[ldquo]Do you want to know what most free thinkers want? Some want the freedom not to think at all...others want the freedom to think badly...and others still want the freedom to despise all authority and tradition.[rdquo] Louis Pasteur3
[ldquo]The good physician treats the disease, but the great physician treats the patient.[rdquo] Sir William Osler23
[ldquo]The value of diversity[mdash]including its extreme form, eccentricity[mdash]lies in creativity, adaptability, and the bypassing of the limits of convention. Innovation springs from dissatisfied minds....[rdquo] Pathologist Carl Gray13
In surgical pathology, diagnostic differences of opinion usually arise as a function of[colon] (1) variation among the classifiers, (2) variation among the classified, and (3) problems with using the systems and methods at the interface between classifier and classified. In the Letters section of the current issue of the journal, Dr. Medline expresses the opinion that variation among the classifiers, [ldquo]individuality[rdquo] to use his term, should be increased to improve diagnosis. 17 Although individuality or personal practice style is unquestionably important both in approaching some cases and in advancing the field of surgical pathology, the letter does not address the point likely to be of most interest to pathologists, clinicians, and patients, that is, when and under what circumstances does diagnostic individuality of the classifier enhance the care of patients, and when is it just another name for quirky, eccentric, or objectively wrong approaches that result in different diagnostic interpretations for the same findings?
Placing a high value on individuality is certainly not Dr. Medline's idiosyncratic opinion. Debates about the degree to which it is appropriate for authorities to constrain the individual's tendency to act as he or she sees fit have been the subject of political debate for over 300 years. Individualism (self-directed, comparatively unrestrained behavior) is classically contrasted with collectivism (individual rights are subordinate to constraints imposed by systems or organizations). In the political sphere, there has been a tendency in recent years for individuality to triumph over collectivism, and historians now view as failures the various attempts to use social policy to strictly standardize how people think and act. 11 Writer David Frum has noted that currently in the United States, there is an [ldquo]obsession with personal freedom[rdquo] that grew out of the cultural and social turmoil of the 1960s and 1970s, essentially replacing the comparatively conformist, obedient attitudes of the past. 21 A similar debate over the right of the individual practitioner to come to his or her own conclusions about patient care 16,24 spans essentially the entire history of modern medicine, and recently this debate has become quite heated. For example, are HMO guidelines an effort to impose evidence-based standards on a workforce prone to irrational variation, or (as recently stated by an Albuquerque physician) are these guidelines an attempt to make physicians [ldquo]paint by numbers[rdquo]?
With regard to the practice of surgical pathology, I am not aware that we have any enumerated right to express individuality in our practices, and in fact there are several powerful constraints on our diagnostic individuality. One constraint is the legal system, which for at least 150 years has identified variation from [ldquo]standard[rdquo] as a reason to pursue legal action. 20 Another constraint on individuality is ethical. Unless the patient has given informed consent to a deviation from standard practice, the pathologist is on extremely shaky ethical grounds when providing anything but standard interpretations. 2,30 It is true that the concept of autonomy is central to ethical standards in medicine, but autonomy is generally viewed as a right of the patient (for example, to be free from coercion) rather than a right of the physician to be free of external rules and guidelines. 12,18 Pathologists find that the legal and ethical environment in which we practice imposes an obligation that we assign to each case the label that represents the best practice of the specialty at the point in time that the case is evaluated. Ideally, identical findings would receive identical labels from all pathologists in all practice settings, although when our role as patient care advocate conflicts with our role as taxonomist, the best practice diagnosis for the same set of pathology findings will sometimes differ.
It is interesting that despite the powerful pressures for diagnostic conformity, surgical pathologists who are attempting to follow current practice standards generate substantial numbers of diagnostic disagreements. 7,15,25 Some disagreements appear to represent examples of cases with an objective [ldquo]right answer[rdquo] that certain pathologists do not recognize, suggesting that there is a gap between what the specialty knows and what individual practitioners do. Other disagreements reveal that there is a gap between what we know as a specialty and what we would like to know.
