Annals of Emergency Medicine
Volume 39, Issue 3 , Pages 334-337, March 2002

Headaches from practice guidelines☆

Department of Emergency Medicine, University of Florida Health Science Center Jacksonville, Jacksonville, FL

Article Outline

Abstract 

Wears RL. Headaches from practice guidelines. Ann Emerg Med. March 2002;39:334-337.

 

See related article, p. 215 .

It has long been noted that the development and dissemination of practice guidelines do not in themselves result in changes in physicians' practices,1, 2, 3 so we should not be surprised by Vinson's4 report in this issue of Annals noting that opioids are more commonly used for treating migraine headache in the emergency department than are the nonopioid drugs promoted as first-line therapy in clinical practice guidelines issued by expert groups such as the Canadian Headache Society5 and the US Headache Consortium.6 Because, in other situations, physicians are overly enthusiastic adopters of new (and as-yet-unproven) treatments,7, 8 their failure in this instance to adopt practices that are based on authoritative, evidence-based recommendations in three fourths of their patients begs for exploration.

We know something about this, although our knowledge has been tersely characterized by Poses9 as, “Some big disappointments, no home runs, and no easy way to predict success.” Cabana et al3 reviewed studies of guideline implementation and developed a framework that included 7 general categories of barriers, grouped in 3 large dimensions reflecting the temporal order of behavioral change (knowledge, attitudes, and behavior).

Knowledge-based problems include issues such as lack of awareness of guidelines and lack of detailed familiarity with their recommendations. It is likely in this situation that lack of awareness and familiarity played significant roles in the physicians' behavior, because both guidelines were issued by specialty bodies whose contact and influence with practicing emergency physicians are extremely limited.

Attitude-based problems include lack of agreement, either with a specific guideline or with guidelines in general, and also such factors as lack of self-efficacy (the belief that one can actually perform the behavior), lack of outcome expectancy (the belief that adherence will result in the desired outcome), and the inertia of previous practice. In this case, it seems likely that inertia, perhaps coupled with specific disagreement, might play a role. For busy clinicians who encounter these guidelines, finding that they both state that opioids are effective might be sufficient to keep them from reading any further—why change if what you're already doing is known to be effective? A second inertial component is that clinicians in this setting are not likely to perceive the choice of therapy in acute migraine to be a problem for them, so from their point of view, the guidelines are solutions looking for problems rather than the other way around. It seems reasonable that change will occur rapidly if changing solves a problem for the clinician—consider the relatively rapid diffusion of rapid sequence intubation in emergency care—but for most clinicians, managing patients with migraines is not a problem that needs solving.

At first glance, lack of outcome expectancy might not seem to be a problem here, but on reflection, it likely is. An element of diagnostic uncertainty surrounds many headache patients seen in EDs, but the agents recommended by the guidelines are specific for migraine. For cases in which clinicians exceed some threshold of diagnostic uncertainty, it might be rational to avoid migraine-specific treatments and choose an agent that is likely to be more general and with which they are more familiar.

Finally, there may be general objections to the entire idea of guidelines by some clinicians. These objections have usually been presented as somewhat romantic appeals to the “art of medicine” and disparagements of “cookbook medicine.” In other cases, general objections might be based on unintended results achieved by some guidelines. For example, back pain guidelines intended to reduce the use of radiographs actually increased usage instead10; a guideline for congestive heart failure was found to increase length of stay without improving outcomes11; and an enforced guideline limiting Medicaid drug payments resulted in increased nursing home admissions.12 It only takes a few such surprises, coupled with reports of guideline developers' poor adherence to methodologic standards,13 to justify feelings of distrust toward expert pronouncements.

