort measures; the patient expired the next day.

Discussion: Stroke is common amongst the elderly, and patients with vascular dementia are at risk for recurrent stroke. Of patients ages 70 to 79 who had a stroke, almost 10% had preexisting dementia. In these patients it is often difficult to distinguish benign episodes of decreased interactivity from life-threatening acute alterations in alertness. Patients with dementia are less likely to be admitted to a stroke unit and more likely to die in the hospital. Up to one-third of neurologic diagnoses in the emergency room carry uncertainty, and anywhere from 5 to 38% of acute ischemic stroke diagnoses are missed in the ER. Presentations with typical symptoms, such as a facial droop, are misdiagnosed 4% of the time. The unpacking bias – the failure to elicit all relevant information eg. the neurologic examination – may have contributed to our patient’s missed diagnosis. The ED clinician anchored on the statement that the patient sometimes “gets like this,” missing the opportunity to investigate clinical changes further.

Peau d’Orange: buzzwords which incorrectly framed a differential

S Shah1.

1 Hospital of the University of Pennsylvania, Philadelphia, PA

Learning objectives:

  1. 1. Recognize the impact terminology can have on framing differentials.

  2. 2. Review the utility of a thorough history and physical in minimizing diagnostic testing.

Case information: A 49-year-old Eritrean woman with RA presented with right arm swelling and was found to have DVTs in multiple vessels. Initial exam also documented peau d’orange changes of the right breast and a skin nodule, treated as cellulitis. The breast’s appearance triggered an extensive search for underlying malignancy. Breast ultrasound, mammogram, abdominopelvic imaging, and excisional lymph node biopsy were all negative. Reexamining the data highlighted a marked peripheral eosinophilia, as high as 50%. Repeat interview and exam uncovered daily episodes of hives, prior keratitis, and other skin nodules. Ten days after hospitalization, a skin snip biopsy was performed, revealing onchocerciasis. She was treated with ivermectin, with resolution of her hives.

Discussion: While seemingly unprovoked DVTs raise suspicion for underlying malignancy, one cannot determine provocation without a thorough history, physical, and data review. This patient had multiple reasons to develop thrombosis, including RA, hypereosinophilia, and underlying infection. Onchocerciasis can present even decades after exposure with hives, pruritus, subcutaneous nodules, limb swelling, and keratitis. Anchoring on the combination of DVT plus breast edema, with blinders against the other available information, the team traveled down a path of inappropriate testing, including a surgical procedure. The textbook term of “peau d’orange” framed the differential, triggered an illness script that skewed the workup towards a particular disease, and resulted in excessive testing, thus delaying diagnosis and appropriate treatment. In addition to the obvious financial costs of the extensive workup, there was a psychological cost to the patient in this search for a nonexistent malignancy. Descriptive terms have powerful connotations that must be considered in the context of the entire clinical picture.

A disclosed cause with near-miss premature closure

T Suzuki1, H Kobayashi1, Y Tokuda2.

1 Tsukuba University Hospital Mito Medical Center., Ibaraki, Japan

2 Japan Community Healthcare Organization, Minatoku, Japan

Learning objectives: Premature closure can be avoided by careful workups for appropriately investigating a cause explaining important clinical problems.

Case information: A 36-year-old woman, hospitalized in obstetrics in a nearby hospital for postoperative observation of cesarean section, was referred to department of orthopedics of our hospital because of pain and progressive edema of the right lower extremity. Abdominal echography was conducted at the obstetrics, showing right side hydronephrosis but its cause was not pursued. An orthopedist commented that the patient did not have “orthopedic” illness and had simply “edema” and a closed case was suggested. However, since the right leg pain became worse and a cause of her illness was not disclosed, the patient was referred to department of medicine. On physical examination, vital signs were normal, but edema and Homan’s sign was observed in the right lower leg. A diagnosis of deep venous thrombosis (DVT) was considered and thus ultrasonography was conducted, revealing DVT from the right the right common iliac vein to the right femoral vein and bilateral hydronephrosis. Further, contrast-enhanced CT scan of the abdomen and the chest was conducted, showing a large mass lesion around the uterus with ureteral compression and pulmonary embolism. Eventually she received a diagnosis of uterine cervix cancer and venous thrombosis and started with its appropriate treatment.

