BY-NC-ND 3.0 license Open Access Published by De Gruyter April 11, 2014

Path dependence and routines: a threat to capability development

Lars E. Sjödahl
From the journal Diagnosis


This article introduces the concept path dependence based on case illustrations mainly from psychiatric health care. The concept path dependence is widely used in economic research, history, social sciences and law, but so far seldom in psychology and health care, although equally applicable here. Two variants of path dependence, one in a narrow sense and one in a broader sense are defined and exemplified by case illustrations from psychiatric health care and patient safety systems.


The concept path dependence as a research tool is widely used in economic research, history, social sciences and law [1–4] but so far seldom within psychology and health care. Definitions vary depending on the subject content. In a narrow context, path dependence refers to studies of how your decisions here and now, under defined conditions, are dependent on earlier decisions under similar conditions. On a more general, broader level, path dependence refers to studies of complex causal relations over time.

Bennett and Elman [5] suggest the following approach to the variety of definitions found in the literature concerning the concept path dependence: “We argue that dissimilarities among political scientists on the concept of path dependence can be represented in terms of different content scholars give to, and emphasis they play on four elements common to most accounts: causal possibilities, contingency, closure and constraint” (p. 252). We agree with this relativistic stand when we exemplify the concept path dependence and apply it to routine behavior and complex causal relations over time.

Path dependence in a narrow sense

In a narrow sense path dependence is synonymous with habitual routines, defined by Gersick and Hackman [6] as follows: “A habitual routine exists when a group repeatedly exhibits a functionally similar pattern of behavior in a given stimulus situation without explicitly selecting it over alternative ways to behaving” (p. 69). This definition also applies to our single cases, which illustrate path dependence in a narrow meaning, i.e., what you do here and now depends on your previous behavior in similar situations. In our first illustration (below), an excerpt from a patient chart, we see a pattern of drug prescriptions, continuously repeated 5 times over a period of 17 months without any commentaries or face-to-face contact between doctor and patient.

Excerpt from a patient chart

14 Jan prescription

  • Tabl Haldol 1 mg No C, ds 1 tabl twice/day, iter bis

  • Tabl Fenfluramine 25 mg No C, ds 1 tabl twice/day, iter bis

  • Tabl Nozinan 5 mg No C, ds 2 tabl at bedtime when needed, iter bis.’

  • Tabl Mogadon 5 mg No L, ds 1 tabl at bedtime when needed. Iter

  • Tabl Tranylaypramine 10 mg No C, ds 1 tabl when needed. Iter.

  • Prescription sent to patient’s home address

12 May prescription

  • Tabl Haldol 1 mg, No C, ds 1 tabl twice/day, iter bis

  • Tabl Fenzfluramine 25 mg No C, ds 1 tabl twice/day, iter bis.

  • Tabl Nozinan 5 mg No C, ds 2 tabl at bedtime when needed, iter bis

  • Tabl Mogadon 5 mg, No L, ds 1 tabl at bedtime when needed iter.

  • Tabl Tranylaypramine 10 mg No C, ds 1 tabl when needed. Iter.

  • Prescription sent to patient’s home address. And so on five times over a period of 17 months.

The following episode illustrates how path dependence, in our narrow sense, breeds rigidity, obstructs creativity and search for other alternatives. In the Swedish daily paper, Dagens Nyheter (DN), Johansson [7], an 80-year-old woman who had been admitted to the geriatric department of a well-known hospital in Stockholm recounts the following episode: “When I wanted to read my book I found that neither of the light bulbs in the bed-lamps were working. They were both broken. However, I was then told that in this hospital the responsibility for changing light bulbs lay with an external contracted firm and that it was too late in the day to contact them. The nurse, who had no authority to change light bulbs, asked me: ‘What can you have to read just now, this evening?’”

As illustrated above, the consequences of habitual routines, path dependence in our narrow sense, will be diminished opportunities for growth in competence and capability development. Gersick and Hackman [6] describe the situation as follows: “…over time both motivation and opportunities for learning may diminish gradually, without anybody recognizing that these changes are taking place” (p. 73).

