A cross-sectional study of the role of men and the knowledge of danger signs during pregnancy in southern Mozambique

Background: The role of the male partner and wider family in maternal health, especially in case of emergencies, has been receiving increasing attention over the last decade. Qualitative research has highlighted that women depend on others to access high quality maternity care. Currently little is known about these factors in relation to maternal health in Mozambique. Methods: A cross sectional household survey was conducted with men and women in southern Mozambique about decision making, financial support and knowledge of danger signs. A multivariable logistic model was used to identify factors associated with knowledge of danger signs and Cohen’s kappa for agreement among couples. Results: A total of 775 men and women from Marracuene and Manhica districts were interviewed. Maternal health care decisions were frequently made jointly by the couple (32-49%) and financial support was mainly provided by the man (46-80%). Parental and parent-in-law involvement in decision making and financial support was minimal (0-3%). The average number of danger signs respondents knew was 2.05 and no significant difference (p = 0.294) was found between men and women. Communication with the partner was a significant predictor for higher knowledge of danger signs for both men (p = 0.01) and women (p = 0.03). There was very low agreement within couples regarding decision making (p = 0.04), financial support (p = 0.01) and presence at antenatal care consultations (p = 0.001). Results suggest women and men have a high willingness for more male participation in antenatal care, although their understanding of what constitutes this participation is not clear. Conclusion: The study findings highlight the important role men play in decision making and financial support for maternal health care issues. Strengthening male involvement in antenatal care services, by investing in counselling and receiving couples, could help accelerate gains in maternal health in Mozambique. Maternal health care studies should collect more data from men directly as men and women often report different views and behavior regarding maternal health care issues and male involvement. The study took place in Maputo province in the neighbouring districts of Marracuene and Manhiça, which had respectively 84,975 and 157,642 inhabitants in 2007 [23]. Formal maternal healthcare is provided entirely by public health services in this area, organised by a broad network of primary health care centers with secondary and tertiary referral centers [15]. At least 94% of women in Maputo Province receive one ANC during pregnancy and 87.5% of women deliver in a health facility [15]. Teenage pregnancy is very common: 25.8% of women between 15 and 19 years old have already been pregnant [15]. The most common direct causes for maternal deaths in Mozambique are hemorrhage, sepsis and eclampsia and among indirect causes HIV and malaria infections take the lead [24–26]. Compared to the rest of the country, Maputo Province has a reasonable coverage of health care centers: 90% of the population has a health care center within a 1 h drive [27]. A cross-sectional descriptive survey was conducted between June and August 2017. The study was nested in a cohort study in which 383 households were followed over a period of 4 years (from 2014 until 2017) in Manhiça and Marracuene districts, Mozambique. All participants within the cohort study were questioned annually about family planning knowledge, attitudes and practices. Additional questions relating to the current study were included in the final round of data collection. The questionnaire can be found as an additional file [Additional file 1]. Families were recruited through a simple, district-stratified random sampling process with allocation proportional to size within each stratum (as shown in Table 1). According to the National Institute of Statistics, 35,454 and 20,712 households lived in Manhica and Marracuene respectively in 2007. Based on the sample size calculation (as shown in Table ​Table1)1) the aim was to include 383 households, of which 242 in Manhiça and 141 in Marracuene. Considering a traditional household usually consists of at least one man and one woman of reproductive age, the aim was to include 766 men and women. Stratified Sample Technique according to Haddad et al. (2004) for calculating the sample size [28] The questionnaire consisted of questions regarding sociodemographic characteristics, knowledge of content of ANC, knowledge of danger signs during pregnancy and level of male involvement during pregnancy. The knowledge of content of ANC was assessed by the open-ended question: “What happens during ANC?”. The items listed by the respondent were categorised by the interviewer under different categories. The categories (see Fig. 1) were based on the minimum package of services to be provided by antenatal care according to WHO and MoH guidelines [29, 30]. Items not fitting in these categories were noted under the option “others”. Items listed under “others” were revised by the research team and if necessary added to a certain category after data collection. Tetanus vaccination, anaemia screening and intermittent preventive treatment of malaria were categorised under “other testing and treatments”. Knowledge content ANC by sex in percentage with confidence intervals for proportions Responses regarding knowledge of danger signs was assessed by the open ended question: “What are the danger signs during pregnancy?”