Discussion: For patients with unilateral edema, local lesion such as DVT should be considered first since pulmonary embolism can be associated with DVT and it may lead to serious complication. When considering DVT, we should also be careful about its etiology, including internal malignancy such as uterine cervix cancer. Near-miss premature closure was involved in this case for unexplained edema for orthopedist and for hydronephrosis for obstetrics. Careful work-up disclosed the single cause explaining these findings before closing prematurely.

Celiac artery compression syndrome: abdominal pain of unknown origin

T Terashima1, T Watari1, Y Tokuda2.

1 Shonan Atsugi Hospital, Kanagawa prefecture, Japan

2 Japan Community Healthcare Organization, Minatoku, Japan

Learning objectives: Celiac artery compression syndrome (CACS) is a rare disorder but should be considered as a cause for unknown recurrent abdominal pains after eating.

Understanding availability bias may help avoid cognitive errors in a diagnostic reasoning.

Case information: A 41-year-old man presented with a sudden onset severe upper abdominal pain for 3 hours that started 30 minutes after eating meats at riverside barbecue. He reported he had the similar abdominal pains usually after meals for a few years but no diagnosis has been made. Physical examination showed the flat and non-tender abdomen with normal bowel sounds.

Initial investigations on admission, including blood tests, abdominal ultrasound, contrast CT scan, and esophagogastroduodenoscopy, were considered normal, and thus he was diagnosed as acute gastroenteritis and hospitalized for close observation. On day 2, reconstructed sagittal abdominal CT angiogram demonstrated a focal narrowing in the proximal celiac artery. He received a revised diagnosis of celiac artery compression syndrome and he made an uneventful recovery with only conservative treatment.

Discussion: Celiac artery compression syndrome (CACS), Danbury syndrome, is a rare disorder characterized by postprandial intestinal angina caused by insufficient celiac blood supply to gastrointestinal organs. It occurs as a result of focal stenosis of the celiac artery due to compression by the median arcuate ligament. The gold standard diagnostic test need to focus on celiac etiology, including 3D-CT angiography, selective catheter angiography, and magnetic resonance angiography.

A rare condition, like CACS, can be a diagnostic challenge for physicians. Most patients with this syndrome may undergo extensive workups for other causes of abdominal pain. The lateral view of the celiac artery led to the diagnosis of CACS in this case. Availability bias was involved in our initial misdiagnosis because of our quick consideration for acute gastroenteritis by history of abdominal pain during eating barbecue meats.

A revised diagnostic list for pulseless electric activity

Y Tokuda1, T Shimizu2, T Watari3, H Inada3, T Nemoto4, K Matsumoto5, I Kitagawa3.

1 Japan Community Healthcare Organization, Minatoku, Japan

2 University of Hawaii, Honolulu, HI

3 Shonan Kamakura General Hospital, Kamakura, Japan

4 St. Mariannna, Kawasaki, Japan

5 Osaka National Hospital, Osaka, Japan

Learning objectives: Overcome anchoring bias and look for hypothesis which would fit to clinical presentations.

Describe a revised list for etiology of pulseless electrical activity.

Case information: A 68-year-old hypertensive man was brought to ER because his wife found him to be unconsciously lying down over the back in the bathroom floor just after he was moaning during taking a shower. He smoked 1 pack and drank 2 cans of beer a day. Visiting EMT personnel could not palpate the peripheral pulses and resuscitation was initiated. The pupils were equally dilated and light reflexes were absent. On arrival, he had pulseless electrical activity (PEA) with apnea and chest compression with tracheal intubation was performed. Jugular venous pressure was elevated and S3 gallop was audible.

