Recent studies have shown that psychopathological rating by the Hamilton Depression Rating scale (HAMD) is well established and highly predictive for later response. in this study we aimed to find out if Clinical Global Impression (CGI) scale is suitable to guide antidepressive treatment under naturalistic conditions.

Inpatients with a major depressive disorder and treatment with citalopram were included and rated using in parallel the HAMD scale and the CGI scale weekly at baseline to day 35. According to CGI the sample has been divided in “CGI improver” (CGI = 2–4) and “CGI non-improver” (CGI = 5–6). Response was defined as HAMD sum score reduction by at least 50%.

55 patients were included. A HAMD score reduction of ≥24% on day 14 was highly predictive (p = 0.016) for both response and non-response on day 35. Among patients who improved on CGI scale at week two 33% became responder at week five vs 44% when improvement on HAMD scale was achieved. HAMD sum scores correlated well on the CGI scale (53–74%, p = 0.000). Patients who were “improved” according to CGI scale on day 14 exhibited less reduction on the HAMD scale on day 35 than patients who were “improved” according to HAMD (35% vs 46%, respectively). CGI rating on day 14 predicted response on day 35 with only 44% specificity (p = 0.255).

Our findings revealed that the CGI rating scale is not predictive for later response on day 35.