Chronic Medication Use And Weight Gain: A Real Or Perceived Risk?

​
​Jenny Sun, PhD

Before starting a new medication, patients and their providers must weigh their options: how do the benefits of treatment compare to the drug’s potential side effects? For chronic medications, patients will receive treatment for the foreseeable future, so the potential risk of medication-induced weight gain can play an important factor in treatment decision making. When evaluating the side effects of chronic medications, how well can we tease apart a real risk from a perceived risk?

​Prescription drug side effects are initially detected through spontaneous reporting, and this is when the perception of risk begins. However, the strength of evidence is low at this stage because a few reports of a medication-induced risk are not enough evidence to conclude that the side effect was caused by the drug. Large-scale studies must be conducted. While randomized controlled trials (RCTs) are often used to evaluate causality, there is rarely sufficient follow-up to observe the long-term metabolic side effects of sustained treatment with chronic medications. Evidence must therefore be generated using non-randomized studies, but the different choices in observational study design can sometimes paint a messy picture for patients who must make treatment decisions.

One example is oral contraceptives and weight gain. The perception of risk began decades ago, and weight gain is a common reason for discontinuation of oral contraceptives. Dozens of studies have explored this potential association, but there is little evidence from well-conducted studies to support the risk of weight gain. While some studies reported an increased risk of weight gain, many of these studies had no control group (patients may gain weight regardless of oral contraceptive use, so a comparison group is needed) or had no actual weight measurements (relying on perceived weight gain overestimated the risk). Furthermore, confounding was considered in only 8 of 17 non-randomized studies of progesterone-only contraceptives and weight (adjusting for confounding reduces the possibility that any observed weight gain might have been caused by other, unobserved factors).

Another instance where differences in observational study design have influenced the perception of risk is for antidepressants and type 2 diabetes. While antidepressants are associated with a modest risk of weight gain, evidence regarding the risk of type 2 diabetes is conflicting. Several observational studies found that antidepressant users had a substantially higher risk of type 2 diabetes compared to untreated patients. However, these studies overlooked the ways that treated patients differ from untreated patients. First, antidepressant users have more contact with the healthcare system compared to untreated patients, which may explain the higher rate of recorded diagnoses for type 2 diabetes (detection bias). Second, patients with more severe depression are at a higher risk of developing type 2 diabetes compared to untreated patients, so studies must account for the differences in depression severity at the time of treatment initiation (confounding by indication). More recently, studies with careful attention to these potential biases reported no association between antidepressants and type 2 diabetes risk, or that the risk is lower than initially reported.

​As in the cases of oral contraceptives and antidepressants, poorly controlled studies have played a role in creating the perception of weight-related risk for some prescription drugs. This is not to say that observational studies of medications are problematic or that all reports of medication-induced weight gain are incorrect. Rather, this is a reminder to researchers about the importance of robust research methodology, especially when observational studies play such an essential role in evaluating drug safety when RCTs are not available. Unfortunately, for the patients who must make treatment decisions, correcting a flawed study can come too late. Even the perception of risk can be enough to influence treatment decision making.