Maternal and perinatal health research priorities beyond 2015: An international survey and prioritization exercise

Background: Maternal mortality has declined by nearly half since 1990, but over a quarter million women still die every year of causes related to pregnancy and childbirth. Maternal-health related targets are falling short of the 2015 Millennium Development Goals and a post-2015 Development Agenda is emerging. In connection with this, setting global research priorities for the next decade is now required. Methods. We adapted the methods of the Child Health and Nutrition Research Initiative (CHNRI) to identify and set global research priorities for maternal and perinatal health for the period 2015 to 2025. Priority research questions were received from various international stakeholders constituting a large reference group, and consolidated into a final list of research questions by a technical working group. Questions on this list were then scored by the reference working group according to five independent and equally weighted criteria. Normalized research priority scores (NRPS) were calculated, and research priority questions were ranked accordingly. Results: A list of 190 priority research questions for improving maternal and perinatal health was scored by 140 stakeholders. Most priority research questions (89%) were concerned with the evaluation of implementation and delivery of existing interventions, with research subthemes frequently concerned with training and/or awareness interventions (11%), and access to interventions and/or services (14%). Twenty-one questions (11%) involved the discovery of new interventions or technologies. Conclusions: Key research priorities in maternal and perinatal health were identified. The resulting ranked list of research questions provides a valuable resource for health research investors, researchers and other stakeholders. We are hopeful that this exercise will inform the post-2015 Development Agenda and assist donors, research-policy decision makers and researchers to invest in research that will ultimately make the most significant difference in the lives of mothers and babies. The CHNRI methods have been previously published together with detailed guidelines for implementation [5]. The goal of our priority setting exercise was to identify research questions with the potential to have an impact on maternal and perinatal health indicators between 2015 and 2025. In this context, maternal health’ relates to conditions affecting women during pregnancy, childbirth/abortion and up to six weeks postpartum/post-abortion, and ‘perinatal health’ relates to conditions affecting offspring from the time of fetal viability to the first 28 postnatal days. This process was managed by the WHO and implemented in three phases: (1) the generation and collection of research questions, (2) thematic analysis and consolidation of research questions, and (3) prioritization of research questions using a scoring system based on five criteria. Figure 1 illustrates this process. Study and analysis flow. Phase I was initiated by establishing a reference group of researchers, health care providers, program managers, and other stakeholders (including representatives of consumer groups and donors). An invitation was sent to a large number of active researchers in the field of maternal and perinatal health, identified through bibliographic metrics and other information available in the “BiomedExperts” database. This database includes over 400,000 registered members and 1.8 million pre-generated profiles of life science researchers (http://www.biomedexperts.com/). Potential participants were identified in the BiomedExperts database using a pre-specified search strategy available in Appendix 1. The identification of researchers was stratified to ensure participation of researchers from both developed and developing countries. In addition, invitations were also sent to program managers and policymakers identified in contact lists of WHO and partner organizations (e.g. the United States Agency for International Development (USAID) Maternal and Child Health Integrated Program (MCHIP)). Those who responded positively to the invitation became members of the reference group. All members of the reference group were invited to provide three research questions in seven domains: obstetric haemorrhage, hypertensive disorders of pregnancy (HDP), maternal sepsis, abortion, difficult/obstructed labour, preterm birth, and stillbirth. Maternal and perinatal health research questions identified through other processes (including a USAID priority setting exercise, published WHO guidelines, and a previous WHO CHNRI intrapartum priority setting exercise) were also included in the index list of research questions [12-16]. In Phase II, this long list of questions was independently assessed by two researchers (MW and SJP) for identification of duplicate questions. Questions that were out of scope (i.e. not pertaining to any of the previously mentioned domains), or that were too broad to be considered research questions (e.g. “research to reduce maternal mortality”, “develop and test interventions for reducing postpartum haemorrhage”), or that were considered epidemiological (non-intervention) research, were excluded. This process was reviewed by a third researcher (JPS), who resolved discrepancies. A reduced list of questions was then submitted to thematic analysis. The thematic analysis consisted of grouping similar questions together to identify research themes and sub-themes. This allowed us to identify additional duplicates and out-of-scope questions. Questions were edited for clarity and similar questions were merged. During this process, we aimed to achieve a certain level of detail compatible with the concept of “research avenues” (i.e. a research question that is not too broad, neither too specific, and could be answered through a set of individual research projects); hence, very detailed and specific questions were made more general. This process resulted in a refined list of questions for the technical consultation meeting held in Geneva in April 2013. The large majority of the participants in this technical consultation was selected from amongst the reference group and composed the technical working group. This technical working group consisted of a diverse group of 22 participants that included clinical specialists, researchers, program managers, WHO officers, donor and consumer representatives, and other stakeholders. During this technical consultation, the product of the thematic analysis was reviewed, new questions were developed where omissions/gaps were identified, and similar questions were further consolidated. Phase III consisted of scoring the final list of research questions. To reduce bias due to participant fatigue, we prepared six spreadsheets that differed in the order in which the research questions were presented. Each member of the reference group received one of these electronic spreadsheets via e-mail, accompanied by a score sheet consisting of five criteria to be used for scoring the questions. These five criteria included answerability, effectiveness, deliverability, maximum potential for disease burden reduction, and equity (Table 1). They are described in detail in the CHNRI guidelines [5]. The participants were instructed to score the questions one criterion at a time using a binary score system (1: Yes, 0: No). If they were not sure, did not know, or were not able to make a judgment, they were asked to leave the question blank. The completed spreadsheets, when returned by the participants, were integrated into a database. Scoring criteria for setting research priorities A research priority score was generated for each question by summing up the scores attributed to each criterion. No special weighting of criteria was applied. Thus, for each individual respondent, each research question could have a priority score ranging from 0 to 5. The overall research priority score was computed as the sum of all individual research priority scores. For each question, the overall research priority score was normalized (i.e. considering all questions, the overall research priority score for the question was subtracted by the minimum research priority score among all questions, and divided by the range: (x - min)/(max - min)). The normalized research priority score (NRPS) was analyzed and the cut-off point, enabling identification of the upper quartile (questions with the highest normalized research priority scores), was determined. Online Google® forms were used to capture online data from the reference group and Microsoft Excel (2010) spreadsheets were used to score and analyze the responses provided.