Article Index

Most people learn of medical progress through the media. Whether through short clips on radio or TV or detailed articles in the press, scarcely a day passes without a report of a health development, and how it might affect you. Yet this news is often unhelpful. Exaggerated cures, contradictions, and plainly misleading information can do harm.

While health care policy is a public issue and freedom of the press mandates the free dissemination of health news, misinformation may be expensive, personally harmful, and detract from issues that are more important.

While public education has changed attitudes towards smoking, seat belts, and impaired driving, much reported information is confusing and counterproductive. One day, margarine is safer than butter, but later we learn that it too may harm arteries. Dietary fiber was once thought to prevent colon cancer, but now we are not so sure.

Some reports, but not others suggest saccharin, a boon to diabetics, causes bladder cancer. Such contradictions perplex those unfamiliar with the workings of science.  The problem is not the science, but how journalists report it, and how the public interprets it. This essay aims to help readers make sense of health news.

Understand the Scientific Method

One need not have a science background to understand how science progresses. While some medical science is self-evident (the setting of a fracture is one example), most medical facts are established through experiment. In clinical medicine, this means experiments with living persons. In a randomized clinical trial (RCT) of a new drug, individuals with a certain disease are randomly distributed into two groups. Those in one group receive the treatment to be tested, and the others receive a placebo. Neither the subjects nor the researchers know the membership of the groups. If the treatment group is more improved at the end of the trial, the difference is called the therapeutic gain, and if it is sufficiently large, the drug is deemed effective.

Read More About Randomized Clinical Trials

However, many things may go wrong with such an experiment. There may be insufficient numbers to make a conclusion; there may be younger patients in the treatment group; patients or doctors may inadvertently find out which patients are on the drug; or there may be a practical or statistical flaw in the trial design. A trial absolutely free of bias is probably impossible. Moreover, the patients selected for the trial may be unrepresentative of the sick population to be treated, side effects may nullify the treatment’s benefit, or similar trials may show contradictory results.

Short-term clinical trials of a treatment are difficult enough, but consider the experimental difficulties inherent in determining whether certain diets, environmental factors, or lifestyles cause or prevent disease. Such studies require the recruitment of dissimilar human beings for observation over years or decades. Usually there is no “treatment,” just an estimation of the degree of exposure of each subject to a diet, toxin, or other environmental factor. This is then related to a health outcome.

Such a study makes several doubtful assumptions: that other determinants of health such as genes, smoking, gender, and occupation are equally distributed among the subjects and will not bias the results; that the estimation of the exposure is accurate and sustained throughout the long study; that no pre-existing risk was overlooked; and that the population studied was such that the results can be generally applied. One report compared colon cancer in Copenhagen and Helsinki, and concluded that the higher incidence in the former was due to the insufficient intake of dietary fiber. This ignored environmental, genetic, and other differences in these two populations, but contemporary press reports encouraged the notion that fiber can prevent colon cancer.

Thousands of medical journals each publish hundreds of scientific papers annually. Only a few of these report a scientific advance, and many of the remainder are flawed, biased, or irrelevant. Very few studies are pivotal, and few valid conclusions rest on a single report. Scientific articles are working documents, subject to criticism, revision, contradictory studies and, for some, eventual confirmation. Each represents a piece of a puzzle which put together over years may reveal an important fact. Smoking as a cause for cancer was not a sudden revelation half a century ago, but rather the building of evidence over many years eventually to inform public health policy. One should interpret media reports accordingly.

Readers should be aware of obvious biases. A cold lasts 5 to 7 days, so any treatment given at the right time will seem to be effective. Those making specific lifestyle or diet changes may simply be the more health conscious among us who also eschew smoking, use seatbelts, and get regular exercise. Bias is almost impossible to eradicate in long-term population studies, so evidence from many sources is required before drawing conclusions.

There are few absolutes in Medicine. Rather, data are interpreted in terms of probability. Suppose a certain diet, if taken over 5 years reduces the chance of death during that period by 2 percent. Note that this is not a large figure – possibly an observation due to chance. Secondly, it suggests that after 5 years, two more of 100 individuals taking the diet will be alive than another 100 not on the diet. Thirdly, many taking the diet will die anyway, and many not on the diet will survive. If the diet is complex, hard to achieve, and expensive over a lifetime, you would want to be certain the data underlying the conclusion was valid before undertaking it. Finally, switching late in life to a diet others have followed for years seems unlikely to benefit you. Risk assessment is difficult and most diet and lifestyle data are works in progress.

Information derived from one population may not apply to another. Genetics, sex, environment, age, nationality, race, and many other factors may have diverse effects on the disease under study. Diet and blood cholesterol levels may have very different impacts upon smokers and diabetics than others.

Science progresses in increments, punctuated by setbacks and only the occasional discovery heralds improved diagnosis or treatment. The “truth” is realized eventually by consensus, based upon data from several sources. Such caveats seldom are declared in media reports, or indeed by the researchers themselves.

Join the Conversation

join conversation

Keep up-to-date on the latest news, stories, tips, research highlights, and more!

Sign up for eNewsletter

Connect through Facebook

Follow us on Twitter