STAR*D sequenced antidepressant treatment study

Before STAR*D, the evidence base for treating major depression beyond a single drug trial was thin. Most antidepressant registration trials were short, enrolled carefully selected patients, and excluded anyone with significant comorbidity. Real-world patients frequently had partial responses or treatment failures, but clinicians lacked data from representative populations on what to try next or how likely a second or third agent was to succeed. That gap was what the STAR*D investigators set out to fill.

STAR*D enrolled 4,041 adults, 3,671 of whom received the first treatment step, with nonpsychotic major depression from 41 primary care and psychiatric centers, deliberately using minimal exclusion criteria to reflect actual clinical populations. A. John Rush designed the sequential protocol, and Madhukar Trivedi led the implementation of measurement-based care, in which clinicians adjusted treatment according to standardized symptom scales at each visit instead of relying on clinical impression alone. Trivedi was also first author of the report on second-step augmentation with bupropion or buspirone, and Maurizio Fava led the report on the third-step switch to mirtazapine or nortriptyline. Patients who did not achieve remission on one treatment could proceed to a next step, which offered either a medication switch or augmentation.

Remission rates fell steeply with each step. At step one, citalopram, roughly 37% of patients achieved remission by QIDS criteria. At step two, approximately 31% of those who had failed step one remitted. Steps three and four each produced remission in around 13 to 14% of the patients who reached them. Cumulative remission approached 67%, meaning that with sufficient persistence, about two-thirds of patients could eventually achieve remission. The critical caveat was durability: patients who required three or four steps to remit had substantially higher relapse rates during follow-up, diminishing the practical benefit of late remission.

The summary report appeared in the American Journal of Psychiatry in November 2006. Most depressed patients did not remit on their first antidepressant. Each failed trial reduces the odds of eventual remission and raises the odds of relapse if remission is eventually achieved. Patients who entered follow-up in remission relapsed less often than those who entered it with only a partial response. The trial had no comparison arm without measurement-based care, so it could not show that the method itself improved outcomes.

STAR*D's methodology attracted criticism as well. The lack of a placebo arm meant that it was impossible to determine how much improvement was drug-specific. H. Edmund Pigott and colleagues argued that the summary had used a non-blinded measure instead of the protocol's blinded Hamilton scale and had counted patients who already scored as remitted when they entered a step. Their 2023 reanalysis of the patient-level data, following the protocol, found a cumulative remission rate of 35.0%, about half the 67% originally reported.