Hospital pneumatic tube systems (PTSs) are becoming frequently used method of transport in hospitals. PTSs are rapid automated delivery systems that can efficiently transport drugs, medical and patient’s reports, X-ray films, tissue samples and blood specimens to and from laboratories, pharmacies, nurse’s stations, blood banks and the emergency department . However, the transport of blood samples from the phlebotomy site to the core laboratory still remains the most common use of PTS. PTS has the advantage of eliminating wait times and reducing manual labor. In particular, it reduces laboratory turnaround time, which is the total time taken between the blood collection and the return of the completed report of final analytical results to the doctor or the clinic . In addition, the hospital staff is freed from the pressure of work as there is no need of courier transport and can be focused on patient-care activities, improving in this way the service quality of the hospitals. However, the transported samples, depending on the system configuration and the speed, are subject to forces of pressure such as such as sudden accelerations/decelerations, high speeds, changes in air pressure generated by the vacuum system, movement of blood in test tubes and vibrations. These forces can potentially lead to a very common error in the preanalytical phase of the testing process known as hemolysis.
Hemolysis designates the release of hemoglobin and other intracellular components from erythrocytes to the surrounding plasma following damage or disruption of the cell membrane . In vitro hemolysis, caused by PTS, has been reported to alter the blood specimen’s quality and potentially affects clinical laboratory measurements such as biochemical, most notably potassium (K), lactate dehydrogenase (LDH) , hematology and coagulation parameters. Moreover, these strong forces can potentially affect blood gas measurements  and spectrophotometric analysis of cerebrospinal fluid . Most recently, the transportation of blood samples through the PTS has been shown to affect in vitro platelet function .
Many studies have been performed on the effect of pneumatic tube transport system on blood specimens in order to determine its association to hemolysis. However, conflicting outcomes did not result in any firm conclusion; thus, the effect of PTS on blood samples is still debated. Therefore, in order to resolve this uncertainty, a comprehensive quantitative synthesis of the published literature is needed. The aim of this systematic review and meta-analysis is to evaluate the effect of PTS transport on routine laboratory samples such as hematological, biochemical and coagulations’ parameters and to give a firmer estimate on the relation of hemolysis and pneumatic tube transportation. The analysis was conducted by comparing the results measured when the sample was transported manually to the core laboratory, to those obtained when the paired samples were transported via the PTS.
Materials and methods
The methods used were in accordance with the preferred reporting items for systematic reviews and meta-analyses (PRISMA) guidelines . A systematic electronic search of MEDLINE and SCOPUS bibliographical databases was performed, without language restrictions, up to November 2016 using the following search terms: “Pneumatic tube system” OR “Pneumatic tube transport” OR “Pneumatic tube transportation” OR “Pneumatic tube transport systems” OR “Pneumatic tube delivery system”. We also performed manual check of the reference list of included papers and related reviews to complement our electronic search as well as e-mail correspondences with authors for additional data, where relevant.
Study selection and data extraction
The outcome of interest was changes to laboratory measurements subsequent to transportation of blood samples via the PTS. Studies that prospectively evaluated the effect of PTS transport on routine laboratory measurements, including hematology, biochemical and coagulations parameters, by comparing the results of hand delivered samples and samples transported via the PTS, were included for meta-analysis. The inclusion and exclusion criteria applied during selection of eligible studies and data extraction were as follow: (i) the blood sample collection had to be performed from a single venipucture. Studies were excluded if stated otherwise, (ii) in order to minimize the sources of variation, the present meta-analysis focused on paired samples that were processed and analyzed at the same time. Studies that evaluated the PTS in different periods or extracted data from the laboratory information system were excluded, (iii) studies that were performed with blood samples collected in primary tubes and not separated in aliquots were only included and (iv) studies that did not provide mean (or median) serum or plasma concentrations of our interested outcomes, and no other estimate that could be used, were excluded from meta-analysis.
