Gendered norms of responsibility: Reflections on accountability politics in maternal health care in Malawi

Background: This paper aims to provide insights into the role of traditional authorities in two maternal health programmes in Northern Malawi. Among strategies to improve maternal health, these authorities issue by-laws: local rules to increase the uptake of antenatal and delivery care. The study uses a framework of gendered institutions to critically assess the by-law content, process and effects and to understand how responsibilities and accountabilities are constructed, negotiated and reversed. Methods: Findings are based on a qualitative study in five health centre catchment areas in Northern Malawi. Data were collected using meeting observations and document search, 36 semi-structured individual interviews and 19 focus group discussions with female maternal health service users, male community members, health workers, traditional leaders, local officials and health committee members. A gender and power sensitive thematic analysis was performed focusing on the formulation, interpretation and implementation process of the by-laws as well as its effects on women and men. Results: In the study district, traditional leaders introduced three by-laws that oblige pregnant women to attend antenatal care; bring their husbands along and; and to give birth in a health centre. If women fail to comply with these rules, they risk being fined or denied access to maternal health services. The findings show that responsibilities and accountabilities are negotiated and that by-laws are not uniformly applied. Whereas local officials support the by-laws, lower level health cadres’ and some community members contest them, in particular, the principles of individual responsibility and universality. Conclusions: The study adds new evidence on the understudied phenomenon of by-laws. From a gender perspective, the by-laws are problematic as they individualise the responsibility for maternal health care and discriminate against women in the definition and application of sanctions. Through the by-laws, supported by national policies and international institutions, women bear the full responsibility for failures in maternal health care, suggesting a form of ‘reversed accountability’ of women towards global maternal health goals. This can negatively impact on women’s reproductive health rights and obstruct ambitions to achieve gender inequality and health equity. Contextualised gender and power analysis in health policymaking and programming as well as in accountability reforms could help to identify these challenges and potential unintended effects. According to the definition of accountability by Brinkerhoff and Wetterberg [21], by-laws can be considered an accountability tool whereby a pregnant woman or her partner have the obligation to provide information about, and/or justification for her/his actions in response to the chief who has the power to make those demands and apply sanctions for non-compliance. This form of accountability, however, reverses roles as common understandings of accountability are about the obligation of agents (powerholders), rather than subjects (subordinates or groups of people that are individually less powerful), to take responsibility for their actions [22, 23]. We rather define by-laws as a social process of the translation of norms and rules that involve a “collective expectation of behaviour in terms of what ought to be; a collective expectation as to what behaviour will be; and/or particular reactions to behaviour, including attempts to apply sanctions…” (Gibbs, 1965: 589) [24]. Norms express particular values and relations of power, and they are gendered in the following ways: By studying the by-laws as a social process of norm formulation and implementation from a gender perspective, we aim to understand how responsibilities and accountabilities are constructed, negotiated and shifted. Data collection took place between April and June 2015. The study was coordinated by a Dutch researcher and conducted in collaboration with a Malawian non-governmental organisation working on maternal health care and Dutch and Malawian research assistants. Mzimba district is divided in ten Traditional Authorities (governed by chiefs); out of these, two were selected for the study; they had been part of the project of the partnering organisation. In the first instance, one research site (health facility and its catchment area) was selected where participants were purposefully selected to represent the diversity of views on community participation in maternal healthcare. Because one health facility only has two skilled health workers, we included four more study sites (four health facilities) for additional data collection. Table 1 provides some maternal health data of the study sites involved. Basis statistics of health centre study sites Data provided by the Health Management Information Systems of the District Health Office aEstimation by facility manager bOn average, northern part of the district cMedical Assistant (clinician) dNurse-Midwife Technician eHealth Surveillance Assistant (Community health worker) fHospital Attendant/Assistant According to the authors of our theoretical framework, norms are expressed in rules (informal conventions as well as formal procedures), practices (behaviour), narratives and mechanisms of enforcement (which may consist of an actual sanction but also of arguments why a norm is to be held valid), which can be observed by researchers. Formal rules are conveyed through documents, practices are conveyed through examples of implementation and narratives are transmitted through storytelling and symbols [26]. We collected data through document search, observation and semi-structured interviews as well as focus group discussions (FGD) to convey points of agreement and contestation regarding norms in maternal health care. Where available, documents such as minutes of meetings (health