The study setting
The Rwandan health system is decentralised with the community level as the first level of maternal health service provision. There are three community health workers (CHWs) in each village including Animatrice de Santé Maternelle (ASM), a CHW who is exclusively in charge of maternal health services [24, 25]. The secondary level of maternal health services is the health center. The majority of women with uncomplicated pregnancies deliver at health centres, while complicated cases are referred to the district hospital level or to the referral hospitals according to the severity of the pregnancy-related problem [25].
This study was conducted in two locations in Rwanda: the city of Kigali and the Northern Province. Kigali includes urban, semi-urban, and rural areas with a total population of 1,135,428 and 1910 villages. The Northern Province includes mainly semi-urban and rural areas with a population of 1,729,927 and 2881 villages [26].
Methodology of the study and study participants
Sampling
This study applied a cross-sectional study design. A sample size of 922 women was calculated based on the estimated prevalence of pregnancy-related hypertension (10 %) [8] with an absolute precision of 5 %, a 10 % possibility of non-responses, and a design effect of 1.5. A sampling frame prepared by the National Institute of Statistics of Rwanda (NISR) for the Rwandan general population and household census conducted in 2012 was used. This sampling frame is a complete list of villages covering the Northern Province and Kigali. From this sampling frame, 48 villages (1 %) were randomly selected. Eligible participants were women who gave birth within 13 months before data collection and who were identified with the assistance of the ASM. The final sample consisted of 921 as one contact did not participate, giving a response rate of 99.9 %.
Questionnaire
A questionnaire was developed by the research team that included questions about sociodemographic and psychosocial factors, pregnancies before the latest pregnancy, latest pregnancy, latest delivery, and postpartum situation. Sociodemographic background characteristics included age, marital status, and educational level. The majority of questions in the questionnaire were closed ended questions with a fixed number of response alternatives and Likert-type scale questions. For five questions in this study, the respondents were also given the possibility to give additional written comments. The questionnaire included questions about previous pregnancies as well as more detailed questions related to health problems during latest pregnancy, latest delivery, and postpartum period. Participants were also asked to report their SRH at one day postpartum, one week postpartum, one month postpartum, and at the time of the interview. First written in English, the questionnaire was translated into Kinyarwanda. Thereafter, the questionnaire was tested in a pilot study. The pilot study included 36 women from a village neighbouring a selected village of the study. All 36 questionnaires were completed, and apart from adjustments of the wording of some questions, no major revision of the questionnaire was needed after the pilot study.
Data collection procedures
All data were collected by a group of 12 female experienced interviewers (nurses, midwives, and clinical psychologists) through individual structured interviews to secure completeness of data. Before conducting the interviews, the interviewers participated in a five-day training. The data collection was performed between July 2014 and August 2014. At the end of each day, during the first three days of data collection, at least one participant per village was re-interviewed in order to check the completeness of the questionnaires and the accuracy of data collected. After the primary data entry, the information from 100 questionnaires, each including the 117 variables used in this study, were re-registered to check the accuracy of the first data entry. In total, 30 errors were detected which corresponds to an error rate of 0.25 % (30/11700). The erroneous data were thereafter corrected.
Dependent variables
The participants retrospectively reported their SRH four times postpartum: one day, one week, one month, and at the time of the interview. There were five available response options: very good, good, neither good nor poor, poor, and very poor. In a sub-part of analysis, the variable was dichotomised into two categories labelled good health status (good-SRH) for those who rated their health as very good or good, and poor health status (poor-SRH) for those who rated their health status as very poor or poor or neither good nor poor.
Independent variables
Sociodemographic and psychosocial variables were analyzed as independent variables. Women’s age was a continuous numerical variable that was divided into five age categories: less than 25 years, 25–29 years, 30–34 years, 35–39 years, and more than 40 years. Marital status included married, cohabiting, separated or divorced, widowed, and unmarried or single. Age at marriage was a continuous numerical variable that was categorized into less than 20 years, 21–30 years, and more than 30 years. Woman’s education was a combination of two primary variables: having attended school (yes or no) and educational level. Educational level included primary level not complete, primary level completed, vocational training, secondary level senior 1–4, secondary school senior 5–6, tertiary level, and a do not know option. The two variables were grouped into four categories: no education, completed primary level, completed secondary school or vocational training, and tertiary university level. Woman’s occupation included student, unskilled worker, skilled worker, civil servant, not employed, and other employment. Place of delivery included delivery at home, on the way to the health facility, at health post/dispensary, at the health centre, at district/provincial hospital, at referral hospital, at a private clinic, at any other health facility, and at any other place. Mode of delivery included delivered vaginally without instruments, vaginally with forceps, vaginally with vacuum extraction, planned caesarean section, and emergency caesarean section. Health insurance included the categories no health insurance, community health-based health insurance, public health insurance, and private health insurance. A new variable handicapping complication was created for women who reported either fistula, urinary incontinence, or fecal incontinence. First trimester was defined as the first three months of the pregnancy. Second trimester was defined as four to six months of pregnancy. Third trimester was defined as seven months or more. Variables about main health problems during pregnancy and delivery were for each variable collected for the first, second, and the third trimesters. Thereafter, each variable – diabetes mellitus, anaemia, and infections (composed essentially by urinary infection) during the first, the second, the third trimester – were combined to become diabetes mellitus during pregnancy, anaemia during pregnancy, and infections during pregnancy. Significant vaginal blood loss within 24 h after delivery and significant vaginal blood loss within the first weeks after delivery were combined into significant blood loss after delivery. Further categorization of variables are presented in first three tables.
Statistical analysis
Prevalence rates were calculated for description of different variables related to pregnancy, delivery, and postpartum. The study identified factors related to poor-SRH postpartum using univariable logistic regression analysis. Variables that were statistically significantly associated with poor-SRH status were considered for the final logistic regression model. Finally, a multivariable logistic regression model was built that calculated odds ratios (OR) and their 95 % confidence intervals (CI). In the multivariable model, forward stepwise regression was used. All statistically significant variables in univariable analyses were entered one by one to identify factors that had a relationship with poor-SRH at one day, one week, and one month postpartum and at time of the interview, keeping in the final model only factors that were statistically significant (p < 0.05). All multivariable models included number of births, women’s age, mode of delivery, and marital status for theoretical reasons. A Kaplan-Meier analysis using the curve “One minus survival” with a log rank test was constructed to illustrate the time-dependent self-rated poor health status in relation to anaemia during pregnancy, women’s level of education, and significant postpartum haemorrhage. This analysis was done to illustrate improvement of health status in women who reported low level of education, anaemia during pregnancy, or significant postpartum haemorrhage compared to those without these factors during the follow-up period. These analyses only included women rating their health as poor-SRH at one day postpartum. Thus, the Kaplan-Meier analyses included a sub-category of 296 participants. The time end point for each participant was the time of the interview. All analyses were performed in SPSS version 22.