Do No Harm

Recent years have seen a number of policies and recommendations from a variety of health care organizations with the stated intent of making delivery of clinical perioperative care better or safer. In some cases, implementation of these policies has been encouraged with financial rewards for compliance and penalties for noncompliance. Unfortunately, many of these “evidence-based” attempts at improving clinical care have, in fact, worsened care delivered to our patients. By implementing these policies, patients have been harmed, and the quality of health care diminished. The following summarizes 3 failed attempts to make things better: 1) Starting in 2000, the Joint Commission on the Accreditation of Healthcare Organizations decided that pain was not adequately treated in postoperative patients and that this failure was a violation of patients' rights.1,2 Based on this conclusion and recommendation, most hospitals in the United States embarked on programs to more aggressively treat postoperative pain. As an unfortunate, but entirely predictable,3 consequence of this emphasis on pain control, patients are suffering the effects of excessive sedation, including hypoxic brain damage and death. The frequency of such events has increased to the point that the Anesthesia Patient Safety Foundation has identified the excessive use of opioids as a major cause of postoperative morbidity and mortality.4 2) Approximately 2 years ago, a variety of expert panels determined that an inadequate number of patients were receiving preoperative beta blockade and that the quality of care would improve if more patients were given beta blockers before surgery. Despite the fact that data in support of such a conclusion and recommendation were equivocal at that time and published manuscripts5 advised caution in implementing protocols in anticipation of the findings of a large ongoing study (the POISE trial), many hospitals and anesthesiology departments felt compelled to write “Beta Blocker Protocols” and expand the indications for using this class of drug in their preoperative clinics. The results of the POISE trial published in May 20086 concluded that, while perioperative beta blockers did have a small beneficial effect on the rate of non-fatal myocardial infarctions, patients randomized to receive beta blockers were twice as likely to have a major stroke and 30% more likely to die than patients receiving a placebo. 3) Within the last year, there has been increased concern about the detrimental effects of perioperative hyperglycemia. Again, despite the lack of good data, institutions and departments were compelled to write policies advocating “tight” glucose control. Since then, 2 large European studies of aggressive treatment of hyperglycemia in intensive care settings (the VISEP and GluControl trials) were terminated by their safety monitoring boards due to excessive morbidity and a lack of beneficial effect for patients randomized to the “tight” control arms of these studies.7 In the United States, enrollment into the intensive treatment arm of a large National Institutes of Health-funded study of type 2 diabetics (the ACCORD trial)8 was terminated because of increased risk of death compared to the standard treatment strategy. I have several suggestions for physicians evaluating clinical practice advisories issued by outside agencies: 1) For recommendations based on “evidence-based medicine,” be aware that publication bias may exist. Studies showing an effect are more likely to be published than those that don't. There is no telling how many studies failed to get published or were terminated because it became apparent to the investigators that there wasn't going to be a “difference”. However, now that many journals insist that all studies seeking approval from an Institutional Review Board must be registered, it may be possible to include results from both “positive” as well as “negative” studies thus minimizing the likelihood of publication bias. 2) If the benefit of an intervention has been under investigation for a long period of time, it is likely that the benefit of that intervention is small, if it exists at all. The use of beta blockers for the prevention of perioperative myocardial infarcts has been studied for decades. Even before publication of the POISE trial, this should have suggested that if beta blockers work, they don't work very well. 3) Be wary of studies that only look at 1 or 2 outcomes and fail to examine likely complications associated with the proposed therapy. Yes, beta blockers may reduce the incidence of non-fatal myocardial infarctions, but they increase the incidence of bradycardia, hypotension, strokes, and death. Yes, strict glycemic control may reduce surgical site infections, but it increases the incidence of profound hypoglycemia. And what will be the costs (increased nurse workload, laboratory testing, protocol development and assessment, staff education etc.) associated with implementing the proposed changes in the safest possible manner? 4) Don't assume that what worked at other institutions will work at your hospital. If the decision is made to implement a practice advisory, insist that careful monitoring be implemented to ensure that the benefits are being reaped and no unintended morbidity is being generated. Clinical practice guidelines can be helpful to improve patient care, but we should be cautious. The “Law of Unintended Consequences” tells us that the hubristic attempts to regulate and control processes often have negative outcomes. What should we do? We should use good common sense. We should not rush to embrace each new edict as it is delivered to us. And above all, we should strive to do no harm.