Two reviewers worked independently, searching the literature, reviewing full manuscripts of eligible studies and extracting data into an electronic database. These data included first author’s name, publication year, country where the study was conducted, number of subjects, characteristic of the participants (healthy or patients) and the distance and the speed of the PTS. For the hematological parameters, the hematology analyzer was also recorded (Table 1). Any disagreement was resolved by discussion and consensus with a third reviewer.
The mean difference (MD) weighted by its standard error (SE) was used as the effect size for the continuous data meta-analysis, along with the 95% CIs, to evaluate the effect of PTS on laboratory measurements. The mean and the standard error of the mean were calculated for both the manual and PTS transport. MD was calculated as follows:
where corresponds to the mean of samples transported via PTS and to the mean of paired samples transported manually. Initially, all measurements were converted in the same units. In case the outcome measures were reported as median (M) and range (R=max–min), mean and SD values were estimated using the methods and the guidelines described by Hozo and coworkers . For small sample sizes (n<25), we used the formula
For n>25, the median itself was used, as it is considered to be better estimator of the mean. Concerning the SD, for small sample sizes (n<15), we used the formula
For data sets of medium size (15<n<70), we used the formula
whereas for larger data sets (n>70), we used the same formula but with 6 in the denominator (R/6), as previously advised .
If the outcome measures were reported as median (M) and interquartile range (IQR), mean and SD values were estimated according to the recommendations of the Cochrane Handbook . In that case, the median was used as an estimator of the mean, whereas the SD was calculated using
In case the standard error of the difference was not reported, it was calculated using SE=SD/√n, where SD is the standard deviation of the difference and n is the number of paired samples. The standard deviation of the difference was estimated according to the recommendations of the Cochrane Handbook using the formula for the calculation of the standard deviation when comparing dependent (paired) samples :
Based on the data given in each study, standard deviation of the difference could also be calculated using
In case the t statistic was not given, it was calculated from the p-values reported in the studies, obtained from Student’s t-test (two tailed) or from the Wilcoxon signed-rank test. Exceptionally, if SD was not reported in a measurement, it was imputed using the largest SD reported for the same measure and the same population, i.e. healthy or patients. Both Spearman and Pearson correlation coefficients were used for this task. If correlation coefficient was not reported as an outcome measure, it was estimated using Eq. (6). Finally, if the standard deviation of the difference could not be calculated from any of the above statistical tests mentioned, a meta-analysis of correlation coefficients  was conducted and the overall correlation coefficient was used to estimate the SD of the difference, using Eq (6).
For the quantification of data synthesis, a meta-analysis was performed using the random-effects model proposed by DerSimonian and Laird . Heterogeneity among studies was assessed using Cochrane’s Q and I2 tests . Subgroup analyses were performed as a mean of investigating heterogeneous results, based on (i) the type of participants (healthy vs. patients), (ii) the different hematology analyzers, where the samples were tested, (iii) the speed of the PTS for each study and (iv) the distance traveled by the PTS samples. Meta-regression analysis was further used to determine the influence of potential effect modifiers (speed and distance of PTS) . Sensitivity analysis was conducted by removing one study each time and repeating the analysis, in order to investigate the influence of each study on the overall effect size. Cumulative meta-analysis was also performed in order toinvestigate whether the summary effect size changed considerably over time as more data accumulated, by visual inspection of cumulative plot. Possible publication bias was examined using the visual inspection of funnel plot asymmetry, as well as the rank correlation method of Begg , the Egger’s  regression method and it’s random-effects analogue . The statistical analyses were performed with the statistical software package STATA 13.1. Results were considered statistically significant for p<0.05.
A total of 287 publications were identified through the search criteria of which 185 remained after removal of duplicates. After a screen on title and abstract, a total of 29 articles remained for further evaluation, five of which were excluded as they did not meet the inclusion criteria. More specific, four studies were excluded from the meta-analysis because of the fact that they studied independent samples, that is, different group of people in different period of time, and one study was excluded as blood samples were not obtained from single venipucture but rather from both hands. As a result, 24 studies met the inclusion criteria and were included in the meta-analysis (see Figure 1).
