Background

Maternal mortality remains a principal measure of maternal health despite the widespread recognition that the maternal death rate is only a single indicator of ill maternal health [1, 2]. Recent studies have quantified problems, such as mental distress, genital infections, breast problems [3, 4], physical complaints, pain, sleep problems [4], urinary incontinence, and anemia [5], occurring during the postpartum period. However, the available data do not fully describe maternal health according to the definition of health as “physical, mental and social well-being and not merely the absence of disease or infirmity” [6]. Women’s health-related quality of life (HRQoL) during the postpartum period is affected by their living conditions [7,8,9,10,11,12], reproductive history [7, 13, 14], and exposure to and use of reproductive health and antenatal care services [5, 15]. Maternal complications can have long-lasting consequences that impact the lives of women and their families for a long period [16].

The umbrella term antenatal care describes medical and social practices performed during pregnancy [17]. Despite the widespread use of antenatal care, its effectiveness in low-resource settings remains unclear [18, 19]. According to various observational studies, antenatal care has positive effects, such as lower maternal and perinatal mortality and better pregnancy outcomes [20]. Evidence of the effectiveness of antenatal care is necessary for decision-makers to establish adequate policies and strategies and allocate proper resources for their implementation. The self-reported HRQoL is an appropriate outcome measure in evaluations of maternal health interventions because this measure is often used in health economic evaluations [21] and can capture health events that are rarely fatal [22].

In this study, we used data obtained from a cross-sectional population-based survey in Rwanda to investigate women’s HRQoL during the first year after delivery. Our main objective was to determine whether adequate antenatal care utilization, which is a binary variable defined according to the number of visits and timing of the first visit, is positively associated with women’s HRQoL. Furthermore, we explored the association between HRQoL and socio-economic and demographic factors. To the best of our knowledge, this study is the first investigation of women’s self-reported HRQoL after delivery in Rwanda. Our study contributes to the literature regarding the burden of maternal morbidity and, ultimately, to governmental health and social policies designed to improve women’s well-being.

Methods

Study design and setting

This cross-sectional population-based survey involved women who gave birth between 1 and 13 months prior to the data collection.

The study population was randomly selected from all districts in Kigali City and the Northern Province. Kigali City has three districts, and the Northern Province has five districts. Kigali City has a population of 1,135,428, and most residents live in areas with urban characteristics (i.e., housing, economic activities, and access to infrastructures), while the Northern Province has 1,729,927 inhabitants living mainly in rural areas [23]. The total population in the selected provinces represents 27.2% of the country’s total population [23]. Kigali City and the Northern Province were chosen not only because of logistical reasons but also because these areas are considered to reflect the situation in other provinces based on various surveys conducted in Rwanda showing no major regional differences in the reproductive health indicators [24].

Participant selection

The sample size was calculated based on the estimated prevalence of hypertension as a pregnancy complication (p = 10%) [25], a precision rate of 5%, and projected non-response rates of 10 and 5% to account for the design effect.

Three-stage sampling was conducted to identify the households to include in the study. First, in eight districts (three in Kigali City and five in Northern Province), 48 primary sampling units, i.e., villages, which represent the lowest administrative unit in Rwanda, were randomly selected from a total of 3918 villages in the two provinces, corresponding to 1.2% of the total number of villages [26]. Twenty percent of the villages were selected from urban areas, and 80% of the villages were selected from rural areas to reflect the rural-urban proportions in Rwanda. Second, the number of households selected in each village was decided according to the proportion to the total number of households in each village. In each village in Rwanda, community health workers maintain records of women expecting to deliver and women with newborns and infants less than 1 year of age. Third, these records were used to randomly select the households to be visited, and the list was reduced to only include households with a woman who fulfilled the main inclusion criterion, i.e., delivery within 1 to 13 months before the survey.

If a village did not have the desired number of households fulfilling the inclusion criterion, the remaining households were selected from a neighboring village. In total, 922 women were selected and invited to participate. All women agreed to participate.

Data collection procedure

A questionnaire comprising socio-economic and demographic factors, health conditions, and the use of maternal health services was developed in English and translated into Kinyarwanda by a professional translator with experience in translating medical and public health questionnaires in Rwanda.

The data collection was performed by the University of Rwanda, School of Public Health between July and August of 2014. In addition to four PhD students who were responsible for leading the fieldwork and their five supervisors, 12 female data collectors were hired. The data collectors had previous training in nursing or a related subject (minimum of 6 years of post-primary school) and experience in population-based data collection. The data collectors received a 4-day training, including 1 day of piloting in one village (not included in the sampled area). Data entry was performed by four experienced data clerks trained on using an SPSS data-entry template (SPSS version 22.0) [27] and supervised by a data manager.

