The Women's Health Initiative Study: A Firebell in the Night?

Antidepressant medications are among the most commonly prescribed. Are these drugs medically riskier than we think? A new study published in the Archives of Internal Medicine found that women who reported taking an antidepressant drug had increased risk for stroke and increased risk for death relative to women not taking antidepressants.The Women's Health Initiative (WHI) of the National Institutes of Health followed postmenopausal U.S. women for up to 15 years. The current report focused on 136,000 women who were not taking antidepressant drugs when they entered the study. The critical comparison was between the subsequent health history of 5,550 women who started to take antidepressants after study entry (1 or 3 years in) and the rest of the women who did not start taking antidepressants. The study did not find any relationship between antidepressant use and heart disease. However, antidepressants were associated with increased risk of death from any cause and of hemorrhagic stroke among those taking SSRIs.

Here are five reasons why we should pay attention:

(1) The large sample size in and of itself makes this study highly credible. The larger the sample, the more statistical power you have to test your hypothesis in a conclusive way. This study was a more powerful test than previous studies, which used smaller samples. Many health events, such as stroke death, will not be common in an otherwise healthy sample over a short follow up, so large samples and long follow up periods are critical if you want to be able to detect an effect. For example, in this sample, there were 2357 strokes, 445 of which were fatal, over the follow up period. This pool of events is sufficiently large to enable good tests of whether antidepressants were associated with elevated risk for stroke or stroke death.

(2) Health risks were observed across different categories of antidepressants and were not confined to the older tricyclic antidepressants, One reason newer selective serotonin reuptake inhibitor (SSRI) antidepressants (such as Prozac) are prescribed so widely is that they are viewed to be medically benign. However, when SSRIs were examined separately, they also had increased risk for stroke and death.

(3) The size of the effects, though not whopping, are certainly attention-getting. For example, SSRIs were associated with double the risk for fatal hemorrhagic stroke and one-third higher risk of mortality from all causes.

(4) The authors appear to have no axe to grind against antidepressants. This was a multi-site study funded by the National Institute of Health. In fact, I find it interesting that the study authors, in the accompanying press material, are somewhat muted in their alarms, "Depression is a serious illness with its own health risks, and we know that antidepressants can be life-saving for some patients. No one should stop taking their prescribed medication based on this one study, but women who have concerns should discuss them with their physicians," said lead author Jordan W. Smoller, MD, ScD.

(5) An accompanying commentary found little to fault with the study. When a particularly impactful or potentially controversial study is published in the health sciences, it is not uncommon for the journal to publish a critical commentary alongside the original report. In this case, the comment agreed that this was a potentially important study. The main caution in the comment was that we cannot be absolutely certain whether observed health risks were truly attributable to the drugs and not to the depression, or to some other unseen characteristic that goes along with starting antidepressant treatment (i.e., women who take antidepressants take worse care of themselves). Smoller and colleagues did the best they could to control for these other factors statistically. While the caution stands, we have to realize that these may be the best data we will ever have --a study that randomly assigned women to antidepressants and controlled the treatment over 15 years would be impossible to perform, for both ethical and logistical reasons.

Let me end with the clinical language of the commentary, "The findings, in the largest cohort of women yet studied, provide additional warning that antidepressant therapy may in fact be detrimental with respect to stroke and total mortality in this demographic population."

Time will tell whether this study is a firebell in the night.