To the Editor Recently, a large nested case-control cohort study1 found that using selective serotonin reuptake inhibitors (SSRIs) was associated with intracranial hemorrhage. Risk set sampling was used for matched case-control pairs and SSRIs were compared with tricyclic antidepressants (TCAs) to minimize the indication bias. However, there was an imbalance in the primary analysis between the case groups using SSRIs and TCAs and the control groups for SSRI and TCA in the usage rates of their respective antidepressant classes that were shown in Table 2 of the article.1 The difference in usage rates of their respective antidepressant classes were 7.4% between the case groups using SSRIs and TCAs (n = 588, 19.4% vs n = 364, 12.0%) and 4.4% between the control groups for SSRI and TCA (n = 14 341, 16.0% vs n = 10 381, 11.6%). While the authors stated that the cohort was “well-defined,” it is belied by the wide age range that was as young as 18 years with a mean (SD) age of 66.6 (16.6 years) shown in Table 1 of the article.1 A study showed that patients older than 40 years were more likely to be prescribed TCAs than SSRIs, and census data of age groups treated for depression showed patients aged 16 to 24 years received an SSRI over a TCA while the older groups up to age 85 years and older received a TCA over an SSRI.2,3 The nested analysis consisted of finding the frequency of SSRI use among individuals in the case and control groups matched by age, sex, and follow-up time and the frequency of TCA use among the matched case and control groups. This approach did not account for differences in ages between the case and control groups for SSRI vs TCA because the matching process for each antidepressant class was separate and independent. Therefore, the comparator group for TCA may have differed in age from the group for SSRI. This presupposes that the comparator group for TCA was older and therefore would have a higher incidence of intracranial hemorrhage, while rates of intracranial hemorrhage were actually lower (Table 2).1 However, these are crude rates that were not adjusted for any covariates, unlike the group for SSRI, because people who used TCAs served as the null reference group. People who used TCAs as a distinct group may have biased the results positively away from the null and led to a narrowly significant higher intracranial hemorrhage rate among the group for SSRI (adjusted relative risk, 1.17; 95% CI, 1.02-1.35; Table 2).1 Therefore, the choice of using people using TCAs as a comparator group in this study is questionable.
Beale DJ. Inappropriate Comparator Group in Study of Selective Serotonin Reuptake Inhibitors and Risk for Intracranial Hemorrhage. JAMA Neurol. Published online May 08, 2017. doi:10.1001/jamaneurol.2017.0538