How am I, really? Perceptions of health and distress by women and their midwives

The depression outcome meta-analysis included 17 studies of 1,998 women.

“Seventeen studies of 1,998 women were included in the meta-analysis of depression outcome42–58 (Table 1) and 7 studies of 851 women included in the anxiety outcome42,43,45–47,58,59 (Table 2).”

Sample size calculations with power justifications were not reported by 13 included studies.

“Although sample size calculations with power justifications were not reported by 13 investigators42–47,50,51,53–56,58 for their various primary outcomes, this does not diminish the validity of the reported prevalence of included studies.”

Selection bias was possible in 7 studies that did not report participation rate and in 1 study with less than 50% participation.

“Selection bias was possible in 7 studies43,49,51,52,55–57 which did not report the participation rate of eligible subjects and one study44 which reported < 50% participation rate.”

Fifteen studies did not measure effect of exposure over time, limiting conclusions about changes in prevalence over hospitalization.

“Fifteen investigators44,46–59 did not measure effect of exposure over time, limiting the conclusions which can be draw regarding changes in prevalence of anxiety or depression over the duration of hospitalization.”

Loss to follow-up greater than 20% was reported in 5 studies that measured exposure over time, introducing possible systematic differences.

“Finally, of those investigators who did measure exposure over time, loss to follow-up was reported as > 20% in 5 studies44,47,50–52, introducing the possibility of systematic differences in drop-outs impacting interpretation of longitudinal results.”

Timing of survey administration relative to exposure was not reported in the majority of studies.

“Timing of survey administration relative to exposure was not reported in the majority of studies44–46,50–52,56–58.”