Characteristics of the included studies
The descriptive and quality data of the included studies are summarized in Table 1. From the 24 eligible studies, 14 investigated the 16 hematological parameters, with a total number of 691 participants. Eighteen studies analyzed measured biochemical parameters such as K with a total of 788 paired samples, LDH with a total of 919 paired samples and aspartate aminotransferase (AST) with a total number of 439 participants. Among the eight studies that examined coagulation parameters, five articles included prothrombin time measurements and eight partial thromboplastin time measurements. Finally, three studies examined the erythrocyte sedimentation rate. Of note, two studies with hematological measurements reported data for two independent populations (i.e. groups of healthy individuals and patients), whereas four studies with biochemical measurements reported data for two independent populations measured under different distance or speed of the PTS. Because the individuals in these populations were different, the estimates are statistically independent, and thus they were treated as different studies in the meta-analysis.
No statistically significant differences between the PTS samples and the hand-carried samples were observed in complete blood cell counts (CBC) and reticulocytes (Retic) (p-values from 0.122 to 0.951). Similarly, no statistically significant difference was found between the manual and the pneumatic tube transport system for the coagulation parameters and the ESR. However, a statistically significant difference between the two delivery methods was found for the biochemical parameters K, LDH and AST. Analytically, the overall meta-analysis results yielded MD values of 0.04 mmol/L for K (95% CI=0.015–0.065; p=0.002), 10.343 U/L for LDH (95% CI=6.132–14.554; p<10−4) and 1.023 IU/L for AST (95% CI=0.344–1.702; p=0.003). For the majority of the parameters used in the meta-analysis, the between-study heterogeneity was high with I2>75%. Table 2 provides information about the pooled effect sizes with 95% confidence intervals of the 21 variables studied along with the number of studies and subjects involved for each parameter, the p-value of each meta-analysis, the I2 and p-value of heterogeneity tests and the presence or not of publication bias and time trend. It should be noted, that other biochemical parameters were not included in the meta-analysis, as the studies did not provide adequate data in order to calculate the effect sizes. Figures 2 and 3 represent the forest plot of K and LDH data set, respectively.
Subgroup analysis was performed for all parameters according to the distance and the speed of the PTS. Specifically, for the set of hematological data, the analyzer used for measuring the tests and the population tested were also examined in order to compare the mean effect across subgroups. Meta-analysis according to population did not yield any significant finding. For the subgroup analysis according to the hematology analyzer, there was a high degree of heterogeneity (p<10−4) between the four groups of hematology analyzers (Cell-Dynn, Beckman Coulter, Sysmex and Advia) for almost all hematology parameters, except for mean corpuscular volume (MCV), HGB, MPV and RDW. Statistically significant heterogeneity was found in the data set of the white blood cells (WBC) with respect to the methods of transport because samples measured with the technology of Cell-Dyn (MD=0.04 K/μL; 95% CI=0.01–0.06) and Beckman Coulter (MD=0.11 K/μL; 95% CI=0.01–0.21) hematology analyzer yield significant difference and the former giving consistently larger difference. Similar result was found for the Platelets (PLT) data set when samples were measured using the technology of Sysmex hematology analyzer (MD=−2.05 K/μL; 95% CI=−3.28 to −0.82). Moreover, the subgroup analysis based on the length of PTS showed that: (a) in the longest distance traveled by the samples via PTS (400–600 m) there was alterations in the results of WBC (MD=0.15 mmol/L; 95% CI=0.09–0.22; p<10−4) and (b) there were alterations in the data set of LDH, particularly for the two longest distances of 200–400 m (MD=15.76 U/L; 95% CI=10.0–21.42; p=<10−4) and 400–600 m (MD=5.97 U/L; 95% CI=0.75–11.19; p=0.025). Finally, the subgroup analysis according to the speed of PTS yielded a statistically significant difference in the results of LDH and K, for all groups, regardless of speed range.