Path dependence in a broader sense

Path dependence in a broader sense is a conceptual tool for studies of complex causal relations over time. A more comprehensive definition can be found in an article by Gawande [8] in The New Yorker: “With path-dependent processes, the outcome is unpredictable at the start. Small, often random events early in the process are ‘remembered’, continuing to have influence later. And, as you go along, the range of future possibilities gets narrower. It becomes more and more unlikely that you can simply shift from one path to another, even if you are locked in one path that has a lower payoff than an alternative one.”

Paul Krugman’s (Nobel laureate for Economics 2008), discovery of path dependence between geographic location and trade patterns (see 8), actualizes the following two sayings “history matters”, and “before considering the future, take a look into the past.” We shall follow this advice by briefly presenting two illustrations from psychiatric diagnostics and health care. The first one deals with Swedish psychiatry during later years, starting with 2003. The latter presentation covers the period from the foundation of the first psychiatric university department in Berlin 1865 [9] up to today’s diagnostic criteria in psychiatry, the DSMs. In a short essay like the present one, the presentations have to be in the form of outlines. First we take a look at the Swedish illustration.

Swedish psychiatry during recent decades

In Sweden during 2003, there was a series of five tragic incidents, some of them with mortal outcome and many leaving seriously injured people. For example, a man drove his car at high speed along a narrow pedestrian street in the old part of Stockholm killing two people and injuring another 20. The driver insisted somebody else had been driving the car – even though he was behind the steering wheel. The incident that had the greatest impact on the general public was, however, the murder of the Swedish foreign minister, Anna Lindh, in a department store in Stockholm. These tragic events have one common denominator: the violence was committed by men in need of psychiatric help, and some of them had been refused admittance to psychiatric clinics. Jennekvist [10] of the Stockholm Police, with overall responsibility for investigating the murder of Anna Lindh, describes the defendant’s situation as follows: “When nobody, not any one psychiatric clinic he had turned to was willing to listen to his problems it was just as well to continue taking his pills then at least his body got some rest” (p. 137). In this case, we see how one routine behavior, prescribing drugs, reinforces another routine behavior, the patient’s relying on taking medication to get some rest.

To learn from this path, this series of events related to each other in a complex causal way, Socialstyrelsen (the Swedish National Board of Health and Welfare) carried out some content analysis of patient charts [11]. Below is a short summary of results from one case study:

  • Anamnesis is fragmented.

  • Information search fragmented, psychological investigation started but not completed.

  • Diagnoses defective, multidimensional judgment non-existing.

  • Treatment plan, based on diagnoses, non-existing.

  • Length of treatment course and care not possible to reconstruct from patient chart and notes.

  • Documentation defective, care notes destroyed after being summarized.

The scenario presented above, literally forced Sweden’s government to start a nation-wide investigation into the responsibilities, competence and work-processes in psychiatry and mental care. This was done in 2003 when Anders Milton, former head of the Swedish Medical Association, was appointed national coordinator for this investigation. Three years later, in November 2006, Dr. Milton presented the results in Sweden’s main morning paper, Dagens Nyheter (DN) under the following heading: “Change the laws for psychiatry so it becomes possible to identify those who fail in their responsibility” [12]. Milton underlines that the deficiencies and flaws he found in psychiatric care were of such a magnitude that they would never have been accepted within any other area of medical care.

History matters

In our second presentation of path dependence in the broader sense, i.e., that “history matters”, we will briefly mention some critical events that have certainly influenced both psychiatric diagnostics and clinical practice from the middle of the 19th century up to today’s construction of descriptive non-etiological criteria manuals like the DSMs. This period of time presents with a series of events, the causal influences of which are complex and important for causal understanding of today’s psychiatric diagnostics.