However, outside the affective realm, there are 3 legitimate, deep intellectual and philosophical issues about practice guidelines that have not gotten much discussion to date. First, the entire idea of prescriptive, “feedforward” guidance in medicine has not been seriously examined with an eye to its proper role. Although the idea of prescriptive guidance has superficial appeal, studies in other fields have suggested that there should be a balance between prescriptive and discretionary controls based on the nature of the task and the nature of the worker.14 Perrow15 pointed out many years ago that there are 2 characteristics of tasks that make them hard to be proceduralized. The first factor is the number of exceptional cases, that is, novel and unexpected situations requiring specific adjustment. The second is the nature of the search for solutions, that is, whether it is easy, yielding to analysis or simple rules, or hard, requiring knowledge-based processing.16, 17 If the likelihood of exceptional cases is low and the search space is simple, prescriptive, specific, feed-forward guidance in the form of protocols is likely to be successful. If, on the other hand, the number of special cases is high and the task of finding solutions is difficult, standardization and prescriptive control will be too rigid to adapt to novel circumstances. Reason14 has commented that “it is curious that such a bastion of discretionary action as medicine should be moving towards a feedforward mode of control when many other hitherto rule-dominated domains … are shifting towards performance-based controls and away from prescriptive ones.” Although it would seem that management of acute migraine would lend itself to prescriptive control, as a general proposition, we in health care have yet to examine the clinical problems we face and to establish for each its appropriate place on the prescriptive-discretionary continuum. Notably, there have been recent calls to examine this issue as a means of improving health care quality.18

The second deep, seldom-discussed problem is the assumption that the attributes of aggregates can (or should) be applied to their individual members. Clinical policies are necessarily based on estimates of average outcomes, but average outcomes are not the only thing that matters.19 The problem is that, even if the average outcomes envisioned by guideline developers could be obtained in large groups of patients, individual patients may have only one “shot” at an outcome. Thus, they might perceive their personal risk of an unfavorable outcome as unacceptably high, even though the aggregate risk experienced by a large group of similar patients is not.20 In addition, although patients can pool their financial risks to reduce them, they cannot do so with their health care risks. Asch and Hershey20 argue that a dairy farmer interested in maximizing milk production might try a recommended intervention knowing that the risk of a decrease in some cows' production will be offset by the increase in others, but that, by taking a societal perspective and assuming a redistribution of outcomes, health policy analysts (including guideline promoters) are treating people like cows.

The third deep problem is related to the second. The risk of a given strategy can be reduced in 2 ways: by spreading it over many individuals as just described or by implementing many different strategies. This is analogous to diversification in financial planning and has been used in agriculture,21, 22 but has not been discussed in health care. The dominant assumption in health care has been that variation is bad, that if it exists someone must be doing something wrong, and that the road to improvement necessarily involves eliminating variation. However, studies of organizational reliability and quality in other domains have emphasized the need for a type of variation to produce reliable outcomes.23 This issue of whether or in what circumstances guidelines should avoid “putting all their eggs in one basket” has yet to be addressed substantively by guideline developers and health policy analysts.

Cabana et al's3 third and final dimension encompasses external barriers to behavioral change, given that knowledge and attitudes are ripe. This includes problems such as patient preferences, the complexity or difficulty of a given guideline, the existence of conflicting guidelines, and environmental barriers such as organizational, financial, or legal constraints. Patient preferences almost certainly played a role in Vinson's4 findings. It would be a brave and curious patient indeed who, in the face of an acute migraine, would choose a new, possibly better drug over one that has worked well without significant side effects in the past.

With all of these difficulties, what would be a good strategy for guideline developers and implementers to follow to have a better chance of success? Poses24 has suggested that 3 questions should be answered, in order, before we attempt to modify physicians' behavior.

First, does current behavior really need to be changed? This implies 2 subquestions. Is there evidence that one alternative is preferable to others in a given situation? And, if so, is there evidence that physicians are not choosing this alternative when they should? Vinson's4 study answers the second subquestion, but that answer may be premature because it is not so clear that the first has been satisfactorily addressed. Although the guidelines certainly recommend using nonopioids as the first choice in migraine headaches, their supporting evidence is not so clear— both guidelines agree that there is evidence that opioids are effective, and neither cites evidence of a direct comparison demonstrating either opioids or nonopioids to be superior. In fact, one group flatly states that there is insufficient evidence to choose between effective therapies.6 If there is not an evidence-based argument for superiority, then there is little moral standing to intervene by changing behavior.