A random-effects meta-regression analysis for the WBC data set revealed a positive linear relationship between the MD of WBC and the distance of PTS (p=0.007; β=0.0002912). A negative linear relationship resulted from the meta-regression analysis of red blood cells (RBC) data set and the distance of PTS (p=<10−4; β=−0.000279) (Figure 4). Finally, the product of the distance over speed was used as a variable in the meta-regression analysis for the LDH data set and yielded a statistically significant result (p=0.015; β=−0.00254942) (Figure 5).
In the leave-one-out sensitivity analyses, the results showed that no individual study influenced the overall effect estimate of all included studies. However, it is worth mentioning that we excluded the study of Koroglu and coworkers, from the meta-analysis of the WBC data set only, because it was conducted in a population with very low number of WBC and reestimated the overall effect size. The result of the meta-analysis yielded a statistical significant p-value of 0.043 with a MD of 0.05 K/μL (95% CI=0.00–0.11).
A small indication of publication bias was found, only in the meta-analysis for the RBC values. This finding was observed visually in the funnel plot graph and validated statistically using Begg’s and Egger’s test (p=0.029, p<10−4, respectively). However, because in the presence of heterogeneity the results of these tests are questionable, we applied the random-effects analogue of the regression method of Egger’s. The random-effects analysis excluded the possibility of publication bias (p=0.924).
The present meta-analysis is the first effort to synthesize the published evidence and evaluate the effect of the PTS on blood specimens. Our attempt was to quantify the level of in vitro hemolysis by evaluating, at the same time, the role of transport speed and distance as well as the impact of the technology of the analyzers used in the measurement of the samples, specifically in the hematology parameters.
Our findings point toward a statistically significant effect of PTS on K, LDH and AST and may lead to the conclusion that the pneumatic tube transport system is associated with a mild degree of hemolysis in the transported blood samples, as these three tests, according to the literature, are first altered when hemolysis is present . It should be noted that when a study was excluded from the meta-analysis of the WBC data set only, a statistically significant result emerged for WBC parameters also. This particular study was conducted exclusively, in a population with acute leukemia (leucopenia) and hence with mean WBC <0.8 K/μL, which means that there is a less risk of cell disruption occurring by the PTS as the WBC cell count is small. At this point, it is important to note that except for RBCs, WBCs have been implicated as a source of excess K in the patient serum. Many studies suggest that a false elevated serum or plasma concentration of K is associated with some hematological disorders, particularly in leukemic patients that have high amounts of WBCs and fragile cell membrane, in which the leukemic cells are more susceptible to disruption (lysis) when exposed to even mild mechanical trauma [42, 43].
In all analyses reported here, there was a strong evidence for between-study heterogeneity, and there are many factors that contribute to this. Some of them include the PTS configuration and type, the different carriers, the technology of the analyzer used to test the blood samples, the speed and distance traveled by the samples via PTS, the sample collection tubes, as well as the pressure and the intensity of the gravitational force caused by the sudden accelerations and decelerations in which samples are subjected during transport. Some of these factors were investigated in our meta-analysis in order to account for the sources of heterogeneity. Subgroup analysis, according to the hematology analyzer, showed large heterogeneity between the four different hematology analyzers. This finding may be due to the fact that different methodologies are used by the analyzer in order to determine the complete blood count, or it may be related to other study-level characteristic (such as the speed of the PTS used in the particular hospital). Subgroup analysis according to the speed of PTS showed statistically significant higher values of K and LDH when samples were delivered by PTS compared to hand-delivered samples, regardless of the rate. Subgroup analysis based on the length of PTS yielded similar results for the data set of WBCs and LDH. These findings give a first indication of the positive correlation between speed and distance of PTS samples and hemolysis.
