Estrogen for schizophrenia

Abstract

Background: In terms of clinical outcomes, women with schizophrenia seem to fare better then men, but appear more vulnerable to psychotic illness in the period after birth and menopause. As these vulnerable periods to psychosis are associated with estrogen withdrawal, this hormone has been proposed as a treatment for schizophrenia.

Objectives: To evaluate the clinical effects of estrogens alone or in combination with progesterone, as a sole treatment or as an adjunctive therapy, for the treatment of schizophrenia or schizophrenia-like illnesses.

Search strategy: Electronic searches of the Cochrane Schizophrenia Group's Register (October 2003) was supplemented with manual reference inspection of all identified studies. Authors of trials were contacted for further material and archive information.

Selection criteria: All randomised clinical trials comparing estrogens with or without progesterone, as a sole or adjunctive treatment for people with schizophrenia or other similar serious, non-affective psychotic illness.

Data collection and analysis: We evaluated data independently and analysed on an intention to treat basis. For binary data we calculated the fixed effect relative risk (RR) and its 95% confidence interval (CI). For continuous non-skewed data, we calculated weighted mean differences.

Main results: All available evidence relates to women. Four studies (n=108) compared estrogen only with placebo. Short-term scores for general mental state showed no significant difference between groups (n=24, 1 RCT, WMD PANSS for 100mcg comparison -2.26 CI -15.4 to 10.9). Data from all four studies showed overall loss from the studies was low ( 5%), with no significant differences between groups (n=96, 4 RCTs, RR 0.95 CI 0.2 to 6.1). Skewed continuous data from two studies showed no clear differences in ratings of movement disorders. One medium-term unpublished study (n=14) compared estrogen and progesterone with placebo. Data at six months showed no difference between groups for total scores (n=9, WMD PANSS -25.3 CI -51 to 0.1). For negative symptoms, results favoured the estrogen and progesterone group (n=9, WMD PANSS negative subscale -9.0 CI -17 to -0.9). For loss to follow up there was no difference between groups (n=10, RR 0.33 CI 0.02 to 6.7). This trial used many cognitive tests and one visual retention test showed statistically significant differences favouring the treatment group: total scores (n=8, WMD -3.5 CI -5.7 to -1.3).

Authors' conclusions: Adjunctive estrogen with or without progesterone does not appear to offer convincing advantages over placebo. Before any more research is undertaken in this area, all completed and unpublished work should be made available in order to ensure that more trials are justified.