Menu
Menu
Menu
Menu
Background: Ghana is attracting global attention for efforts to provide health insurance to all citizens through the National Health Insurance Scheme (NHIS). With the program’s strong emphasis on maternal and child health, an expectation of the program is that members will have increased use of relevant services. Methods: This paper uses qualitative and quantitative data from a baseline assessment for the Maternal and Newborn errals Evaluation from the Northern and Central Regions to describe women’s experiences with the NHIS and to study associations between insurance and skilled facility delivery, antenatal care and early care-seeking for sick children. The assessment included a quantitative household survey (n = 1267 women), a quantitative community leader survey (n = 62), qualitative birth narratives with mothers (n = 20) and fathers (n = 18), key informant interviews with health care workers (n = 5) and focus groups (n = 3) with community leaders and stakeholders. The key independent variables for the quantitative analyses were health insurance coverage during the past three years (categorized as all three years, 1-2 years or no coverage) and health insurance during the exact time of pregnancy. Results: Quantitative findings indicate that insurance coverage during the past three years and insurance during pregnancy were associated with greater use of facility delivery but not ANC. Respondents with insurance were also significantly more likely to indicate that an illness need not be severe for them to take a sick child for care. The NHIS does appear to enable pregnant women to access services and allow caregivers to seek care early for sick children, but both the quantitative and qualitative assessments also indicated that the poor and least educated were less likely to have insurance than their wealthier and more educated counterparts. Findings from the qualitative interviews uncovered specific challenges women faced regarding registration for the NHIS and other barriers such lack of understanding of who and what services were covered for free. Conclusion: Efforts should be undertaken so all individuals understand the NHIS policy including who is eligible for free services and what services are covered. Increasing access to health insurance will enable Ghana to further improve maternal and child health outcomes.
Data came from a mixed-methods baseline assessment for an evaluation of the Maternal and Newborn Referrals Project. The project is being implemented by the Institute for Healthcare Improvement (IHI), the National Catholic Health Service (NCHS) and the Ghana Health Service (GHS). Fieldwork for the evaluation’s baseline assessment was conducted between May and June 2012 in the Northern and Central Regions of Ghana. We used a simultaneous approach whereby we collected and analyzed quantitative and qualitative data at the same time and then integrated our findings related to the study aims of this paper [22]. Such an approach enabled us to obtain a richer and more comprehensive understanding of individual behaviors and community and contextual dynamics related to insurance, than a single method alone [23]. Mixed-methods approaches are increasingly being used in the study of diverse models of health insurance across Africa, reflecting the need to integrate multiple perspectives and sources of information to understand the complex determinants of insurance uptake and the relationships between insurance and outcomes [21,24,25]. Information from the baseline assessment was used to inform interventions under the Maternal and Newborns Referrals Project. Ethics review approval for the study was obtained by the University of North Carolina at Chapel Hill and the GHS. Informed consent was obtained from all study participants. The quantitative assessment included a household survey with 1267 women and interviews with 62 community leaders, which is about one leader per community. (In two large communities two leaders were interviewed.) The purpose of the household survey was to obtain information on knowledge, attitudes and practices regarding maternal and child health services. The focus of the community leader questionnaire was to understand community-level factors and barriers to the use of health services. The household survey employed the 30 by N cluster sample design, which is commonly used in child survival programs [26]. The overall sampling strategy was designed to meet the evaluation objectives for the Maternal and Newborn Referral Project. At baseline, the goal was to include a large sample of recently pregnant women to identify their experiences with pregnancy, childbirth, and newborn health. Thus, we started with a sampling strategy of a 30 by 7 approach to identify thirty clusters per region (Northern or Central), and seven recently pregnant women (pregnant in the last 12 months) in each cluster were to be randomly selected for interview (see below). To supplement the sample of 210 recently pregnant women, we also included 14 nearby neighbor women (ages 15–49) who were not necessarily recently pregnant to permit an examination of maternal and newborn health knowledge, attitudes, and behaviors of women in the community. The target sample was 630 women (210 women with a recent birth and 420 additional women) in both the Northern and Central Regions. (The target sample size was actually exceeded by seven for a total of 1267 instead of 1260 women). Thirty randomly selected communities within three districts in the Northern Region and 30 randomly selected communities in three districts of the Central Region were included in the sample. The districts were chosen by the project implementation team based on current and planned