In 1865 psychiatry was given the status of an own institution at the University of Berlin. In 1887 Karl Kraepelin published the second edition of his famous textbook Compendium, Textbook of Psychiatry, where he tried, based on a large number of case reports, to construct a nosological classification system comparable to that used in somatic medicine. Today’s DSMs can be seen as remains of Kraepelin’s big ideas about a nosological system for psychiatry. Further, two Nobel laureates, Egas Moniz (1949) with his lobotomy, and Arvid Carlsson (2000) with his dopamine hypothesis, may also have influenced both treatment alternatives and diagnostics within psychiatry up to today’s treatment practices in psychiatry. It would be an interesting enterprise to describe and analyze eventual causal relations under this period, a path leading up to today’s non-etiological criteria manuals and drug prescriptions within psychiatric diagnostics and treatment.

Patient safety – risks for path dependence

Referring to the World Health Organization (WHO), Gigerenzer [13] presents the following picture of the safety situation within medical care: “In the U.S. the average physician-patient contact is 5 min. Most of the information is presented in a vocabulary that is unintelligible to the patient. Patients tend to views of fate or ‘Inshalla’ rather than learning to practice informed consent. All is God’s will, or the physician’s, why should they worry? The Institute of Medicine estimated that some 44,000 to 98,000 patients are killed every year in U.S. hospitals by preventable medical errors and misadventures.” Comparing the medical field with commercial aviation Gigerenzer [13] points out a reasonable cause for the poor situation within medical care: “…But safety systems such as in commercial aviation …have not been set up in hospitals” (p. 21). Safety systems in commercial aviation are based on Flanagan’s critical incident method [14, 15]. This approach rests on the following four assumptions [15]:

  • “No planning and no evaluation of specific behaviors are possible without a general statement of objectives.”

  • “…discussions have failed to emphasize the dominant role of the general aim in formulating a description of successful behavior adjustment in a particular situation.”

  • “It is clearly impossible to report that a person has been either effective or ineffective in a particular activity by performing a specific act unless we know what he is expected to accomplish.”

  • “A basic condition necessary for any work on the formulation of a functional description of an activity is a fundamental orientation in terms of the general aims of the activity.”

In two different studies by Sjödahl [16, 17] the critical incident method has been used to bridge the gap between practice and theory in the fields of forestry research and health care. Both these studies, partly dealing with the safety aspect, exemplify how the Brunswikian concept of representative ecological sampling can be defined as aspects considered important for man’s adaption to the real world [18].

The Swedish medical safety system is regulated by the Social Service Law from 2001. It contains, among others, two chapters dealing with medical practice called Lex Maria and Lex Sarah. Both describe safety duties and reporting routines; the former general in scope, the latter restricted to the care of elderly people.

For many years, a special agency, the Medical Responsibility Board (HSAN), working in cooperation with the Swedish National Board for Health and Welfare, had been in charge of complaints from patients and hospitals. In 2011 a new Patient Safety Law was introduced. Critical incidents and iatrogenic damage, due to wrong diagnosis or faulty treatment were to be reported directly to the National Board. However, HSAN should still decide on corrective measures. In a follow-up study [19] the actual situation looks grim. During 2011, only half the number of cases usually dispatched during a year were referred from the National Board to HSAN for corrective measures. Besides thousands of cases dated 2011, there was a backlog of about 1400 complaints from 2010 waiting to be dealt with. This delay means that some clients may have to wait more than 2 years before receiving a response to their complaints. These long waiting periods illustrate how path dependence, i.e., routine delays, with no cognitive or task feedback, creates a very inefficient experiential learning situation, called “wicked” by Hogarth [20]. In June, 1, 2012 the Swedish parliament initiated a new agency called the Inspectorate of Health and Care, that was to work independently. Further, during recent years two new professorships of Patient Safety have been established in Sweden [Richard Cook (USA), and Hans Ruthberg (Sweden)].

Psychiatric diagnostics: risks for path dependence

The following comments and illustrations refer to the first four versions of the DSM criteria.

Kaplan and Sadock [21] present the DSM as follows: “The approach to DSM-IV as it was in DSM-III is a-theoretical with regard to causes. Thus DSM-IV attempts to describe the manifestations of mental disorders; only rarely does it attempt to account for how the disturbances come about. The definitions of the disorders consist of descriptions of the clinical features” (p. 29).