Of importance, diagnostic disagreements are now occurring in a setting in which medical issues of all sorts are becoming increasingly politicized, and the topic of medical error is of intense interest to groups outside of medicine. The issue of diagnostic individuality raised by Dr. Medline is central to a discussion of error, because this individuality is a source of variation that theoretically can be modified, and in the mind of the public and of many healthcare leaders, variation is becoming increasingly equated with error. For example, variation in clinical care now tends to be labeled by the elite media with pejorative terms such as [ldquo]overuse,[rdquo] [ldquo]underuse,[rdquo] or [ldquo]misuse[rdquo] of proper therapy, 1 and medical specialty organizations are grappling with the failure of individual practitioners to adhere to what these organizations perceive to be compelling evidence-based practice standards. 4,23,29 One study projected up to a 14-year gap between release of trial results and incorporation of results into some physicians' practices. 19 Is nonconformity[colon] (1) simply ignorance or eccentricity, (2) a reflection of the inability of practitioners to apply standards in a consistent fashion, or (3) have practitioners with no opportunity for input on standards concluded that many guidelines are GOBSAT (good old boys sat around a table) that does not apply to their practice setting? 10
In the present political climate, it may be appropriate for surgical pathologists to take a closer look at how well we understand the pathogenesis of our diagnostic disagreements, and whether we are prepared to propose, evaluate, and implement [ldquo]treatment[rdquo] options. We certainly have an extensive experience with studies of how often diagnostic disagreements occur in various settings, but these latter studies do not treat or cure the problem of disagreement any more than documenting that 3000 cases of polio occurred in a certain region in 1952 created or administered a vaccine. Furthermore, pressure to fix the error problem may lead to [ldquo]do something[rdquo] approaches that waste resources or are counterproductive. Especially when the pressure for change is political, there is a danger that the response will also be political, with science added as a window dressing.
Similar to any other medical problem, diagnostic disagreement should be approached as much as possible using a scientific framework. Questions that should be answered include[colon] (1) Is the current incidence of error/disagreement too high? (2) What is a realistic goal for the rate of error/disagreement? (3) What would be the benefits of a reduction? (Ideally, the resources allocated to achieving the goal should be proportional to the benefits of achieving the goal.) (4) What methods do we have or can we develop to meet this goal? (5) Will implementing these methods cause important negative consequences? (6) Will we know that the goal has been met, that is, will the problem be attacked with testable hypotheses accompanied by objective measures of success or failure? (7) Will the measures of success or failure be patient care outcomes or easier to measure surrogates for outcome such as number of cases referred for a second opinion?
Unfortunately, in surgical pathology, it does not appear that we are currently even at step one of this process because there is no consensus within the specialty about what level of disagreement is a problem. A leading pathologist recently interviewed in the New York Times suggested that diagnostic disagreement affecting 1.4[percnt] of cases referred to his institution was sufficient to change the practice of pathology to incorporate routine second opinions. 28 Other leading pathologists have concluded that a study with a case disagreement rate of 33[percnt] at the benign malignant threshold immediately following a tutorial has been [ldquo]incorrectly cited as a study that documents diagnostic disagreement.[rdquo]22 It is true that even one diagnostic disagreement can be an extremely distressing, perhaps even humiliating experience for the pathologist. However, disagreement is not polio, and it is unlikely that with current technology, zero tolerance for diagnostic disagreement is a rational expectation. If zero tolerance cannot be achieved, it falls to spokesmen for the specialty to explain to the public why this is so. Likewise, denying that high rates of disagreement are really disagreement makes pathologists sound like tobacco executives denying that cigarette smoking causes cancer. Finally, while the legal principle of stare decisis (stand by things decided) is not appropriate for retrospective medical review, it is also not appropriate to exaggerate differences of opinion or to use hindsight bias to create the perception of error when none exists.
Current dogma in the field of error reduction holds that too much emphasis has been placed on the individual's contribution to error, and that many errors are caused by system flaws that inevitably lead to error. In pathology, major system factors that could be contributing to error include flawed classification systems, weak (for example, personality-based) diagnostic gold standards, poorly functioning sources of clinical information, and improper specimen preparation and handling. However, if pathology error is more closely linked to variation in individual diagnostic behavior rather than to flawed systems, then efforts to decrease error by changing systems would be misguided. Attention would be better directed to other tasks, such as modification of the training, certifying, and monitoring of pathologists in an attempt to create a less variable workforce. The lesions that we classify are diverse, but we cannot lose sight of the fact that there is also substantial diversity of the classifiers. This pathologist diversity arises out of differences in such factors as intelligence, inquisitiveness, willingness to work hard, interest in the field, quality of residency training, commitment to education after formal education, and volume of cases. Another important difference between pathologists is where they practice on a receiver operator curve (ROC) (favor false-positive or false-negative).
