In a rhetorical response to Wald and Law, a group of researchers proposed the “Polymeal.”6
Based largely on the Mediterranean diet, the Polymeal incorporates healthy portions of dark chocolate, omega-3-rich fish, olive oil, garlic, red wine, and nuts. Many authorities feel that universal adoption of the Polymeal would trump the benefits of the Polypill at less expense and without the side effects.
Number needed to treat
To better appreciate the statistical underpinnings of predisease designations, one must touch on the concepts of relative vs. absolute risk and number needed to treat (NNT). For example, a drug may substantially reduce the risk of a clinical end point (disease or death). “Drug X reduces the risk of [pick a disease] by 50%” may be the claim, backed by clinical research data. But if the patient you are treating has a low risk of contracting the disease to begin with, the change in absolute risk may be negligible (2% vs. 3%, for example).
NNT introduces the consideration of how many patients would need to be treated to achieve prevention of a single clinical end point (a disease event or death). The lower the NNT, the more beneficial and cost-effective the intervention. The higher the NNT, the greater the number of patients who would need to be subjected to the risk and expense of taking a medication over a specified period of time to achieve prevention in just a single patient. When NNTs extend into the hundreds or even thousands, the justification for intervention becomes problematic.
A recent randomized trial among patients underscored the challenges of decision making for medical consumers. Regardless of how information about risk/benefit was conveyed — even in this relatively well-educated study population (38% had at least a college degree) — only about one third of participants fully comprehended the implications of data.7 The challenge, the authors assert, is to make information available to the public in a form that is understandable, and moreover, actionable.
To gain a better understanding of the controversies surrounding predisease, consider four prototypical preconditions: prehypertension, hyperlipidemia, prediabetes and osteopenia.
Hypertension is arguably the first condition for which treatment was proposed for people not suffering from an actual disease. A test parameter—the height achieved by a column of mercury in a sphygmomanometer—was incontrovertibly linked to risk of stroke, heart disease, and kidney failure in otherwise well individuals. The progress made in medicine is laudable when one considers that President Franklin Delano Roosevelt succumbed to the then little-recognized and now largely avoidable effects of poorly controlled hypertension a mere 67 years ago.
The benefits of intervening early to treat frank hypertension are incontrovertible. Also evident, however, is the fact that people with milder abnormalities stand to benefit less from treatment than those with severe abnormalities. Aggressive pharmacologic treatment of borderline increases in BP in patients at high-risk of MI, valvular disease, congestive heart failure, stroke, or renal disease might be justified, but indiscriminate drug treatment of prehypertensive patients without concomitant risk factors is not supported by the evidence and might invite unnecessary side effects, particularly in the fragile elderly.
In 2003, the National Heart, Lung, and Blood Institute (NHLBI) revised its BP guidelines. The new guidelines state that systolic BP (SBP) levels between 120 and 139 mm Hg or diastolic BP levels (DBP) of 80 to 89 mm Hg indicate prehypertension. By this expanded definition, the prevalence of prehypertension in healthy US adults is 36.3%. This revision means that 100 million Americans whose BP had previously been considered normal now have cause to watch their BP more closely.8
A recent meta-analysis confirms that prehypertension appears to be associated with a greater risk of stroke. In a pooled analysis of 12 studies, an SBP of 120 to 139 mm Hg or a DBP of 80 to 89 mm Hg at baseline was associated with a 55% greater risk of incident stroke. The effect was mostly driven by an increased risk in patients whose BP was in the higher range of prehypertension—an SBP of 130 to 139 mm Hg or a DBP of 85 to 89 mm Hg. But the relationship did not reach statistical significance for the lower range of prehypertension (SBP, 120-129 mm Hg; DBP, 80-84 mm Hg).9
The authors of the meta-analysis note there is no clinical trial evidence that lifestyle changes will prevent stroke in prehypertensive individuals. On the other hand, there is no evidence that drug treatment to lower BP will prevent strokes in this population either. “It really makes us wonder whether we should be prescribing medications for those individuals,” says Amytis Towfighi, MD. “However, to know for sure, we’d have to do a study to see if prescribing medications will actually lower stroke risk for [them].”10
Correspondingly, NHLBI public pronouncements emphasize nonpharmacologic treatment with diet and exercise. Current guidelines say that prehypertension is best treated with:
- Weight loss
- Restriction to ≤1,500 mg sodium per day
- A diet rich in fruits, vegetables, and low-fat dairy products and reduced consumption of saturated and total fat
- Moderate use of alcohol (no more than two drinks a day for men or one drink a day for women. A drink is 12 oz of beer, 4 oz of wine, 1.5 oz of 80-proof spirits, or 1 oz of 100-proof spirits.)
Indeed, the potential efficacy of this approach in prehypertension is suggested by the robust results of the Dietary Approaches to Stop Hypertension (DASH) study but remains an extrapolation from research on frank hypertension. The DASH diet reduced BP by 5.5/3.0 mm Hg. That reduction in BP was estimated to reduce coronary heart disease (CHD) by 15% and stroke by 27%.11