Truth in transition - can clinical guidelines carry the load?

A few decades ago, clinicians were led to believe that it was beneficial to prescribe encainide and flecainide to patients experiencing ventricular extrasystoles after a myocardial infarction. Estimates of harm vary widely but range from 50 000 excess deaths.(1,2) Similarly, prescribing corticosteroids to head-injured patients was reasoned to be safe and possibly beneficial until the large CRASH-1 trial proved harm, again with an estimated 10 000 unnecessary deaths associated with that belief.(3) Peri-operative beta-blockade is a third example, with widely ranging estimates of harm (from 10 000 annual iatrogenic deaths in the United Kingdom)(4) to a later review which failed to substantiate any increased mortality from this practice.(5) The first two examples - there are many others - date back some decades and highlight the importance of careful evidence accumulation and scrutiny prior to widespread implementation of new strategies. The erroneous thinking 'it may do some good, and it can't do much harm' is a lesson that clinicians continue to ignore. We need to trust evidence of both safety and efficacy before prescribing, and society trusts us to have made that assessment scrupulously. When we have not and people die, we are responsible. The third example highlights the potential clinical impact of guideline recommendations on healthcare outcomes and demonstrates that more than a decade after a major trial(6) enrolling more than 8 000 patients, it is still unclear what to recommend. The evidence space was muddied by trial fraud in the DECREASE trial stable(7) and by patient selection issues.