If the answer to the first main question is affirmative, then the next question is: What specifically is the problem with the physicians' decisionmaking that leads them to choose the incorrect alternative? It is unlikely that physicians purposely choose suboptimal strategies out of general obtuseness or perversity. A detailed understanding of the barriers, risks, and rewards associated with the various courses of action is required. We have learned what some of the general issues are,2, 3, 25, 26, 27, 28 but each particular situation will need an assessment of exactly how those factors play out in its own context. Interventions implemented without a clear understanding of why physicians did not choose the preferred behavior are likely to be congenitally doomed.

Poses's24 last question is: Given what we have learned from the second question, what is the best method to effect change? In principle, finding the cognitive and other problems that produce the “wrong” behavior could lead to targeted interventions that should be more effective. Education tends to be the most commonly proposed intervention, but its track record has been dismal,29, 30, 31, 32 although specific modalities (eg, interactive programs with opportunities for practice, programs targeted at changing specific types of behaviors) have demonstrated some success in changing practice33 and, occasionally, patient outcomes.31 Because practice guidelines can be thought of as a form of didactic education,2 it should not be surprising that their effectiveness has been underwhelming. Other modalities, such as feedback, direct participation in change management, administrative rules, and financial incentives, have potential but have been less commonly used and seldom studied. Given the risk of unintended consequences from seemingly reasonable guidelines,10, 11, 12 the more coercive a change modality is, the stronger the evidence answering the first 2 questions must be.

So, should the attempt to formulate practice guidelines be abandoned in light of the theoretical problems of their role and the practical problems of implementation? I think not. A better course of action would be to examine the theoretical and practical issues carefully and to use them to come up with better guidelines or a portfolio of guidelines that are more sensitively tuned to the realities of practice and patients and more thoughtfully implemented in the real world of clinicians.