In the first number of Diagnosis, Phillips [22] discusses consequences for psychiatric diagnostics without biological of other scientific support. This situation raises many questions, for example: how do you in diagnostic practice integrate idiographic and nomothetical knowledge? The former is about single events and situations, the latter about more generalized knowledge. In absence of biological bio-makers this integrating process becomes a decisive moment in psychiatric diagnostics. Maybe, Lewin’s [23] advice could be of some help in this situation: “The application of laws presupposes the comprehension of the individual case. One can apply a law only if one knows the nature of the concrete case with which one is dealing.” Certainly, the DSM’s descriptive syndrome clusters may lead to mixed dimensional/categorial or prototype decisions as described by Phillips [22]. The idea to switch over to dimensional judgements has probably been considered more than once. It is easy to understand that such an enterprise would require substantial work with constructing rating scales. From the introductory pages of DSM-IV it is clear that the manual makes no assumptions about absolute boundaries between mental disorders; it is not a category system in any strict meaning. Further, the textual symptom descriptions are meant to be used by individuals with appropriate training and experience in mental diagnostics, not in a cookbook fashion, mechanically by the untrained. For training purposes, the manual is therefore supplemented with a casebook, seldom mentioned in discussions about the use of the DSM criteria. Even though a shallow criteria system like DSM-IV may foster prototype matching or routine categorization, i.e., path dependence in our narrow sense, we do not know for certain how the manual is used in clinical practice [24]. The following strategies could all be represented more or less among diagnosing psychiatrists:

  • I follow the rules of DSM-IV descriptions of symptoms and syndromes as strictly as possible.

  • I usually compare with my colleagues: what is typical for a certain diagnostic category.

  • Usually I rely on memories of recent and earlier patients with similar symptoms.

  • I usually support my judgments and diagnoses by reflecting on possible causes behind the patient’s symptoms or syndromes.

  • I am very careful about the boundaries between symptoms and syndromes to avoid, for example, categorizing a non-psychotic patient as psychotic.

  • A patient’s psychosocial context and personality characteristics have also influenced my final diagnosis.

  • I have to take into account time restraints and availability of existing treatment alternatives before I make the final diagnosis.

It is likely that psychiatrists, psychologists and therapists make at least some causal inferences about their patients’ symptoms [25]. Then, the crucial question becomes: What information should be searched for besides descriptive symptom labeling, according to a non-etiological criteria manual? This is partly an empirical question: To what extent do different psychiatrists and psychologists differ in their diagnostic information search?

Haskel [26] has pointed out that the vast majority of diagnostic errors have their roots in patient-provider relations. This is even more so in psychiatric diagnostics, where the patient-provider relations can be very extensive in both time and space. Evaluative, retroactive first-person accounts, based on patients’ chart notes, would be one way to detect unknown error routines, i.e., path dependence. The need for highly organized follow-up periods, pointed out by Weed and Weed [27], seems also to be of crucial importance in psychiatry where relying on routines, path dependence, sometimes results in unbelievable misery. In a post-mortem study by Nilsson and Lögdberg [28] many of the patients had been lying dead for weeks or months before being discovered.

In somatic medicine it has been underlined that diagnostic information search is teamwork involving many partners [29, 30]. This integrative aspect, involving different information sources and knowledge systems, is of special importance in psychiatric diagnostics.