Pathologist variation is like flawed systems in that it can theoretically be modified to reduce diagnostic disagreement. However, experience in a practice setting suggests that despite attempts to homogenize the members of the specialty workforce through mechanisms such as residency training, board testing, CME, and standard textbooks, individuality remains a major contributor to diagnostic disagreement. For example, the spread between the best and the worst diagnosticians completing pathology training is substantial, and directing underachievers into nonpathology careers is a task that can vary from unpleasant to dangerous. 9,14 The Anatomic Pathology Boards reject a subgroup of pathologists who complete residency, but what is the rationale behind the pass/fail threshold and how is the threshold validated? Following initial specialty credentialing, control of performance is left to the tort system, to hospital committees, and to state agencies, and the specialty itself plays no formal role. For example, what steps are taken by academic pathologists when a case is referred labeled with a diagnosis that is not a differential consideration? For private pathologists, what do you do if a consultant makes an implausible diagnosis? How do pathology organizations respond to objectively wrong interpretations of survey cases? 8 Presumably, these responses differ from aviation's response when a United pilot steers his plane toward a mountain. 5 Board recertification is a plausible mechanism to improve and/or standardize the skills of pathologists practicing surgical pathology, but as currently structured, it is not apparent that recertification addresses either of these goals.
An interesting system work-around for diagnostic individuality is increased case referral, and this plan has the distinct advantage of simply changing the frequency of a practice behavior that is already well-established. We would find that if all of the odd breast tumors in a geographic area are prospectively referred to one pathologist, then the individuality of many pathologists would be reduced to the individuality of one pathologist.
Like modern industry, clinical medicine finds itself under pressure to both conform and to innovate. In medicine, these twin goals are pursued though a division of cognitive labor, with some physicians largely responsible for the innovation that advances the field and other physicians doing their best to apply established rules to cases. Pathologists attempting to follow rules theoretically work within the status quo, whereas a smaller group of pathologists works to revise the status quo. However, the division of labor in medicine is not nearly so clear-cut as that found in industry. Routine practice presents the pathologist with cases for which there is either no evidence-based literature to directly guide diagnostic behavior or, alternatively, the literature comes to conflicting conclusions. The result is that attempts to export [ldquo]one size fits all[rdquo] standards into pathology laboratories are substantially more difficult than standardizing the manufacture of a product. Even within industry, extreme standardization is recognized to create a workforce that cannot respond to exceptions.
At the same time that we face difficulties in standardization, we also face barriers to the generation, study, and acceptance of good new ideas and the retirement of bad ideas. The practicing pathologist who is not happy following rules is the ideal candidate to perform innovative studies structured according to standard scientific and ethical principles. 6 It is clear that there is no place for the pathologist who expresses individuality by subjecting unsuspecting patients to uncontrolled diagnostic self-expression. Meeting the public's expectations for innovation will require that the surgical pathology leadership remain open to contrary opinions, resisting the natural tendency of leaders to establish a tyranny of the status quo. Theories and classification systems are created to be replaced, but complaints about the resistance to new ideas go back at least as far as Virchow, who is credited with noting that good new ideas go through three phases. First, they are ignored, then they are attacked, and finally, after a prolonged struggle, they are met with we knew that all along. 27 What we knew all along about diagnostic disagreements is that they will continue to happen, that additional concrete steps should be taken to minimize them, and that it is time to move beyond documentation and denial into well-constructed studies of disagreement. If pathologists conclude that disagreement is like the weather, something to complain about but not fix, we will face externally devised fixes. A fundamental shift has occurred in public awareness and attitude toward error. Collecting incidence data is an important first step, but a leap from incidence to cure without a full understanding of the problem is unlikely to be successful. Likewise, we probably do not have time for the luxury of analysis paralysis. It remains possible that the cure for error and disagreement is as simple as sending large numbers of cases for a second opinion. However, this tactic may distract us from potentially more useful tasks such as developing a better understanding of the classification system failures and the pathologist individuality that contribute to disagreement and error.
1. Altman LK. Study to develop way to monitor cancer care . The New York Times National
May 21, 2000[colon]18.
2. Bulger RJ. The quest for therapeutic organization. JAMA 2000; 283[colon]2431[ndash]3.
3. Burke DS. Pasteur. Book review of Louis Pasteur
, by Patrice Debr[eacute]. JAMA 2000; 283[colon]2587[ndash]8.
