Sortie hospitalière précoce dans le post-partum

OBJECTIVE: To assess a policy of early discharge from hospital after a vaginal birth.

DATA SOURCES: Reports of studies on this topic were identified in Medline and Cochrane Collaboration database (from 1985 to the beginning of 1995, in english or latin publications). This search was supplemented by referenced studies in book chapters and in other published bibliographies. Key words were: early discharge or ambulatory care and mother or infant or post-partum or obstetrics.

STUDY SELECTION: One hundred and ten articles have been retrieved. We have included in the meta-analyse the randomized clinical trials comparing a group of women discharged early (less than 48-72 hours after delivery) with a control group leaving hospital after an usual stay (more than 48-72 hours). Quality of the studies was analysed through the Chalmers et al. recommendations.

DATA EXTRACTION: Data were retrieved independently by two authors, and results were compared. Selected outcomes were: hyperbilirubinemia, infant feeding problems and skin rashes; breast feeding; maternal satisfaction on the hospital stay; readmission and non-routine clinics for both the mother and her child.

DATA SYNTHESIS: Data from five included studies were pooled by the Peto method. There are no significant modifications of the risks of re-admission to hospital. The frequency of non-routine clinics and the risk of skin rash are not different in the two groups. The risk of hyperbilirubinemia and infant food problems are not different. The frequency of breastfeeding one month after delivery was higher in the early discharge group (OR = 1.88; 95% CI: 1.09-3.23), but this difference decreased over time. Women satisfaction on hospital stay was lower in case of early discharge (OR = 0.56; 95% IC: 0.44-0.72).

CONCLUSION: The meta-analysis suggests an advantage of early discharge policy on breastfeeding at one month after the delivery, but mothers seem to prefer longer hospital stays. These results must be cautiously interpreted as included studies are of small sample size or rather old. This policy remains to be evaluated more thoroughly before recommendations could be made.