project activities. The recently pregnant women were randomly sampled from a list of all recently pregnant women in the community (determined through interviews with community leaders and health workers). Cluster sampling is advantageous because it provides a means to obtain a representative sample from the region without undertaking a census of households in the community. In this case, based on an exhaustive list of communities in the six districts (three in Northern region and three in Central region), it was possible to select a random sample of communities to represent the study districts. This is an efficient sampling method, but it leads to biased standard errors due to the correlation between observations from the same cluster. We explain our approach for accounting for the biased standard errors in the quantitative analysis section. The target sample size of 1260 was determined based on the broader objectives of the evaluation study of looking at changes in key outcomes over time. For the purpose of our descriptive paper which used baseline data only and accounting for plausible design effect, our sample size is adequate to obtain precise estimates of our key outcomes. Two maternal outcomes, facility delivery by a SBA and four or more ANC visits were studied for a woman’s most recent pregnancy in the last three years. One child health outcome, health seeking by severity of illness for children under-five, was also included in the analysis. The World Health Organization (WHO) defines a SBA as an individual trained to proficiency in the skills needed to manage normal pregnancy, childbirth and the immediate postnatal period, and in the identification, management and referral of complications in women and newborns [27]. Increasing the percent of deliveries attended by a SBA in a health facility is widely regarded as a key strategy to reduce maternal mortality [28]. The World Health Organization (WHO) promotes at least four ANC visits as a strategy to improve both maternal health and birth outcome [29]. ANC is also a means to link women to health services, and women who attend ANC visits have been shown to be more likely to have a facility delivery [30]. The child health outcome was defined as a binary variable indicating a women’s perspective on whether or not an illness had to be severe (either very serious or somewhat serious versus slightly serious or non-serious) for her to bring her under-five child to a health facility. Early care-seeking is an important strategy to reduce under-five mortality because illnesses such as malaria and pneumonia can progress rapidly in young children if treatment is delayed. It is estimated that poor or delayed care-seeking contributes to up to 70% of under-five deaths [31]. Two health insurance variables were studied – NHIS coverage in the past three years and coverage while pregnant. In the quantitative household questionnaire a matrix was used to ask women about NHIS coverage currently and during the past seven years, but only data on current coverage and coverage during the past three years was used for this variable because of missing responses beyond three years. A decision was made to make the coverage in the past three years variable categorical because conceptually we expected differences between women with coverage for all three years versus women with no coverage and women with coverage for 1–2 years (which often reflected the calendar year(s) that they were pregnant). The coverage variable was categorized as all three years, one to two years and no coverage and was included in the analysis for all three outcomes. The NHIS coverage while pregnant variable was binary and was included in the analyses for the maternal health outcomes for women with a pregnancy in the past three years. We expected both health insurance variable be associated with increased use of services. Distance and transportation (either the availability of transport or money for transport) were noted as barriers to maternal and child health services in this study population and other populations in sub-Saharan Africa [32-37]. Staffing of health facilities and shortages of health workers remain a challenge for many rural areas of low and middle income countries [38]. Within Ghana the Northern region is a particular concern and has the lowest doctor-to-population and nurse-to-population ratios [39]. Though Ghana health policy indicates that health centers should be staffed by a midwife, many health centers in the Northern region face gaps in coverage. To capture these community-level factors, community leaders were asked questions about access (distance to a health facility) and availability of health services (whether midwives are present at the nearest health facility). Two distance variables were created: 1) distance to the nearest health center or hospital used in the analysis of maternal health outcomes, and 2) distance to the nearest health post, health center or hospital used in the analysis of the child health outcome. Separate variables were created because the most common source of ANC and skilled delivery are health centers and hospitals that are staffed by midwives. Conversely, for the child health outcomes families have a broader choice of health care options to meet the needs of sick children including health posts, health centers, and hospitals. The midwife staffing variable was included in the analyses for both the maternal health outcomes; it is measured based on the presence of a midwife at the nearest facility (categorized as throughout the year, part of the year, not at all or don’t know). We expect that distance and lack of a midwife at the nearest facility will be deterrents to care-seeking. Several individual level