Perspectives on the methods of a large systematic mapping of maternal health interventions.

BACKGROUND: Mapping studies describe a broad body of literature, and differ from classical systematic reviews, which assess more narrowly-defined questions and evaluate the quality of the studies included in the review. While the steps involved in mapping studies have been described previously, a detailed qualitative account of the methodology could inform the design of future mapping studies. OBJECTIVES: Describe the perspectives of a large research team on the methods used and collaborative experiences in a study that mapped the literature published on maternal health interventions in low- and middle-income countries (2292 full text articles included, after screening 35,048 titles and abstracts in duplicate). METHODS: Fifteen members of the mapping team, drawn from eight countries, provided their experiences and perspectives of the study in response to a list of questions and probes. The responses were collated and analysed thematically following a grounded theory approach. RESULTS: The objectives of the mapping evolved over time, posing difficulties in ensuring a uniform understanding of the purpose of the mapping among the team members. Ambiguity of some study variables and modifications in data extraction codes were the main threats to the quality of data extraction. The desire for obtaining detailed information on a few topics needed to be weighed against the benefits of collecting more superficial data on a wider range of topics. Team members acquired skills in systematic review methodology and software, and a broad knowledge of maternal health literature. Participation in analysis and dissemination was lower than during the screening of articles for eligibility and data coding. Though all respondents believed the workload involved was high, study outputs were viewed as novel and important contributions to evidence. Overall, most believed there was a favourable balance between the amount of work done and the project's outputs. CONCLUSIONS: A large mapping of literature is feasible with a committed team aiming to build their research capacity, and with a limited, simplified set of data extraction codes. In the team's view, the balance between the time spent on the review, and the outputs and skills acquired was favourable. Assessments of the value of a mapping need, however, to take into account the limitations inherent in such exercises, especially the exclusion of grey literature and of assessments of the quality of the studies identified.