The 131st Annual Meeting (November 15-19, 2003) of APHA

Abstract #64798

Nancy L Sloan, DrPh¹, Suellen Miller, PhD, CNM², and Beverly Winikoff, MD, MPH¹. (1) International Programs Division, Population Council, One Dag Hammarskjold Plaza, New York, NY 10017, 2123390601, nsloan@popcouncil.org, (2) Department of Obstetrics, Gynecology, and Reproductive Sciences, University of Cali, Women’s Global Health Institute Initiative, 74 New Montgomery Street, Suite 600, UCSF Box 0886, San Francisco, CA 94105-344

There has been limited success in reducing maternal mortality in developing countries. Theoretically promising interventions to improve maternal survival are often implemented because evaluating the effect of obstetric interventions is especially difficult where accurate birth and death registration is uncommon. Measuring maternal mortality requires large samples and overcoming the logistic difficulties of data collection under challenging circumstances, particularly in rural areas. Assessing morbidity requires smaller samples, but may be more challenging than measuring mortality where diagnostic capacity is limited because verbal history information from individuals without medical training is fraught with error. Properly controlled experiments that demonstrate a temporal, adequately strong preventive association in multiple evaluations have been the foundation for dramatic achievements in child survival, yet have been discouraged to assess safe motherhood interventions.

Conducting well designed research to produce evidentiary data to reduce maternal mortality is challenging, but possible. Cluster randomization and quasi-experimental study designs can reduce the financial burdens and logistical difficulties of controlled trials while providing valid evidence. More easily measured mediating variables may be evaluated (but do not permit valid inference on mortality effect). Cross-sectional surveys with limited retrospective periods are an alternative to longitudinal household surveillance systems to assess change in mortality over time. Finally, limiting the complexity of research teams (avoiding “too many cooks”) and the amount of replication is cost-efficient and maintains human subject research ethics (particularly in assuring potential benefit). Unlike trial and error, causal investigation produces the knowledge for effective policy and programming to reduce maternal mortality.

Learning Objectives:

• Participants will be able to identify the criteria that constitute evidentiary data, recognize why lack of or incomparable comparison groups do not permit valid inference, understand why evidentiary data are necessary for policy and programming, and identify how common challenges of evidentiary research can be overcome.

Keywords: Epidemiology, International MCH

Presenting author's disclosure statement:
I do not have any significant financial interest/arrangement or affiliation with any organization/institution whose products or services are being discussed in this session.