control variables were included in the analysis based on previous studies that have demonstrated that these variables are associated with the maternal and child health outcomes of interest [32-37,40-43]. These control variables include the woman’s age, parity at the time of last birth, education level, religion, working status, urban/rural residence and ethnicity (dominant for the community or not dominant). A variable indicating region – Northern or Central was also included. Age of child (categorized as <1 year, 1–2 years and 3–4 years) was included in the analysis for the early care-seeking outcome. A household-level wealth variable was included and was constructed by first examining characteristics of households which distinguished poor and non-poor households in the 2008 Ghana DHS and characteristics which were also included in the Maternal and Newborn Referrals Survey. Type of toilet, source of drinking water, and cooking fuel were the characteristics that best differentiated the lowest and highest wealth quintiles in the Ghana DHS. These variables were then used to create a wealth variable distinguishing households that had all characteristics as the wealthiest, one or two as middle income and zero as the poorest. A similar approach applying three characteristics to classify poor and non-poor households has been used previously in studies in sub-Saharan Africa [44,45]. As a check, our wealth variable was examined relative to the number of assets (i.e. television, radio and other household items) and was found to be consistent in the Northern and Central Regions such that poor women had the least number of household assets compared to wealthier women. The study included qualitative interviews including birth narratives with 20 mothers, 18 fathers, in-depth interviews with 5 health care providers and 3 focus groups with community leaders and key informants. The purpose of the birth narratives with mothers and fathers was to elicit their personal experiences with complications during pregnancy and delivery and to understand the role of referral systems. The in-depth interviews and focus groups aimed to obtain the perceptions, opinions and norms of health care providers and community leaders regarding barriers to effective referrals, community context and strategies for improvement. The field team asked health workers at health centers in the two study regions to generate a list of approximately forty women and/or their newborns who experienced complications prior to or following their arrival at a health facility during pregnancy or delivery in the past year. With the assistance of a community health worker or assemblyman, the team requested interviews with twenty mothers and twenty fathers on this list. One health facility was selected in both the Northern and Central Regions, and two to three health workers were recruited to participate. For the focus groups, community assemblymen and other community mobilizers recruited community health volunteers, traditional birth attendants (TBAs), chemical sellers, transport workers, and community leaders. Qualitative interviews and focus groups were conducted using semi-structured interview guides that included questions and probes related to the assessment aims but also allowed for flexibility. Birth narrative interviews with mothers and fathers elicited in-depth descriptions of experiences with complications and referrals in an attempt to situate these experiences in the broader context. Interviews with health providers elicited their experiences and opinions about referrals. Focus groups with community leaders enabled an understanding of community-level norms and attitudes around maternal and child health. Across all interviews key topics included family and community dynamics, socio-cultural beliefs, structural barriers and facilitators (i.e. transport), and quality of care. While some participants discussed post-natal complications, the majority of interviews focused on antenatal and delivery experiences. There were two components to the quantitative analysis – 1) bivariate logistic regression and multivariable logistic regression to understand the influence of health insurance on the outcomes of interest after controlling for individual, household and community factors and 2) chi-square analyses to understand differences between those who had and did not have insurance by wealth and educational status. As explained earlier the variables for the multivariable analysis were selected based on theory and prior research. Because of the cluster sampling methodology, the analysis accounts from cluster level variation in the outcome variables. This was done by presenting robust estimates of variance. For the qualitative analysis, all interviews were audio-recorded, transcribed verbatim and translated. We used an inductive approach to the qualitative analysis in which we initiated the analysis without apriori hypotheses and we constructed our interpretation based on the salient themes we identified in the participant’s narratives related to their insurance experiences [46]. We prepared analytic summaries for all individual and group interviews to capture the main story of the birth complication or key themes related to ANC, skilled delivery and the referral system [47]. Based on these summaries, we developed a coding scheme that we systematically applied to all transcripts using the Atlas.ti software including codes related to insurance coverage, insurance registration and use of insurance during referral and delivery experiences [48]. We then developed analytic matrices based on the outputs of the coding to summarize findings for each key theme and compare between different categories of participants [49].
N/A