Laboratory data were not contributory except for positive blood troponin level. A 12-lead ECG showed atrial fibrillation, right bundle branch block and ST depression in lead V1-6. Chest X-ray showed lung congestion compatible with heart failure. Echocardiography showed left ventricular (LV) diffuse severe hypokinesia (ejection fraction, about 20%). Acute coronary syndrome was suspected and cardiology team was called. However, general internist examined him and ordered brain CT, showing severe subarachnoid hemorrhage (SAH). Neurosurgeon was called but only comfort care was recommended for SAH and he died after several hours after the admission.

Discussion: SAH can be associated with ECG changes, elevated cardiac biomarkers and impaired LV function known as neurogenic stunned myocardium, which shares similar mechanism of sympathetic overactivity to takotsubo cardiomyopathy. A physician needs to overcome anchoring diagnosis and look for more probable hypothesis, like SAH, fit to overall clinical findings.

Current PEA etiology includes 5H (Hypovolemia, Hypoxia, Hydrogen ion, Hyper/Hypokalemia, Hypothermia) and 5T (Tablets, cardiac Tamponade, Tension pneumothorax, coronary or pulmonary Thrombosis). SAH may be included into revised 6H (subarachnoid Hemorrhage).

Pulmonary tuberculosis masquerading as urinary tract infection in Japanese older adult with malnutrition

S Tomiyama1, S Nishiguchi1, S Sato1, T Watari2, I Kitagawa1, Y Tokuda3.

1 Shonan Kamakura General Hospital, Kamakura, Japan

2 Shonan Atsugi Hospital, Kanagawa, Japan

3 Japan Community Healthcare Organization, Minatoku, Japan

Learning objectives: It is important to consider endogenous reactivation of pulmonary tuberculosis as a cause of hypoxia with sputum in older adult with malnutrition.

Case information: An 83-year-old Japanese bed-bound thin male with a prior surgery history of gastrostomy presented with fever. Vital signs showed low blood pressure and oxygen desaturation with a nasal cannula at two liters flow. On examination, there was thickening of skin over extremities, face and trunk. Blood tests revealed a white blood cell count of 6000 /ul and C-reactive protein 12.8mg/dl. Urinalysis showed bacterial pyuria, so that he received intravenous antibiotic treatment for urinary tract infection. Escherichia coli was detected in both blood and urine culture on the third admission day. He had a persistent slightly fever and produced much sputum from the admission, so we consider aspiration as those causes. On 20th admission day, he had high fever. Plain chest CT demonstrated small calcified lesion in the right lung middle lobe and multiple bronchiolitis finding in bilateral lung lobes. Sputum smear examination was positive for acid-fast bacilli. Mycobacterium tuberculosis was detected in his sputum at a later date. A final diagnosis of urinary tract infection and sputum smear-positive pulmonary tuberculosis was made.

Discussion: One patient usually have one disease, however patient in older adult sometimes may have two or more diseases. There is high prevalence of pulmonary tuberculosis in Japan than any other developed countries in the world. Delay in the diagnosis of pulmonary tuberculosis is associated with higher risk of mortality and spreads the infection in the hospital. It is important to consider endogenous reactivation of pulmonary tuberculosis as a cause of hypoxia with sputum for the aged patient with malnutrition despite evidence for other one diagnosis such as urinary tract infection.

Another pharyngitis mimic: intra-oral hematoma of the soft and hart plates in a patient on warfarin and drug-drug interaction

Y Watanabe1, T Abe1, M Yoshii1, Y Tokuda2.

1 Mito Kyodo General Hospital, Mito, Japan

2 Japan Community Healthcare Organization, Minatoku, Japan

Learning objectives:

* Potential killer lists of sore throats should include intra-oral sub-mucosal hematoma especially in patients on warfarin anticoagulation.

* Hemorrhagic complication due to over-anticoagulation from drug-drug interaction should be considered for patients with warfarin who receive a new medication.

Case information: A 64-year-old man presented to our hospital with a 3-day history of painful swelling of the right ankle. A physician in the outpatient clinic saw him and prescribed amoxicillin/clavulanate for suspected cellulitis. Two days later, although the ankle swelling was improved, he developed sore throat, sputum and voice change and visited our hospital again. A diagnosis of pharyngitis was made by a second physician who placed the patient on a wait-and-see approach. However, on the following day, he visited our hospital for the third time. His past history included atrial fibrillation and cerebral infarction and regular medications were warfarin and aspirin.

