Practice guidelines – systematically developed statements to assist practitioner and patient decisions about appropriate health care for specific clinical circumstances [2].

Guidelines and their siblings (protocols and algorithms) are not new [3]. However, their popularity has burgeoned, especially when compared with their more restrictive first cousins (standards, prediction rules, and decision rules) (Table 1). They have been hailed as the salvation from the wide variations in medical care and simultaneously lamented as harbingers of the ruination of medicine as we know it [4]. A friend expressed concern about the potential for lawsuits now that his subspecialty organization is developing guidelines. In Maine, steps have been taken to protect guideline users against liability [5]. One wonders how such a seemingly harmless term became invested with such great and terrible expectations!

In this issue, Tunis and colleagues [6] discuss how American College of Physicians (ACP) members view guidelines whose development the College has fostered through the Clinical Efficacy Assessment Project [7-9]. The disconcerting finding that 7% of respondents indicated familiarity with fictitious guidelines must be considered when assessing the 11% to 59% who reported familiarity with actual guidelines. Although the lack of familiarity with the guidelines recently issued by the federal Agency for Health Care Policy and Research, which was authorized to promote guideline development and dissemination [3], is understandable, the 11% Figure was for ACP guidelines issued 2 years before. This finding confirms other reports about the limited impact of guidelines [10, 11]; for example, when pressed to recount the content of guidelines with which physicians said they were familiar, few could accurately do so [10].

The study by Tunis and coworkers [6] is particularly intriguing with respect to how physicians accorded credibility. Although most were unfamiliar with the specific guidelines, they clearly had an opinion about how much they trusted them. Their reactions related primarily to the issuing institution. College members (82%) trusted ACP guidelines. Subspecialty members trusted their organization's guidelines. Only 6% trusted Blue Cross guidelines, probably reflecting concern about financial motives and the degree of expertise used in their development. Ironically, Blue Cross supported the original ACP project on the use and interpretation of common diagnostic tests [4, 12]. Thus, to some extent, this illustrates the "halo effect" whereby one's assessment of a person or institution carries over from one judgment to another.

Finally, although 65% of respondents thought that guidelines could improve quality of care, a similar percentage (68%) said they thought guidelines would be used to discipline physicians. The oft-heard reference to guidelines as "cookbook medicine" [5, 13] was made by 25%. Robert Brook's apt response to this peremptory dismissal was that great cooks like Julia Child write cookbooks [14]. Many good cooks have learned from Child's structured approach but are free to alter the recipes or use others. In the end, "the proof of the pudding should be in the eating"—ergo, the recent emphasis on specifying and validly measuring outcomes.

Why do guidelines scare so many doctors? Probably because the stakes have escalated with the incorporation of guideline development into federal legislation in 1989 [3] and the President's health care reform proposal [5]. Those who are fearful focus on the hazards of navigating between the Charybdis of proliferating guidelines and the Scylla of third-party payers and lawyers ready to use the rules against them [4, 15]. Unfortunately, fear is self-defeating and dire pronouncements serve merely to elevate guidelines to a higher status than they deserve. Rather, we should sail on, carefully and confidently.

Although guidelines vary in their validity and complexity [16], they represent a reasonable desire to assure quality care. Explicit strategies for caregivers have a long tradition in programs involving physician assistants and nurse practitioners. The assumption has been that doctors did not need them because they knew what they had to do based on expert opinion gleaned from colleagues or authoritative textbooks and manuals. In 1971, when establishing a walk-in clinic [17] that relied on nurse practitioners, a house officer, and a supervising attending physician, I sought guidance from physicians engaged in problem-oriented practices [18, 19]. The result was a decision to develop protocols for nurses and physicians [20] to deal with common problems. The intent was to guarantee some modicum of consistency from patient to patient. In effect, what we were asking was, "What would the most knowledgeable doctor do in the situations we commonly face?" We used the protocols to orient new staff members and as guides to audit each encounter and did the same in our sexually transmitted disease clinic [21]. Although they removed some element of chance, our protocols did not guarantee good care. Indeed, if the caregiver defined the patient's problem incorrectly, applying the protocol would be meaningless. Furthermore, we recognized that because patients differ, individualization was essential, albeit within an acceptable framework. Undoubtedly, most good practices operate in the same manner.

Developing protocols or guidelines impresses one with how little is known about managing common conditions and how technology often takes on a life of its own with little supporting evidence of effectiveness. For example, Kent and coworkers [22] identified 3125 interim citations in peer-reviewed medical journals during a recent update of 1987 ACP guidelines on magnetic resonance imaging for neurologic conditions. After screening out reviews, technical reports, case reports, and studies with fewer than 30 original case-patients, they were left with 285 papers. On rating the quality of the evidence, 1 article was rated A; 15 were rated B, 16 C, and 118 D. The remainder were below D. This is an extraordinary finding. Here we have an expensive, very widely used procedure and only 5% of recent reports are judged sufficient to determine its diagnostic and therapeutic efficacy, and many of those are hardly adequate.

As Woolf [23] has clearly outlined, guidelines must be based on sound evidence, supplemented where necessary by truly expert opinion. Except where supporting data are incontrovertible, guidelines should be regarded as works in progress. They are most useful for initiating discussion with those physicians whose practices do not systematically incorporate important advances and for establishing a research agenda to help physicians deal with the situations they face daily. Compliance is best when guidelines are accompanied by incentives or reminders. A recent study of guidelines for managing low-risk patients with chest pain [24] showed that when the reminders stopped, implementation of the guidelines stopped. Memory is fallible, and the more we can do to assure patients of the consistent application of knowledge at the highest levels, the better. Coupled with the revolutionary changes in computers, sound evidence-based guidelines can be potent tools to assure high-quality care.

College members seem to be greeting the guidelines era with ambivalence. In the ideal world, the issue would not be whose guidelines but their content. The ideal world is not a game of "gotcha" for developers and users. Still, in a world where trust is in short supply, it is comforting that member surveys reveal a high level of confidence in the College and Annals. This suggests that ACP should continue to develop and evaluate guidelines, especially contentious ones, and that Annals should remain a resource for delineating their supporting evidence.

In short, developing guidelines represents common sense. However, as Voltaire noted, "People sometimes say: ‘Common sense is very rare’ ... . A man, who judges very sensibly in one matter, will always grossly deceive himself in another" [25]. Similarly, common wisdom is not always very wise. The real challenge then is to construct guidelines carefully and to revise them continually by collecting the appropriate data. Unless we do so, we will be operating as in the age of Galenic dogma. At the same time, we need to keep them in perspective; they are neither saviors nor villains, just imperfect attempts to care for patients better. Their development can act as beacons into the still vast scientific unknown.