RESUMO
This retrospective study assessed the utility of women's self-reports to identify obstetric complications in rural Ghana. All consenting obstetric and postpartum inpatients, presenting from the seventh month of gestation to 42 days postpartum, were interviewed at the Holy Family Hospital, Techiman and were asked about their signs and symptoms. A combination of clinical examination and laboratory testing of urine and blood samples was used for determining case status. Self-reported obstetric complications of 340 women were compared with the corresponding diagnostic status for their sensitivity, specificity, predictive value, and test-efficiency. Using algorithms that could not be practically applied at the community level, self-reported symptoms correctly identified the majority (75%) of complicated and uncomplicated pregnancies, but missed one-quarter of cases requiring emergency obstetric care. The positive predictive value of 50% indicates that women's self-reported symptoms should not be used in estimating the incidence of these conditions or in identifying women requiring referral in this population.