Recent meta-analysis is misleadingBMJ 2010; 341 doi: https://doi.org/10.1136/bmj.c4699 (Published 27 August 2010) Cite this as: BMJ 2010;341:c4699
- 1Women’s and Gender Studies, Carleton University, Ottawa, ON, Canada
- 2Faculty of Epidemiology and Community Medicine, University of Ottawa, 451 Smyth Road, Ottawa, ON, Canada K1H 8M5
Delamothe queries why the authors of an American meta-analysis on planned home and hospital birth shifted focus from perinatal mortality to neonatal mortality “despite having relevant data for these calculations on only 9% of their total sample.”1 2
The authors found no difference in perinatal mortality between planned home and planned hospital births when they included the Dutch study, over 90% of their sample. Isolating the neonatal risk from the perinatal risk, they chose only studies that included both early (0-7 days) and late (8-28 days) neonatal mortality, conveniently excluding the Dutch study, which reported only early neonatal mortality.
In high resource countries two thirds to four fifths of neonatal deaths occur in the first seven days.3There is no reason to expect any difference in safety had the Dutch late neonatal mortality simply been reported or requested.
The high quality Dutch study was consequently displaced as the largest contributing study to the neonatal risk estimate by a study based on birth certificates. This American study does not meet the standards for home birth research that since the 1980s have required comparisons of home and hospital birth to stratify for whether the home births in the studies were planned and had a midwife or physician in attendance.4
Excluding the birth certificate study or including the Dutch study would have meant that the authors could not have concluded that less medical intervention or home birth creates higher neonatal risk. Rather, the more accurate conclusion of the meta-analysis would read, “planned home birth produces the same intrapartum and neonatal outcomes as planned hospital birth with far less intervention.” Not so savoury for the international media but fairer for birthing women.
Given its shortcomings, this meta-analysis should be withdrawn.
Cite this as: BMJ 2010;341:c4699
Competing interests: None declared.