Meta-regression analysis confirmed quantitatively the above findings. Specifically, for the WBC data set, meta-regression analysis revealed that the greater the distance, the higher the MD of WBCs. In most cases, heterogeneity was reduced by 20% (for WBCs) to 40% (for RBCs) when these covariates were used in a meta-regression. Moreover, it is worth mentioning, although the result was not statistical significant, that the overall effect size for the RBC data set was negative, meaning that the values obtained by the hand delivered samples were higher than those obtained by the PTS samples. Accordingly, a reverse relationship was found for the RBC MD and the distance of PTS, that is, the greater the distance traveled in a PTS the more the decrease of the MD of RBCs. This can be explained by the fact that the longer the PTS, the more the switches and bends that samples have to suffer and the higher the mechanical trauma, i.e. the extent to which samples suffer cell (erythrocyte) damage during transport. Finally, meta-regression analysis of the product of distance over speed of PTS, in the LDH data set, showed that the higher the speed and the longer the distance, the higher the MD of the LDH between the two methods of transport. This finding can be explained considering the fact that acceleration is not the only factor responsible for hemolysis because blood samples during regular centrifugation (with a force ≅2.500–3.500 for 10 min) are not subjected to hemolysis. As Streichert and coworkers showed in their work , rapid and large acceleration changes and sudden deceleration or shock forces (three-axis acceleration) may lead to hemolysis during transportation. Moreover, Mullins and coworkers reported  that plasma LDH had a positive linear relationship with the number of shock forces (>3 g) experienced during transport through the PTS as recorded using a Smartphone.
Publication bias was found using the standard tests, only for the analysis of RBC, but it can be attributed mainly to the large heterogeneity because the application of the random-effects regression method for detecting the bias yields no evidence for it. In the cumulative meta-analysis, the studies are accumulated in chronological order, and this allows evaluating the consistency of the results of studies and identifying the point at which no further studies are necessary. The presence of time trend identified in some variables examined in this meta-analysis suggests that an early, overestimated or extreme oppose finding occurred in the first studies investigated. This problem is common in clinical studies and may be explained by the age of the methodologies and the technology of the PTS that evolves over time. A factor that may explain these discrepancies is the fact that older studies reported higher speed of the PTS and greater distance traveled (correlation coefficients for speed and distance vs. time equal to −0.51 and −0.41, respectively).
The overall results of this meta-analysis suggest a statistically significant difference for the samples transported by PTS compared to those transported by hand. This difference, however, is small and may not have a significant clinical effect on laboratory results. Nevertheless, considering the fact that hemolysis is a major preanalytical factor that interferes with laboratory measurements, it is suggested that hospitals should validate their PTS and investigate their blood specimen’s susceptibility to hemolysis. In order to reduce this preanalytical error, some interventions can be applied such as shortening transport times, if possible, or reducing the speed of transportation. As the method of parallel processing for sample analysis is inconvenient and time consuming, the use of data loggers or even smartphones have been applied recently in order to monitor the forces encountered in a PTS and probably as a replacement of blood sampling and laboratory testing.
The authors would like to thank the associate editor and the anonymous reviewers for their valuable comments and suggestions to improve the quality of the paper.