Measures

Dependent variable

The dependent variable is the HRQoL, which was measured according to the EQ-5D praxis. The following two methods [28] were used:

(1) The EQ-5D-3L descriptive system (designated EQ-5D) comprises the following five dimensions: mobility, self-care, usual activities, pain or discomfort, and anxiety or depression. The respondents selected among three options (i.e., no problems, some problems, severe problems) in response to each dimension. Each response provides a combination (e.g., 1, 1, 2, 1, and 2) that has a corresponding value, often called a score [28]. Given the lack of an official version of the EQ-5D-3L in any language spoken in Rwanda, we translated the English version of the EQ-5D-3L into Kinyarwanda and obtained retrospective permission to use the translated questionnaire from EuroQol. To calculate the EQ-5D scores, we used the weights used in a population study conducted in the UK [29].

(2) Using the visual analogue scale (designated EQ-VAS), the women were asked to express how good or poor their health was on the day of the interview by indicating a point on a scale from 0 to 100. A score of 100 represented the best imaginable health state, while a score of 0 represented the worst imaginable health state.

Independent variables

The main independent variable was adequacy of antenatal care utilization, which is a binary indicator of whether a woman had received antenatal care according to Rwandan guidelines. This variable was constructed using Kessner’s index [30] as a prototype and adapted to the Rwandan antenatal care guidelines. The Rwandan antenatal care policy was developed based on the 2002 WHO guidelines, which suggested four focused antenatal visits for normal pregnancies [31]. However, most Rwandan couples do not adhere to this recommendation. For example, in 2014/15, only 45% of women completed four visits, and the average month of initiation of the first antenatal visit was the fifth month of gestation [24]. The following two categories of this variable were constructed:

  1. (1)

    The women were categorized as having adequate antenatal care (ANC) utilization if they had completed at least four ANC visits and the first visit occurred during the first trimester.

  2. (2)

    The women who did not meet the abovementioned criterion were categorized as having inadequate ANC utilization.

The women’s socio-economic and demographic characteristics were divided into the following groups: residence was divided into two groups (i.e., rural and urban) according to the location of the village; age was divided into groups in 5-year intervals; educational level was divided into four groups (i.e., some primary school, completed primary school, lower secondary or vocational education, and upper secondary or higher education); and marital status was divided into four groups (i.e., married, cohabitating, separated/divorced/widowed, and unmarried/single).

A household wealth index was constructed by performing principal component analysis and using information regarding housing characteristics (i.e., materials used to construct walls, source of water, type of cooking fuel, and connection to electricity) and ownership of durable assets (i.e., mattress, iron, TV, mobile phone, and computer). The household scores were standardized based on a normal distribution with a mean of zero and a standard deviation of one [32]. The households were ranked according to these scores and divided into the following five equal groups: lowest, second, middle, fourth, and highest wealth quintiles.

The social support variable was constructed from seven questions regarding the respondents’ access to four main types of social support, i.e., emotional, tangible, or instrumental, informational and appraisal support, which are often referenced in the medical literature [33, 34]. The responses were dichotomized, given the values 0 (never) or 1 (sometimes, often or always) and summed for all respondents. Thus, the respondents were assigned social support values ranging between zero and seven. Finally, the values were grouped into the following two categories: poor social support (access to 0–4 types of social support) and good social support (access to 5–7 types of social support).

Statistical analysis

The average EQ-5D scores and 95% confidence intervals were calculated in relation to all independent variables. First, a bivariate analysis was performed using independent t-tests and one-way analyses of variance to identify the significant differences in the mean HRQoL values among different groups of independent variables at a 0.05 significance level. Second, two multivariable linear regression models were constructed using the EQ-5D scores and EQ-VAS scores as the outcomes and the adequacy of ANC utilization and socio-economic and demographic variables as the predictors. All predictors included in the bivariate analysis were considered in the initial model, except for the variable “Number of children” because this variable had a high proportion of missing values (31.4%), which led to biased model results. A backward stepwise procedure was performed at a p value of 0.05 to identify the variables that remained significant in the final model. The co-linearity among the covariates was assessed by performing a variance inflation factor (VIF). None of the covariates presented a VIF value above the maximum acceptable value of 10. The data analysis was performed using STATA 13 [35]. Following Pullenayegum et al. [40]. Thus, the effectiveness of antenatal care in low-income countries depends on how well this service is provided [41] and the availability of other services, such as obstetric care [42].

