Junkfood Science: What timing

December 13, 2007

What timing

Update: Today, the FDA advisory panel rejected a bid by Merck to sell its statin, Mevacor over the counter. In a 10 to 2 vote, the FDA advisory panel recommended the FDA not approve Mevacor.

The evening news reported that the average total cholesterol level of U.S. adults is now in the ideal range, falling below 199 for the first time, according to new statistics released today from the Centers for Disease Control and Prevention. Dr. Timothy Johnson, medical correspondent for ABC World News, attributed the drop to the widespread use of statins which he called miracle drugs. Sales of statins in the U.S. alone totaled $22 billion last year, up from $13 billion in 2002, according to IMS Health, which monitors pharmaceutical sales.

There's a bit more to this story.

Tomorrow, the FDA’s Advisory Committee is coincidentally set to vote on whether to let Merck sell its off-patent statin, Mevacor, over-the-counter without a prescription. Merck has tried twice before and each time the FDA has shot down their applications. Last month, GlaxoSmithKline bought the exclusive U.S. marketing rights for Mevacor and there is a lot at stake in convincing the FDA and consumers about the safety and benefits of statins, bringing cautionary notes to advertising claims. Merck-GlaxoSmithKline anticipates selling the pills for $1-1.50 per pill, per day.

As the Chicago Tribune reported, Mevacor is generally considered less effective than newer cholesterol drugs so, rather than market to those at high risk for heart attacks, Merck is targeting the 20 million Americans who are at “moderate risk” and not currently taking statins. That’s $20 to $30 million a day in sales, $7.3 to $10.95 billion a year.

According to FDA News, the FDA has repeatedly rejected permitting the statin Mevacor to be sold over-the-counter (OTC) because of concerns that consumers could not adequately understand from package labels and drug company advertising, risk factors warranting statin use, contraindications, and how to monitor or evaluate the seriousness of side effects, such as liver toxicity or neuromuscular degeneration. The FDA committee was not convinced OTC would enable their safe and effective use. Past studies had found, for instance, that 25% of people who developed muscle pain, indicative of rhabdomyolysis, a common and potentially fatal side effect, were not able to make appropriate and safe decisions about discontinuing Mevacor.

The Associated Press and other news media made the connection between lower cholesterols and statins, leaving consumers to make the leap and assume that statins are the cause for lower risks for heart attacks by lowering cholesterol. But two questions went unanswered:

· Are the differences in total cholesterol levels over recent decades among all adults over age 20 significant?

· Does the correlation between lower cholesterol levels and rates for heart disease mean that lowering cholesterol is the cause?

The CDC Data Brief opened by stating that elevated serum total cholesterol is a risk factor for heart disease, the leading cause of death in the United States. When most people read that, they likely think that they just read that high cholesterol causes heart disease deaths. Did you catch the leap between a correlation (“risk factor”) and “cause” of death? Actually, the CDC Data Brief doesn’t say or demonstrate that at all.

The CDC simply reported that between the NHANES surveys in 1999-2000 and 2005-6, the mean serum total cholesterols of U.S. adults aged 20 years and older dropped from 204 mg/dL to 199 mg/dL. These are estimations of the mean levels population-wide across the entire country, based on surveys of varying demographic groups and sample sizes. But there is one big piece of information missing that makes these numbers meaningless alone: their standard errors.

These figures have standard errors (what some might call wiggle room or uncertainties) that give each estimated mean cholesterol number such wide ranges, they can overlap and make any real differences less notable. Standard errors are different from standard deviation, as was explained by statistics professors in a recent issue of the British Medical Journal. To greatly simplify, standard deviation is a calculation of the variability of a group by generalizing from a sampling taken from the same group. The mean of each sample group, however, will vary from sample to sample, which is called the "sampling distribution" of the mean. When statisticians then estimate how much sample means vary from the standard deviation of this sampling distribution, it’s called “the standard error of the estimate of the mean.” Standard errors depend on both the standard deviation in each sample and the sample size.

The standard error falls as the sample size increases, as the extent of chance variation is reduced — this idea underlies the sample size calculation for a controlled trial, for example. By contrast the standard deviation will not tend to change as we increase the size of our sample.

So, we can’t know if differences in mean cholesterol levels beween different surveys are meaningful if we don’t know the standard error. The CDC hasn't reported the standard errors for its 2005-6 figures, but they are available for the 1988 through 2004 figures. As an illustration, let’s look at men of various ages. According to that just-released Health United States 2007 report, CDC data shows that in 2004 the actual heart disease death rates for men aged 65-74 years was 0.54%; for men 75-84 years, it was 1.5%; and for men over age 85 it was 4.9%. Risks were more nominal for young men. Cholesterol levels also rise naturally with age. By lumping average total cholesterol levels for everyone over age 20 together, the overall average will be skewed downward.

CDC figures showed no change between 1988-94 and 2001-2004 in total cholesterol levels for men under age 35 (186 each survey). Among men 35-44 years of age, mean total cholesterol levels actually increased during those years, from 206 to 210. And among men 45-54, mean levels went from 216 (SE 1.8) to 213 (SE 2.5), but the standard error ranges overlapped.

Since statin prescriptions have most increased since 2002, what happened more recently to mean total cholesterol levels among men most at risk for dying from heart attacks? The CDC reports the NHANES 1999-2002 and 2001-2004 total cholesterols for men as:

Age 65-74:

202 (200.3-203.7) to

194 (192.3 -195.7)

Age 75+:

195 (192.2-197.8) to

194 (191.4-196.6)

Should we assume that all of these small changes in total cholesterol levels were the causes for lower death rates from heart attacks, or might other things be more likely to point to causative factors? After all, the CDC reported increases for rates of other risk factors for heart disease, such as hypertension by 18%, and we know that most of the most popular risk factors haven't been shown to predict heart attacks or premature deaths.

For instance, might total cholesterol be irrelevant and the fact that rates of smoking among men dropped by one-third between 1985 and 2005, going from 33% to 22%, be more clinically relevant? Or, just perhaps, might medical advances in the treatment for heart attacks over recent decades also have had something to do with it? And, to complicate things further, there’s that undeniable growing obesity paradox. All of these things, for example, make more sense than minor tweaks of a health indice within normal ranges.

As the Chicago Tribune reported, FDA advisory panels have worried that people who don’t need statins will take them under the influence of advertising and not recognize the sales pitches from the science.

Bookmark and Share