4. Cabana MD, Rand CS, Powe NR, et al. Why don't physicians follow clinical practice guidelines? A framework for improvement. JAMA 1999; 282[colon]1458[ndash]65.
5. Carley WM. Pull up! United 747's near miss sparks a widespread review of pilot skills. The Wall Street Journal
March 19, 1999[colon]A1.
6. Emanuel EJ, Wendler D, Grady C. What makes clinical research ethical? JAMA 2000; 383[colon]2701[ndash]11.
7. Farmer ER, Gonin R, Hanna MP. Discordance in the histopathologic diagnosis of melanoma and melanocytic nevi between expert pathologists. Hum Pathol 1996; 27[colon]528[ndash]31.
8. Fitzgibbons PL. A typical lobular hyperplasia[colon] a study of pathologists' responses in the College of American Pathologists performance improvement program in surgical pathology. Arch Pathol 2000; 124[colon]463[ndash]4.
9. Greene J. Suing the system. American Medical News
May 8, 2000[colon]9[ndash]10.
10. Grilli R, Magrini N, Penna A, et al. Practice guidelines developed by medical specialties[colon] the need for critical appraisal. Lancet 2000; 355[colon]103[ndash]6.
11. Fukuyama F. A milestone in the conquest of nature. The Wall Street Journal
June 27, 2000[colon]A30.
12. Gostin LO. Public health law in a new century. Part III[colon] public health regulation[colon] a systematic evaluation. JAMA 2000; 283[colon]3118[ndash]22.
13. Gray C. Eccentricity and conformity. BMJ 2000; 319[colon]S2[ndash]7220.
14. Jamieson RL, Schubert R. Two die in UW medical school shooting. Seattle Post-Intelligencer
June 29, 2000. http[colon]//seattlep-i.nwsource.com
15. Kronz JD, Westra WH, Epstein JI. Mandatory second opinion surgical pathology at a large referral hospital. Cancer 1999; 86[colon]2426[ndash]35.
16. McAlister FA, Straus SE, Guyatt GH, et al. Users' guide to the medical literature. XX[colon] integrating research evidence with the care of the individual patient. JAMA 2000; 283[colon]2829[ndash]36.
17. Medline PB. Pathologists should retain their individuality! [lsqb]Letter[rsqb] Am J Surg Pathol 2000; 24[colon]1577.
18. Miller B. Autonomy. In[colon] Reich WT, ed. Encyclopedia of Bioethics
, vol 4. New York, NY[colon] Simon [amp] Schuster MacMillan, 1995[colon]215[ndash]20.
19. Mitka M. Statins help[mdash]if they're used. JAMA 2000; 283[colon]1813.
20. Mohr JC. American medical malpractice litigation in historical perspective. JAMA 2000; 283[colon]1731[ndash]7.
21. Oshinsky DM. Brat pack. Book review of How We Got Here. The 70's[colon] The Decade That Brought You Modern Life (For Better or For Worse)
, by David Frum. The New York Times
March 12, 2000[colon]13.
22. Page DL, DuPont WD, Jensen RA, et al. When and to what end do pathologists agree? J Natl Cancer Inst 1998; 90[colon]88[ndash]9.
23. Prager LO. Doctors often not practicing what guidelines preach. American Medical News
September 13, 1999[colon]8.
24. Reynolds T. Clinical trials[colon] finding balance in randomization. J Natl Cancer Inst 2000; 92[colon]370[ndash]2.
25. Schnitt SJ, Connolly, JL, Tavassoli FA, et al. Interobserver reproducibility in the diagnosis of ductal proliferative lesions using standardized criteria. Am J Surg Pathol 1992; 16[colon]1133[ndash]43.
26. Shaywitz DA, Ausiello DA. Back to the future[colon] medicine and our genes. The New York Times on the Web
April 16, 2000. http://www.nyt
27. Sommer A. Clinical research and the human condition[colon] moving from observation to practice. Nat Med 1997; 3[colon]1061[ndash]3.
28. Tarkan L. Value of second opinions is underscored in study of biopsies. The New York Times
April 4, 2000[colon]D7.
29. Veatch RM. Reasons physicians do not follow clinical guidelines. JAMA 2000; 283[colon]1685.
30. Weijer C. Ethics and Society. Book review of A Philosophical Disease[colon] Bioethics, Culture, and Identity
, by Carl Elliott. JAMA 2000; 283[colon]2452[ndash]3.