Regarding the second objective of this study, according to our results, having good social support positively affects women’s HRQoL during the first postpartum year. Our findings are consistent with those of several other studies investigating social support as a predictor of the HRQoL during the postpartum period [7, 10, 13, 43]. Possible explanations include that poor social support is a general predictor of postpartum depression [44], stressful events, panic disorders, psychological distress [45], and a poor mental health status [46]. Furthermore, of the five EQ-5D domains, anxiety or depression had the highest proportion of women reporting moderate or severe problems. Social support after giving birth is particularly important in the Rwandan culture. Considering the numerous practices by family and friends that comfort women during the postpartum period, new mothers find it challenging to cope with the situation with inadequate social support.

Household wealth was another socio-economic determinant of women’s HRQoL during the postpartum period. Thus, our results add to the vast literature on health inequalities in different contexts. Several previous studies investigating women’s HRQoL during the postpartum period have demonstrated the influence of household wealth or income [7, 8, 11] or other closely related indicators, such as employment and standard of living [47]. In the model using the EQ-VAS as the outcome, marital status was also significantly associated with HRQoL. In this study, higher educational levels were not associated with better HRQoL. This finding is surprising given that other studies have found good education to be a key predictor of HRQoL [7, 11]. The lack of association between the educational level and HRQoL may be partially attributable to the community health program in Rwanda, which has reduced inequalities in the use of services [48]. Other studies have reported findings similar to ours regarding marital status [8, 37]. However, notably, marital status is not the same among women in different societies [49]. In the Rwandan context, being a single woman is often associated with stigma and has been associated with poor utilization of maternal health services such as antenatal care [50].

In this study, rural women had better HRQoL than did urban women. No consensus exists in the literature regarding the rural-urban differences in HRQoL. Several scholars have argued that rural populations are less likely to be educated and have limited access to infrastructures; therefore, these populations are likely to have lower HRQoL [51]. Other scholars argue that life in urban areas is more stressful, translating to a higher prevalence of anxiety, depression, or other mental health problems [9, 52]. This hypothesis may be supported by our findings because anxiety or depression was the dimension with the widest gap between the rural and urban women as follows: on average, 85% of the women living in rural areas reported not having any anxiety or depression problems, compared with 70% of the women living in urban areas.

Regarding the number of children, although this variable was excluded from the regression analysis, according to the bivariate analysis, women with six or more children had a significantly lower HRQoL than women with fewer children using the EQ-5D. Akýn et al. [7] also established a negative association between the number of children and women’s HRQoL during 12 months postpartum. For women in settings such as Rwanda, where many individuals are rural farmers who must perform farming activities along with caring for their families, partner support is limited, and having many children may be particularly stressful.

The association of the maternal age and age of the baby with HRQoL has been previously reported [7, 14, 53] but was not supported in this study.

Methodological limitations

A major strength of our study is that we present data on women’s HRQoL collected through a population-based survey. In low-income countries, such as Rwanda, most existing evidence on postpartum morbidity is health facility based; however, very few women with postpartum health problems seek health care [3, 54].

A limitation shared by this study and other studies in the literature is that the quality and content of the antenatal care service were not explored. ANC quality is potentially a determinant of postpartum HRQoL, and thus, there is a risk of omitted variable bias. Including the ANC quality in the analysis may have also enabled a better disentanglement of the role of adequate timing and number of visits. In the case of poor quality or when the content does not meet the required standard, ANC would unlikely have much impact on HRQoL. This limitation has been noted in other studies that assessed the effectiveness of antenatal care [18].

The difference between the two outcome measures (i.e., EQ-5D and EQ-VAS) is important. In our data, more variation was observed in the HRQoL using the EQ-VAS compared to that using the EQ-5D. The social support and wealth results were similar using both methods, but using the EQ-5D, the magnitude of the estimates was small, and the model only explained a small proportion of the overall variation. Lastly the cross-sectional nature of this study does not allow for inferences regarding causal-effect relationships.

Conclusions

Based on the findings in this study, the adequacy of antenatal care utilization, which is defined as at least four ANC visits and the first visit occurring during the first trimester, is associated with women’s HRQoL during the first postpartum year. Furthermore, having good social support and living in a wealthy household are associated with higher HRQoL. These results support the current call for decision-makers to consider the role of social determinants in maternal health and well-being. Ministries with health as one of their responsibilities should work more closely with other sectors, such as those in charge of economic development and social protection. Strategies that favor the creation and reinforcement of social networks among women of reproductive age should be promoted. For example, group antenatal care could be tested in Rwanda because it has been shown to have positive effects on women’s satisfaction and in creating social support networks [55].

Finally, the quality of antenatal care in Rwanda should be assessed by not only evaluating practice based on the Rwandan antenatal guidelines but also benchmarking the guidelines based on the most recent evidence of effective interventions that should be included in antenatal practice.