. Responses were categorised by the interviewer under predetermined categories or noted under “others”. Items listed under “others” were revised by the research team and if necessary added to another category after data collection. Final categories of danger signs included: 1. Vaginal bleeding 2. Convulsions or fitting 3. Severe headache and/or blurred vision 4. Fever 5. Painful urination 6. Severe abdominal/epigastric pain 7. Reduced fetal movements 8. Swelling of fingers, face, and legs 9. Abnormal vaginal discharge 10. Others (see Fig. 2). The category “others” included answers referring to a feeling of extreme weakness, weight loss or fast and difficult breathing. Abnormal vaginal discharge included responses referring to leaking amniotic fluid or discharge with itching or smell. The categorisation of danger signs was based on the WHO handbook for health care providers and evidence regarding knowledge of danger signs from Tanzania and Madagascar [31–33]. Knowledge danger signs by sex in percentage with confidence intervals for proportions The selection of questions regarding male involvement was based on a literature review of relevant items that reflect male involvement during pregnancy and childbirth [20, 34, 35]. Different items were included: decision making regarding maternal health care issues, financial support for ANC and delivery, and male attendance at ANC consultations. Questions regarding decision making about ANC and delivery (see Table 3) only allowed for one response option; respondents were asked to select the final decision maker. Questions regarding financial support (see Table ​Table3)3) about ANC and delivery were multiple option questions. Decision making and financial support during pregnancy and delivery among participants experiencing a pregnancy in the last 5 years aMore than one response possible A team of 21 local field workers received a five-day training on ethical issues and data collection procedures, terminology used in the questionnaire and correct translation to the local language (Changana). The team of fieldworkers went from door to door, interviewing all eligible members of the selected households included in the cohort with an electronic questionnaire using tablets. Before the start of the interview, all participants received information regarding the content and objective of the questionnaire, after which written consent was obtained. The questionnaire took on average 30 to 60 min. Inclusion criteria in the cohort study included: speaking Changana or Portuguese, being in a relationship and being between 15 and 49 years old. Ethical approval was obtained from the National Health Bioethics Committee of Mozambique ((187/CNBS/15), Health Bioethics Committee of Universidade Eduardo Mondlane and Hospital Central de Maputo (CIBS UEM&HCM/0008–17) and the Bioethics Committee of Ghent University Hospital (EC/2018/1319). All data was analyzed using the statistical software package R. During data cleaning two data entries were deleted because the same participant was interviewed twice, resulting in a final dataset of 775 participants. The χ2 test was used for comparing sociodemographic characteristics by sex. The x2 test was also used for assessing a relationship between sex and a higher maternal health knowledge (dangers signs and content of ANC), together with confidence intervals for proportions (see Fig. ​Fig.11 and Fig. ​Fig.2).2). The Fisher exact test was computed for cell counts < 5. Descriptive statistics were used for exploring male presence at ANC, decision making and financial support during pregnancy and delivery. Only men and women who experienced a pregnancy in the last 5 years were included in the analysis regarding decision making, financial support and male participation at ANC to control for recall bias. Within the group of participants that experienced a pregnancy in the last 5 years a subset was selected of couples that were linked based on the question “who is your partner/husband” and year of the last pregnancy. This subset was created for examining level of inter-rater agreement with regard to male presence at ANC, decision making and financial support during pregnancy and delivery. For decision making and financial support answers were categorised under “Man/Woman/Couple together/Others”. Percent agreement was calculated by giving 0 if the man and woman of the same couple had conflicting results (eg the man says he was the final decision maker while the woman says they decided together) and 1 if they had corresponding results (eg the man says the woman was the final decision maker and the woman also says she made the final decision). Cohen’s Kappa was also calculated to examine inter-rater reliability between man and women of the same couple, as it is recommended to use both percent agreement and Cohen’s Kappa in health care studies [36]. A total score of knowledge of danger signs (ranging from 1 to 10) was calculated for all participants in the study by making the sum of danger signs listed by the participants. The Mann-Whitney U Test was used to compare the knowledge score of danger signs between men and women. We examined predictors of knowledge of danger signs for men and women that experienced a pregnancy in the last 5 years by building a binomial logistic regression model. Poor knowledge was defined as knowing less than two danger signs, this cut off value was based on the average number of danger signs respondents knew in this study (=2) and cut off values used in other studies about danger signs during pregnancy [37, 38]. Predictors included education, age, marital status, place of delivery of last child, number of antenatal care consultations during the last pregnancy, if they discussed antenatal care with their partner, male presence during ANC at the last pregnancy and number of living children. The Akaike information criterion (AIC) was used for model selection [39, 40]. P-values of less than 0.05 were considered to have significant association between the outcome and the explanatory variables, and P-values of less than 0.1 were considered borderline significant.