Back to Article Outline

References 

  1. Lomas J, Anderson GM, Domnick-Pierre K, et al.  Do practice guidelines guide practice? The effect of a consensus statement on the practice of physicians. N Engl J Med. 1989;321:1306–1311
  2. Greco PJ, Eisenberg JM. Changing physicians' practices. N Engl J Med. 1993;329:1271–1273
  3. Cabana MD, Rand CS, Powe NR, et al.  Why don't physicians follow clinical practice guidelines? A framework for improvement. JAMA. 1999;282:1458–1465
  4. Vinson DR. Treatment patterns of isolated benign headache in US emergency departments. Ann Emerg Med. 2002;39:215–222
  5. Pryse-Phillips WE, Dodick DW, Edmeads JG, et al.  Guidelines for the diagnosis and management of migraine in clinical practice. Canadian Headache Society. CMAJ. 1997;156:1273–1287
  6. Matcher DB, Young WB, Rosenberg JH, et al.  Evidence-based guidelines for migraine headache in the primary care setting: pharmacological management of acute attacks. Available at http://www.aan.com/public/practiceguidelines/03.pdfNovember 30, 2001; Accessed
  7. Holley MT, Morrissey TK, Seaberg DC, et al.  Ethical dilemmas in a randomized trial of asthma treatment: can Bayesian statistical analysis explain the results?. Acad Emerg Med. 2001;8:1128–1135
  8. Lewis RJ. Are clinician-investigators Bayesian?. Acad Emerg Med. 2001;8:1179–1181
  9. Poses RM. Changing physician behavior: what should we do? What can we do [course documentation]?. Toronto, Ontario, Canada: Society for Medical Decision Making Short Course; 1996;
  10. Suarez-Almazor ME, Belseck E, Russell AS, et al.  Use of lumbar radiographs for the early diagnosis of low back pain. Proposed guidelines would increase utilization. JAMA. 1997;277:1782–1786
  11. Weingarten S, Riedinger M, Conner L, et al.  Reducing lengths of stay in the coronary care unit with a practice guideline for patients with congestive heart failure. Insights from a controlled clinical trial. Med Care. 1994;32:1232–1243
  12. Soumerai SB, Ross-Degnan D, Avorn J, et al.  Effects of Medicaid drug-payment limits on admission to hospitals and nursing homes. N Engl J Med. 1991;325:1072–1077
  13. Shaneyfelt TM, Mayo-Smith MF, Rothwangl J. Are guidelines following guidelines? The methodological quality of clinical practice guidelines in the peer-reviewed medical literature. JAMA. 1999;281:1900–1905
  14. Reason J. Managing the Risks of Organizational Accidents. Aldershot, United Kingdom: Ashgate Publishing Co; 1997;
  15. Perrow C. Framework for the comparative analysis of organizations. Am Sociol Rev. 1967;32:194–208
  16. Rasmussen A, Jensen J. Mental procedures in real-life tasks: a case study of electronic trouble-shooting. Ergonomics. 1974;17:293–307
  17. Reason J. Human Error. Cambridge, United Kingdom: Cambridge University Press; 1990;
  18. Committee on Quality of Health Care in America . Crossing the Quality Chasm: A New Health System for the 21st Century. Washington, DC: National Academy Press; 2001;
  19. Deber RB, Goel V. Using explicit decision rules to manage issues of justice, risk, and ethics in decision analysis: when is it not rational to maximize expected utility?. Med Decis Making. 1990;10:181–194
  20. Asch DA, Hershey JC. Why some health policies don't make sense at the bedside. Ann Intern Med. 1995;122:846–850
  21. Galligan DT, Marsh WE. Application of portfolio theory for the optimal choice of dairy veterinary management programs. Prev Vet Med. 1988;5:251–261
  22. Galligan DT, Ramberg C, Curtis C. Application of portfolio theory in decision tree analysis. J Dairy Sci. 1991;74:2138–2144
  23. Weick KE, Sutcliffe KM. Managing the Unexpected: Assuring High Performance in an Age of Complexity. 1st ed. San Francisco, CA: Jossey-Bass; 2001;
  24. Poses RM. One size does not fit all: questions to answer before intervening to change physician behavior. Jt Comm J Qual Improv. 1999;25:486–495
  25. Veatch RM. Reasons physicians do not follow clinical practice guidelines [letter]. JAMA. 2000;283:1685–1686
  26. Poses RM. Conclusions about why doctors change their practice were not supported by the data [letter]. BMJ. 1997;314:1908
  27. Freeman AC, Sweeney K. Why general practitioners do not implement evidence: qualitative study. BMJ. 2001;323:1100–1102
  28. Fairhurst K, Huby G. From trial data to practical knowledge: qualitative study of how general practitioners have accessed and used evidence about statin drugs in their management of hypercholesterolaemia. BMJ. 1998;317:1130–1134
  29. Haynes RB, Davis DA, McKibbon A, et al.  A critical appraisal of the efficacy of continuing medical education. JAMA. 1984;251:61–64
  30. Cantillon P, Jones R. Does continuing medical education in general practice make a difference?. BMJ. 1999;318:1276–1279
  31. Davis D, O'Brien MA, Freemantle N, et al.  Impact of formal continuing medical education: do conferences, workshops, rounds, and other traditional continuing education activities change physician behavior or health care outcomes?. JAMA. 1999;282:867–874
  32. Davis DA, Thomson MA, Oxman AD, et al.  Evidence for the effectiveness of CME. A review of 50 randomized controlled trials. JAMA. 1992;268:1111–1117
  33. White CW, Albanese MA, Brown DD, et al.  The effectiveness of continuing medical education in changing the behavior of physicians caring for patients with acute myocardial infarction. A controlled randomized trial. Ann Intern Med. 1985;102:686–692

☆ Reprints not available from the author.Address for correspondence: Robert L. Wears, MD, MS, Department of Emergency Medicine, University of Florida Health Science Center Jacksonville, 655 West 8th Street, Jacksonville, FL 32209; 904-244-4124; E-mail wears@ufl.edu.

PII: S0196-0644(02)75569-8

doi:10.1067/mem.2002.122180

Refers to article:

  • Treatment patterns of isolated benign headache in US emergency departments

    David R. Vinson, Author contribution is provided at the end of this article. Author contribution: DRV is the sole author and takes responsibility for the paper as a whole.
    Annals of Emergency Medicine March 2002 (Vol. 39, Issue 3, Pages 215-222)

Annals of Emergency Medicine
Volume 39, Issue 3 , Pages 334-337, March 2002

Article Tools