01/03/2011 05:11 pm ET Updated May 25, 2011

Why Some Medical Studies Can Be Harmful

Medical care should be based on firm science and treatment plans should be backed by rigorous studies. With something so important as our health in the balance, we should expect no less. Unfortunately, that is often not the case. Medical practice frequently is guided by inadequate studies in which doctors and patients place their trust. One important reason why many studies are deficient is they are measuring results, or endpoints, that have little to do with the health of real patients. Instead of testing for firm clinical endpoints, these studies look at substitute or replacement results called surrogate endpoints.

When a study is done, what we want to find out is whether a medical or surgical treatment actually benefits a patient, that is, does the patient live longer or better. Rather than real outcome, what is often determined, instead, is simply a lab value that may, or may not, have any clinical relevance.

Surrogate endpoints are used for a variety of reasons. Frequently surrogate results can be found relatively quickly, sometimes months or years before the more important clinical results. Often they are easier to measure, requiring only a simple blood test. Studies using surrogate endpoints are generally less expensive to carry out than the lengthier, but more reliable, studies that use clinical endpoints. The cost savings and convenience in using surrogate endpoints make them the commonly used technique for drug trials by pharmaceutical companies and surrogate endpoint studies are accepted as proof of drug efficacy by regulatory agencies like the FDA.

An example of a surrogate endpoint would be the measurement of CD4 counts in HIV. In HIV, CD4 counts decrease as the disease progresses. When a new medication is tested in HIV, the pharmaceutical company typically tests its effect on CD4 counts, not its effect on the clinical course of the disease. If the CD4 count rises, the medication is assumed to be of benefit in the treatment of HIV and the study can be used as evidence towards the medication's approval. What physicians really want to see, though, is clinical evidence that the medication prevents complications of HIV or prolongs life, not simply normalization of a lab value. However, for HIV the surrogate endpoint works. Typically improvement in CD4 counts is predictive of clinical improvement. Higher CD4 counts, fewer HIV complications. But the problem is that not all surrogate endpoints are as valid.

For example, diabetes is a disease characterized by impaired insulin utilization in the cells of the body. Diabetics frequently have problems with their eyes, kidneys and nervous system, but the most serious complication of diabetes is vascular disease. Seventy-five percent of diabetics die from heart attacks caused by disease in the coronary arteries. Any medication used to treat diabetes would be expected to lessen its complications, especially the vascular ones. With many FDA approved oral medications for diabetes, though, we find the opposite.

The impaired insulin use in the body causes some easily measured blood abnormalities in diabetics. Insulin is needed for the normal removal of sugar, or glucose, from the blood into the cells and diabetics typically have elevated blood sugars. The HBA1c is a measurement of the average blood sugar over several months and diabetics also have high HBA1cs. When studies are done to evaluate treatments for diabetes, the surrogate endpoint usually measured is either blood sugar or HBA1c. The assumption is that by lowering the blood sugar, the true clinical endpoints, especially the cardiac complications, would decrease. This assumption is unfortunately true only to a very minimal extent. Eye and kidney problems can be lessened slightly by tight blood sugar control, but the major life threatening complications of diabetes in the heart and blood vessels are unaffected by any improvement in these surrogate endpoints.

On the contrary, when very sick ICU diabetic patients were studied, something quite disturbing was discovered. Those patients whose blood sugar were controlled very tightly not only did no better than those diabetics who were allowed to have higher blood sugars but, alarmingly, had more cardiac death than those whose sugars were treated less tightly. This type of result was not an isolated finding confined to just the ICU. The ACCORD study, which involved more than 10,000 diabetics, showed similar findings. When HBA1cs were lowered through more intensive medical treatment, cardiac death rate increased. It seems that low blood sugars are not good for the heart.

These types of results have been known for over 40 years. The earliest oral medications for the treatment of diabetes, Orinase and Diabenase, were required by the FDA in the 1970s to include a warning about an increased risk for heart disease which accompanied their use. When Avandia similarly lowered blood sugar and HBA1c yet increased cardiac events by 30 percent, no one who has paid attention to diabetes treatments could be surprised. Lowering blood sugar and increasing heart complications often go hand-in-hand. As new (and expensive) medications come on the market for diabetes, drug companies in their promotions try to reassure concerned physicians that the medications do not significantly increase heart attacks. One might hope that they might actually decrease the number one killer of diabetics.

What, of course, is amazing is that such surrogate endpoints are still used for drug approval for diabetic medications. These new medications are prominently promoted on TV and in magazines. Patients are often overly focused on slight elevations of their blood sugar and HBA1c and clamor for the newer and, incorrectly assumed, better medications..

Now, blood sugar levels really do matter and lowering blood sugar, especially by diet and exercise with the addition of insulin when necessary, is the backbone of good diabetic control. But attempting to control HBA1c tightly by adding more and newer oral medications can be expensive and dangerous. Sometimes less care can be better care.