On exam, the temperature was 37.1 degree Celsius. There was significant sub-mucosal hematoma above the soft and hart plates. Purpura was identified on the right dorsal foot. In laboratory data, hemoglobin was 6.2 g/dl and PT-INR was 19.95. He was then admitted and the laryngoscopy showed sub-mucosal hematoma from the posterior wall of the pharynx to the piriform recess. He received red cell transfusion and vitamin K and he was eventually discharged home without a sequel.

Discussion: Drug-drug interaction is a common cause for increased warfarin over-anticoagulation effects and increased INR was previously observed in patients treated with warfarin and amoxicillin/clavulanate combination. Over-anticoagulation can lead to serious hemorrhagic complications such as intra-oral sub-mucosal hematoma requiring hospitalization and blood transfusion. There are many grave diseases mimicking pharyngitis but intra-oral sub-mucosal hematoma should be considered in patients on warfarin and with sore throat.

Step beyond Sutton’s Law

S Watanuki1, Y Tokuda2.

1 Tokyo Metropolitan Tama Medical Center, Tokyo, Japan

2 Japan Community Healthcare Organization, Minatoku, Japan

Learning objectives: Diagnosis of Whipple’s disease requires a higher index of suspicion beyond Sutton’s law to request PAS stain for undiagnosed intra-abdominal lymphadenopathy.

Case information: A 70-year-old Japanese man presented with a 3-year history of weight loss followed by anorexia and loose stools for the last several months. He denied fever, chills, night sweats, arthralgia or rash. On physical examination, he had generalized edema, anemia and abdominal distension, but no superficial lymphadenopathy. Laboratory data showed anemia and hypoalbuminemia but normal erythrocyte sedimentation rate. Abdominal CT and MRI scans showed multiple retroperitoneal lymphadenopathy with a maximum size of 2 cm. Microscopic exams of the biopsy specimens obtained twice by laparoscopy revealed non-caseating granuloma and surrounding adipose tissue with infiltration of foamed macrophages and foreign body giant cells. Because of the unexplained intra-abdominal lymphadenopathy, he was transferred to our department 5 months after the admission. Additional staining was requested, showing positive PAS and Grocott stains. Electron microscopy revealed numerous bacilli and immunohistochemistry and PCR confirmed these were Tropheryma whippeli. Enteroscopy for small intestine also showed disseminated yellowish-white villi. He received a diagnosis of Whipple’s disease and recovered gradually from antibiotic treatment and was followed up at outpatient basis.

Discussion: It is important to make a correct diagnosis if a disease causes a serious morbidity and is potentially curable illness, like Whipple’s disease. Diagnosis of Whipple’s disease requires a high index of suspicion to request PAS stain for undiagnosed intra-abdominal lymphadenopathy. Sutton’s law states we need to directly approach a tissue by biopsy, if etiology for specific organ disease is unclear, based on the true story of a British bank robber suggesting a bank as the best place to rob. However, this case illustrates a need for additional step for requesting special stains for unexplained histopathologic findings by conventional stains.

Not seeing the forest for the trees

T Watari1, T Shimizu2, K Matsumoto3, S Sato1, H Inada1, S Tomiyama1, I Kitagawa1, Y Tokuda4.

1 Shonan Kamakura General Hospital, Kamakura, Japan

2 University of Hawaii, Honolulu, HI

3 Osaka National Hospital, Osaka, Japan

4 Japan Community Healthcare Organization, Minatoku, Japan

Learning objectives: Cogan’s syndrome is a multi-system autoimmune disease causing vestibuloauditory and ocular manifestations.

Diagnosis of multisystem disease requires seeing whole clinical pictures as well as local processes.

Case information: A 69-year-old Japanese woman presented with a month history of spiking fever, proximal myalgia with weakness, weight loss, intermittent abdominal pain, and low back pain. She also had a history of tinnitus and vertigo for one year. Oral levofloxacin for several days was prescribed for possible abdominal infection at the emergency department but it did not improve her symptoms and she also developed progressive bilateral blurred visions and hearing disturbances.

