Selections from current literature: the fight over fat: is pharmacological lipid lowering useful for coronary primary...

The optimistic bias favouring intervention in medicine has long been known, and it may be fair to say that compensating for this characteristic tendency of human judgement is a primary reason for the need for blind trials and evidence-based medicine. Packer has observed that "Physicians frequently decide to prescribe a drug because of the therapeutic gains it might provide (based on pathophysiological theories) rather than the benefits it actually delivers (as demonstrated by the results of controlled clinical trials)." In the present case, sorting out evidence from optimism, what we know from what we wish to be so, is a challenging task for the clinician trying to determine what to recommend to his/her patients in day-to-day practice. In the case of lipid-lowering therapies, the optimistic preference for positive results is widely evident. Ravsknov showed that trials finding a positive result have been cited six times as often as equally sound trials finding no effect. (The LRC alone was cited 612 times in the first 4 years after its publication.) The LRC trial results themselves (interpreted with appropriate statistical tests) were inconclusive and difficult to generalize, but were presented as definitive and generalizable with certainty. Subsequent papers have been more circumspect, but none the less have generally focused on CHD reduction (a positive result) while giving minimal discussion to overall mortality. In this vein, the finding of increase in mortality due to violence has been dismissed outright (e.g. p. 1243 of the Helsinki Study) though it has appeared in fibrate and bile acid sequestrant trials, there is a dose-response relationship, and the same effect is observed in non-human primates. Clinically, increased violence as a side effect may no longer be relevant, as recent analyses suggest that it is specific to fibrates (and to hormones, not used any longer); the currently favoured statin drugs do not so far appear to share this side effect, but it was dismissed by enthusiasts for lipid lowering long before such data existed. Estimates of the effect of lipid lowering are often inflated by including secondary prevention studies, as well as by assertions that 5-year NNTs underestimate the benefits of lipid lowering although analysis of primary prevention trials indicates that the full benefit of risk reduction is evident within 5 years. What is the family doctor to do? Clearly, not all the claims in the literature can be taken at face value, particularly when advanced by content-area experts invested in lipid research. Hence, it seems to fall to the family physician to translate these claims into honest expectations of benefit for the variety of patients we see, with their various levels of risk. Perhaps the best solution is to be found in combining the sound clinical epidemiology approach taken by Rembold with the honest extrapolation of effect according to baseline risk used by the CTF. We do now have the necessary estimates of NNT for primary prevention to inform our higher-risk patients: an NNT of 53 for CHD events certainly does justify our recommending statin drugs to middle-aged significantly hyperlipidaemic men, especially if they have multiple risk factors. The situation is less clear for lower-risk patients, such as women, the elderly and mildly dyslipidaemic men with no or few other risk factors. Even including the recent data on the statin drugs, the 5-year NNT for prevention of a CHD event in average-risk mildly hyperlipidaemic clinical populations (extrapolated using the CTF method) is 212. What intervention if any the family physician wishes to make, and the patient wishes to take, should be a matter negotiated between them, informed by the family physician's realistic appraisal of the patient's likely expected benefit.