In a patient’s chart we read: “First psychoactive drugs, then deal with the patient’s concrete problems”. However, nowhere in that chart, covering 10 years, is there any documentation from discussions about the patient’s social and vocational problems, which were of a considerable magnitude and complexity. In another patient’s chart with a vague first diagnosis of “depressive reactions” we find a series of 10 anti-psychotic drugs (neuroleptics) prescribed over a long period of time interspersed here and there with one and the same diagnostic note: “no signs of psychosis”. This looks very much like what is called post-diagnosis, i.e., first search for an effective drug, then decide on a diagnosis. Kirk and Kutchins [31] describe this diagnostic routine with the following metaphor: “If it was pounded with a hammer it must have been a nail” (p. 235). In a report from the Swedish Council of Technology Assessment in Health Care [32] the overuse of antipsychotic drugs receives severe criticism. In about 30% of all cases, prescriptions are regarded unacceptable from an ethical and medical point of view. Björn Beerman, head of the Swedish Medical Products Agency, asserts that about 50% of patients on antipsychotic drugs have illnesses that should not be treated with these drugs [33]. Nobody knows to what extent routine post-diagnoses are responsible for these alarming figures. Lack of treatment alternatives may also lead to routine path dependence in psychiatric diagnostics. Kirk and Kutchins [31], when practicing in a psychiatric clinic, report the following observations: “The staff at one agency explained that diagnoses were the art of making distinctions without differences. Psychiatrists would frequently debate the fine points about the correct diagnosis for a disturbed client, but would always prescribe Haldol regardless of the outcome of the diagnostic debate” (p. 235).

Suggestions which may help to improve psychiatric diagnostics and prevent path dependence

  • Specify purposes, goals with diagnoses in psychiatry.

  • Plan and organize follow-up periods for patients and search for criteria like quality of life aspects.

  • Integrate patient-specific and context-dependent information with diagnostic symptom descriptions.

  • Evaluate diagnoses with help of retroactive first-person accounts.

  • Try for a more holistic information search.

  • Better information to patients about drug effects and treatment alternatives.

  • Better understanding among diagnosing personnel of mental attributes as relational events, i.e., caused by interaction between organism and environmental conditions.

  • Further professional education of ward personnel.

  • Extended interdisciplinary research, e.g., about causal relations, etiological models and theories.

Corresponding author: Lars E. Sjödahl, Persikevagen 11, SE22355 Lund, Sweden, E-mail: ,

  1. Conflict of interest statement

  2. Author’s conflict of interest disclosure: The author stated that there are no conflicts of interest regarding the publication of this article.

  3. Research funding: None declared.

  4. Employment or leadership: None declared.

  5. Honorarium: None declared.


1. Holmberg R. Survival of the unfit: path dependence and the Estonian oil shale industry. PhD thesis. Linköping, Sweden: Linköping University, Department of Technology and Social Change, 2008. Search in Google Scholar

2. David PA. Path dependence – a foundational concept for historical social science. Stanford, CA: Stanford University, Department of Economics, 2006. Search in Google Scholar

3. Page SE. Path dependence. Quart J Polit Sci 2006;1:87–115. Search in Google Scholar

4. Larsson S. The path dependence of European copyright. 2011:8:1 SCRIPTed 8. Available at: Search in Google Scholar

5. Bennett A, Elman C. Complex causal relations and case study methods: The example of path dependence. Polit Anal 2006;14:250–67. Search in Google Scholar

6. Gersick CJ, Hackman JR. Habitual routines in task-performing groups. Organ Beh Hum Dec Process 1990;47:65–97. Search in Google Scholar

7. Johansson G. Gärd Johanssons egen berättelse om dagarna på geriatriken. [Gärd Johanssons own story about her stay in the Dept. of Geriatric Care]. Dagens Nyheter 2013:09-02 [In Swedish]. Search in Google Scholar

8. Gawande A. Getting there from here. How should Obama reform health care? The New Yorker 2009:01-26. Search in Google Scholar

9. Bentall R. Madness explained. Psychosis and human nature. London, UK: Penguin Books, 2003. Search in Google Scholar

10. Jennekvist L. Mordet på Anna Lindh. [The murder of Anna Lindh]. Falun: Bonnier Fakta, 2005 [In Swedish]. Search in Google Scholar

11. National Board of Health and Welfare. SOS Dnr.41-7638/03. 2004 [In Swedish]. Search in Google Scholar

12. Milton A. “Ändra lagarna för psykiatrin så att ansvariga kan pekas ut”. [Change the laws pertaining to psychiatry so that those who are responsible can be pointed out]. Dagens Nyheter 2006:11-11 [In Swedish]. Search in Google Scholar