Shibani W, Zulkafli M, Basuno B, editors. Methods of transport technologies: a review on using tube/tunnel systems. IOP Conference Series: Materials Science and Engineering. IOP Publishing, 2016. Google Scholar
Guss DA, Chan TC, Killeen JP. The impact of a pneumatic tube and computerized physician order management on laboratory turnaround time. Ann Emerg Med 2008;51:181–5. PubMedCrossrefWeb of ScienceGoogle Scholar
Lippi G, Blanckaert N, Bonini P, Green S, Kitchen S, Palicka V, et al. Haemolysis: an overview of the leading cause of unsuitable specimens in clinical laboratories. Clin Chem Lab Med 2008;46:764–72. Web of SciencePubMedGoogle Scholar
Lippi G, Plebani M, Di Somma S, Cervellin G. Hemolyzed specimens: a major challenge for emergency departments and clinical laboratories. Crit Rev Clin Lab Sci 2011;48:143–53. CrossrefWeb of ScienceGoogle Scholar
Wenham PR, Hanson T, Ashby JP. Interference in spectrophotometric analysis of cerebrospinal fluid by haemolysis induced by transport through a pneumatic tube system. Ann Clin Biochem 2001;38(Pt 4):371–5. PubMedCrossrefGoogle Scholar
Dyszkiewicz-Korpanty A, Quinton R, Yassine J, Sarode R. The effect of a pneumatic tube transport system on PFA-100 trade mark closure time and whole blood platelet aggregation. J Thromb Haemost 2004;2:354–6. PubMedCrossrefGoogle Scholar
Liberati A, Altman DG, Tetzlaff J, Mulrow C, Gotzsche PC, Ioannidis JP, et al. The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate health care interventions: explanation and elaboration. J Clin Epidemiol 2009;62:e1–34. CrossrefPubMedGoogle Scholar
McClellan EK, Nakamura RM, Haas W, Moyer DL, Kunitake GM. Effect of pneumatic tube transport system on the validity of determinations in blood chemistry. Am J Clin Pathol 1964;42:152–5. CrossrefPubMedGoogle Scholar
Delbruck A, Poschmann H. On the influence of pneumatic post transport on clinical investigation material under different operating conditions. Z Klin Chem Klin Biochem 1968;6:211–6. PubMedGoogle Scholar
Nosanchuk JS, Salvatore JD. Improved pneumatic tube system shortens stat turnaround time. LabMedicine 1977;8:21–5. Google Scholar
Weaver DK, Miller D, Leventhal EA, Tropeano V. Evaluation of a computer-directed pneumatic-tube system for pneumatic transport of blood specimens. Am J Clin Pathol 1978;70:400–5. CrossrefPubMedGoogle Scholar
Keshgegian AA, Bull GE. Evaluation of a soft-handling computerized pneumatic tube specimen delivery system. Effects on analytical results and turnaround time. Am J Clin Pathol 1992;97:535–40. CrossrefPubMedGoogle Scholar
Cui M JRaWH. Changes in lactate dehydrogenase and porassium levels produced by a pneumatic tube system. LabMedicine 2009;40:728–31. Google Scholar
Kratz A, Salem RO, Van Cott EM. Effects of a pneumatic tube system on routine and novel hematology and coagulation parameters in healthy volunteers. Arch Pathol Lab Med 2007;131:293–6. PubMedGoogle Scholar
Wallin O, Soderberg J, Grankvist K, Jonsson PA, Hultdin J. Preanalytical effects of pneumatic tube transport on routine haematology, coagulation parameters, platelet function and global coagulation. Clin Chem Lab Med 2008;46:1443–9. Web of SciencePubMedGoogle Scholar
Koessler J, Kobsar AL, Brunner K, Stolz H, Dossler B, Walter U, et al. The preanalytical influence of two different mechanical transport systems on laboratory analysis. Clin Chem Lab Med 2011;49:1379–82. PubMedWeb of ScienceGoogle Scholar
Tiwari AK, Pandey P, Dixit S, Raina V. Speed of sample transportation by a pneumatic tube system can influence the degree of hemolysis. Clin Chem Lab Med 2011;50:471–4. PubMedWeb of ScienceGoogle Scholar
Evliyaoglu O, Toprak G, Tekin A, Basarali MK, Kilinc C, Colpan L. Effect of pneumatic tube delivery system rate and distance on hemolysis of blood specimens. J Clin Lab Anal 2012;26:66–9. Web of SciencePubMedCrossrefGoogle Scholar
