SSRIs: An Update

An insightful update to Scott's earlier SSRI analysis. (1) Cipriani et al's meta-analysis confirms the antidepressant effect size is ~0.3. (2) That anomalously low number isn't SSRI-specific — it's a general feature of psychiatric (and many other) drugs: melatonin, even benzodiazepines, even morphine (~0.4) all underperform their subjective effects in studies, so either 'nothing works' or we should rethink effect sizes. (3) The key reconciliation of studies (SSRIs barely beat placebo) with clinical experience (psychiatrists/patients find them lifesaving): responder heterogeneity — if ~1/3 of patients respond strongly (effect size ~1.0) and 2/3 not at all, the trial average is ~0.33, while in clinic the 1/3 obvious responders get kept on it and the rest get switched. So 'SSRIs work' is an existence claim, not a universal one. This supports a throughput / cheap-experiment model (cf. Recommendations vs Guidelines, Anxiety Sampler Kits): what matters is whether a drug is a cheap experiment to run per-patient, not just its average effect size. (4) SSRIs are equally anti-anxiety drugs (maybe a larger effect there), which may also explain clinicians' positive glow.

Quality 72 · Strong. Strong, high (72). A clear, clinically-grounded update with a genuinely clarifying resolution (responder heterogeneity reconciles the study/clinic gap) and a durable idea (the throughput / cheap-experiment model). Upper-Strong; an 'update' building on prior work and explicitly tentative ('I don't know if this is consistent with the studies').

Claude’s paradigm shift 55 · Moderate. Moderate, high (55). The responder-heterogeneity reconciliation and the throughput model are fresh syntheses (he credits Gueorguieva & Krystal for a sophisticated version), building on the existing effect-size debate.

Real-world impact 3 · Moderate. Moderate (3). A clinically-relevant reframe in the consequential SSRI-efficacy debate; influential within the psychiatry-adjacent sphere.