13. Gigerenzer G. Reckoning with risk. Learning to live with uncertainty. London: Penguin Books, 2002. Search in Google Scholar

14. Greenwell WS. Learning lessons from accidents and incidents involving safety-critical software systems. MSc. thesis. USA: Faculty of the School of Engineering and Applied Science, University of Virginia, 2003:05. Search in Google Scholar

15. Flanagan J. The critical incident technique. Psych Bull 1954;51:327–58. Search in Google Scholar

16. Sjödahl L. Representative design and sampling. The Brunswik Society Newsletter, 2007:20. Access at: Search in Google Scholar

17. Sjödahl L. Brunswik’s lens model adapted to aspect representation. The Brunswik Society Newsletter 2009:37. Access at: Search in Google Scholar

18. Björkman M. On the ecological relevance of psychological research. Scand J Psychol 1969;10:145–57. Search in Google Scholar

19. Örstadius K. Anmälningar mot läkare lagts på hög. [Reports about doctors put aside]. Dagens Nyheter 2013:02-10 [In Swedish]. Search in Google Scholar

20. Hogarth RM. On the learning of intuition. In: Plesner H, Betsch C, Betsch T, editors. Intuition in judgment and decision-making. New York, London: Lawrence Erlbaum Associates, 2008:91–105. Search in Google Scholar

21. Kaplan HI, Sadock BJ. Concise textbook of clinical psychiatry. Baltimore, Maryland 21201-2436: Williams & Wilkins, 1996. Search in Google Scholar

22. Phillips J. Detecting diagnostic error in psychiatry. Diagnosis 2014;1:75–8. Search in Google Scholar

23. Lewin K. Principles of topological psychology. New York and London: McGraw-Hill Book Company Inc., 1936. Search in Google Scholar

24. Sjödahl L, Kaufmann E. Patient–clinician cooperation in psychiatry. The Brunswik Society Newsletter 2009:45. Access at: Search in Google Scholar

25. Popper KR. The logic of scientific discovery. London and New York: Routledge, 2000. Search in Google Scholar

26. Haskell HW. What’s in a story? Lessons from patients who have suffered diagnostic failure. Diagnosis 2014;1:53–4. Search in Google Scholar

27. Weed LL, Weed L. Diagnosing diagnostic failure. Diagnosis 2014;1:13–7. Search in Google Scholar

28. Nilsson L-L, Lögdberg B. Dead and forgotten – post-mortem time before discovery as indicator of social isolation and inadequate mental healthcare in schizophrenia. Schizophr Res 2008;102:337–9. Search in Google Scholar

29. McDonald KM. The diagnostic field’s players and interactions: from the inside out. Diagnosis 2014;1:55–8. Search in Google Scholar

30. Graedon T, Graedon J. Let patients help with diagnosis. Diagnosis 2014;1:49–51. Search in Google Scholar

31. Kirk SA, Kutchins H. The selling of DSM. The rhetoric of science in psychiatry. Hawthorne, New York: Aldine de Gruyter, 1992:235. Search in Google Scholar

32. Wiesel F-A, Thelander S. In: Swedish Council of Technology Assessment in Health Care SBU-rapport nr 133/1, 1997:05-27. Search in Google Scholar

33. Spross Å. Psykosmedel används fel. Läkemedelsverket vill minska förskrivningen [Anti-psychotic drugs are misused. Swedish Medical Products Agency wants to greatly reduce the use of anti-psychotic drugs.] Interview with Professor Björn Beermann, Läkemedelsverket. Uppsala Nya Tidning 2000:01-14 [In Swedish]. Search in Google Scholar

Received: 2014-1-2
Accepted: 2014-3-17
Published Online: 2014-4-11
Published in Print: 2014-6-1

©2014 by Walter de Gruyter Berlin/Boston

This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 3.0 License.