committee, local government) and texts on the by-laws were collected. In one Traditional Authority a large meeting took place on by-laws during the study period; the main researcher (author 1) used it to observe interactions, collect information on the formulation of by-laws and to talk to local chiefs and local government councillors. Interviews and FGDs took place with participants of different groups involved in maternal health care. The interview and FGD guidelines included four main topics: (1) experiences/engagement with maternal health services; (2) perceptions on the quality of care; (3) community participation and responsibilities in maternal health care; and (4) priorities for better maternal health care. In the FGDs, participants were given additional exercises to discuss challenges in the organisation and quality of maternal health care and to discuss responsibilities and accountabilities. The researchers encouraged discussions through probing questions (e.g. “who do you think is responsible for improving maternal health in this community?”; “if you were to introduce a code of conduct for health workers and service users in this health centre, what would be the most important duty you would formulate?”). The term “by-law” was not explicitly part of the interview guidelines but emerged from the participants as they shared their perceptions on responsibilities in maternal health care under topics 3 and 4 of the interview guideline and in the FGD. Interview and FGD guides were translated from English to Tumbuka, back translated and tested with key informants familiar with interviewing techniques in the research area. Interviews lasted between 50 and 60 min, and FGDs ranged in size from 5 to 8 participants and lasted on average 1.5 h. Only the interviews and FGDs with health workers were held in English; a translator assisted the others. One researcher (author 2) conducted most individual interviews whereas two researchers (authors 1 and 2) conducted the FGDs together or separately. Table 2 presents the type and number of study participants included in this study. The final number of participants is 137 of which 36 participated in individual interviews and 101 in FGDs. Participants included 35 female maternal health service users of which three were guardians (women escorting pregnant women to the health centre for childbirth); 19 men with experience with the health centre as escorts or as husbands; 25 health workers, including auxiliary staff, health surveillance assistants and facility managers; 34 community leaders (of which 24 local government representatives and 10 chiefs); 20 community representatives of the health committee and 4 key informants (from the district health office and non-governmental organisations). Female representation in the health worker group was 36%, in the community leaders group 25% and in the health committee group 50%. Number and types of participants per study site In both Traditional Authorities, local government secretaries helped the organisation of the interviews and FGDs with members from local government, health committees, village development committees and chiefs. The researchers contacted health facility managers directly and informed them about the research. Women were identified with the assistance of health surveillance assistants or at the health centre when attending postnatal services; they were asked to participate in an interview or FGD. A point of saturation approach was applied to the sample of all participants when the range and distribution of actors, views and experiences were covered in different sites and when the researchers expected no further insights to emerge. The sample of chiefs is small; in particular the views of different levels of chiefs may be underrepresented. For example, it was hard to recruit the higher level of chiefs for individual interviews. Six chiefs were briefly interviewed during the by-law meeting; the interviews were not recorded and did not follow the interview guidelines of the other groups but focused on the by-laws. Interviews, FGDs, and the by-law meeting were recorded, transcribed and translated into English by research assistants and checked by the researchers. Transcripts and collected documents were introduced and analysed in Maxqda (Version 11). The first step of the analysis involved exploring the (type of) information provided in each data source and using a few transcripts from different participants to explore stories, their contexts and variations as well as the terms participants used to talk about responsibilities and accountabilities in maternal health. In the second step, we extracted segments relating to the by-laws under one general code and later categorised them according to the topic of the by-laws (ANC, male involvement and institutional deliveries). A third step involved the identification of themes (main codes) using the four dimensions of the theoretical framework: formulation, interpretation, implementation and effects, that contained segments on the background, descriptions, perceptions and accounts on the implementation of the by-laws by participant group. We used the notes and report of the by-law meeting to identify the wording of the by-laws and the main points of discussion. Results of this third step of analysis are presented in the results section. The fourth step involved the analysis of the gendered nature of the by-law process and effects. For this, we applied a gender analysis to the coded segments. This was followed by a separate analysis (step 5) of the actors involved across the themes of formulation, interpretation and implementation. Data were first analysed and collated per participant group and then compared to identify gender differences in answers. A description of data coding and analysis questions for steps three to five is provided in Table 3. The results of the gender and actor analysis are mainly reflected in the discussion section. Coding framework and steps of analysis