She visited multiple specialties in our hospital. Ophthalmologist gave her diagnosis of uveitis with iridocyclitis and episcleritis. Otolaryngologist provided a diagnosis of bilateral sensorineural hearing loss with canal paresis. Gynecologist and surgeon suggested no diagnosis. Since she received no unified plausable hypothesis for explaining her symptoms, she was admitted to Department of Medicine. Laboratory data showed negative immunological and serologic findings including syphilis. She underwent temporal artery biopsy, showing diffuse lymphocytic infiltrates. Because her whole clinical pictures were compatible with Cogan’s syndrome, she started with glucocorticoid therapy and her fever and vestibular symptoms gradually improved, but her visual and hearing loss showed little improvement.

Discussion: Cogan’s syndrome is a rare multi-system inflammatory disease involved with ocular and audio-vestibular organs. This syndrome should be considered for unexplained non-infectious inflammatory processes for ophthalmic and vestibuloauditory organs. Patients with Cogan’s syndrome usually visit ophthalmologists for ocular symptoms and otolaryngologists for vestibular and auditory symptoms. However, diagnostic delay is observed when ophthalmologists are not careful about vestibular and auditory symptoms and vice versa. Thus, we should make a diagnosis based on overall clinical pictures beyond only focusing on the details, like the above proverb.

Is it a delayed or prudent diagnosis of Kawasaki disease?

L Xiong1, S Phongjitsiri2.

1 Baylor College of Medicine, Houston, TX

2 Texas Children’s Hospital, Houston, TX

Learning objectives: To illustrate a case of delayed diagnosis of Kawasaki Disease that may have been prevented using evidence-based diagnostic guidelines

Case information: A 5 year old male presented with 7 days of fever, maculopapular rash, cracked red lips, bilateral conjunctivitis, and new leg pain. He was previously seen by his pediatrician who suspected viral infection. At presentation, he had no adenopathy or peripheral edema, improving fever curve, and albumin 3.7g/dL, ALT 97, no anemia, platelets 349,000/mm3, WBC 8980/mm3, urinalysis with no WBC, ESR 31mm/hr, and CRP 3.5mg/dL that was down-trending from CRP 7.8mg/dL at day 2 of fever. As his labs appeared reassuring and his leg pain resolved with ibuprofen, he was diagnosed with transient synovitis. He later developed cervical lymphadenopathy and periungual desquamation. An outpatient echocardiogram showed suspicion for coronary artery dilatation. Although repeat echocardiogram showed prominent but normal coronary arteries, he met criteria for KD and received IVIG and aspirin.

Discussion: Clinicians used clinical judgment to differentiate Kawasaki disease (KD) and synovitis. Anchoring on leg pain resolved with NSAID and suspected viral etiology led to a diagnosis of transient synovitis. False reassurance from labs resulted in premature closure. Although evolving diagnostic data strengthened the correct diagnosis, sufficient evidence existed at the initial presentation for an echocardiogram per the incomplete KD algorithm, which may have led to the correct diagnosis sooner. As IVIG is most effective in the first 10 days of illness, prompt diagnosis of KD may reduce the risk of coronary artery changes. Diagnosing KD is challenging due to its gradual course and time-sensitive therapy. The AHA recognized these challenges and proposed guidelines for “incomplete” KD as a continuum of the complete presentation of KD. This demonstrated how professional societies developed diagnostic guidelines in a standardized, consensus-based and evidence-based fashion to make diagnosis safer.

About the article

Published Online: 2015-01-24

Published in Print: 2015-02-01

Citation Information: Diagnosis, Volume 2, Issue 1, Pages eA1–eA46, ISSN (Online) 2194-802X, ISSN (Print) 2194-8011, DOI: https://doi.org/10.1515/dx-2015-0001.

Export Citation

©2015 by De Gruyter.Get Permission

Comments (0)

Please log in or register to comment.
Log in