Sari I, Arslan A, Ozlu C, Hacioglu S, Dogu MH, Isler K, et al. The effect of pneumatic tube system on complete blood count parameters and thrombocyte donation in healthy donors. Transfus Apher Sci 2012;47:81–3. CrossrefWeb of SciencePubMedGoogle Scholar
Lima-Oliveira G, Lippi G, Salvagno GL, Dima F, Brocco G, Picheth G, et al. Management of preanalytical phase for routine hematological testing: is the pneumatic tube system a source of laboratory variability or an important facility tool? Int J Lab Hematol 2014;36:e37–40. Web of ScienceCrossrefGoogle Scholar
Al-Riyami AZ, Al-Khabori M, Al-Hadhrami RM, Al-Azwani IS, Davis HM, Al-Farsi KS, et al. The pneumatic tube system does not affect complete blood count results; a validation study at a tertiary care hospital. Int J Lab Hematol 2014;36:514–20. Web of ScienceCrossrefGoogle Scholar
Kocak FE, Yontem M, Yucel O, Cilo M, Genc O, Meral A. The effects of transport by pneumatic tube system on blood cell count, erythrocyte sedimentation and coagulation tests. Biochem Med (Zagreb) 2013;23:206–10. PubMedGoogle Scholar
Toth J, Lenkey A, A VO, Koteles J, Kissne Sziraki V, Kerenyi A, et al. Pneumatic tube system for transport of laboratory samples: preanalytical aspects. Orv Hetil 2014;155:1113–20. PubMedCrossrefGoogle Scholar
Boss DS, Hackeng CM, Vlot EA, van Loon D. Het effect van buizenpost op stolling-en hemolyseparameters: “To walk or not to walk”. Ned Tijdschr Klin Chem Labgeneesk 2015;40:234–8. Google Scholar
Kapoula G, Filntisis I, Karampousli E, Kalivas M, Pagioulas K, Parcharidis P. Pneumatic tube transport system for blood samples: evaluation of its effect on hemolysis. Scientific Chronicles [Epistimonika Chronika] 2015;20:194–8. Google Scholar
Koroglu M, Erkurt MA, Kuku I, Kaya E, Berber I, Nizam I, et al. Assessing safety of pneumatic tube system (PTS) for patients with very low hematologic parameters. Med Sci Monit 2016;22:1329–33. PubMedCrossrefGoogle Scholar
Higgins JP, Deeks JJ. Selecting studies and collecting data. In: Higgins JP, Green S, editors. Cochrane handbook for systematic reviews of interventions. Version 5.0.1 [updated September 2008], ed2011. Google Scholar
Higgins JP, Deeks JJ, Altman DG. Special topics in statistics. In: Higgins JP, Green S, editors. Cochrane handbook for systematic reviews of interventions. Version 5.0.1 [updated September 2008], ed2011. Google Scholar
Prevost AT, Mason D, Griffin S, Kinmonth AL, Sutton S, Spiegelhalter D. Allowing for correlations between correlations in random-effects meta-analysis of correlation matrices. Psychol Methods 2007;12:434–50. CrossrefPubMedGoogle Scholar
Borenstein M, Higgins JP, Rothstein HR. Introduction to meta-analysis. Chichester, UK: John Wiley & Sons, Ltd, 2009. Google Scholar
Dickinson H, Webb NJ, Chaloner C, Wynn RF, Bonney DK. Pseudohyperkalaemia associated with leukaemic cell lysis during pneumatic tube transport of blood samples. Pediatr Nephrol 2012;27:1029–31. CrossrefPubMedWeb of ScienceGoogle Scholar
Streichert T, Otto B, Schnabel C, Nordholt G, Haddad M, Maric M, et al. Determination of hemolysis thresholds by the use of data loggers in pneumatic tube systems. Clin Chem 2011;57:1390–7. CrossrefWeb of ScienceGoogle Scholar
About the article
Published Online: 2017-05-05
Published in Print: 2017-10-26
Author contributions: All the authors have accepted responsibility for the entire content of this submitted manuscript and approved submission.
Research funding: None declared.
Employment or leadership: None declared.
Honorarium: None declared.
Competing interests: The funding organization(s) played no role in the study design; in the collection, analysis and interpretation of data; in the writing of the report; or in the decision to submit the report for publication.