Comparative study of the effect of National Health Insurance Scheme on use of delivery and antenatal care services between rural and urban women in Ghana

Health Economics Review, Volume 12, No. 1, Year 2022

Interpretation

Background: Despite the focus of the National Health Insurance Scheme (NHIS) to bridge healthcare utilisation gap among women in Ghana, recent evidence indicates that most maternal deaths still occur from rural Ghana. The objective of this study was to examine the rural-urban differences in the effects of NHIS enrolment on delivery care utilisation (place of delivery and assistance at delivery) and antenatal care services among Ghanaian women. Methods: A nationally representative sample of 4169 women from the 2014 Ghana Demographic and Health Survey was used. Out of this sample, 2880 women are enrolled in the NHIS with 1229 and 1651 being urban and rural dwellers, respectively. Multivariate logistic and negative binomial models were fitted as the main estimation techniques. In addition, the Propensity Score Matching technique was used to verify rural-urban differences. Results: At the national level, enrolment in NHIS was observed to increase delivery care utilisation and the number of ANC visits in Ghana. However, rural-urban differences in effects were pronounced: whereas rural women who are enrolled in the NHIS were more likely to utilise delivery care [delivery in a health facility (OR = 1.870; CI = 1.533–2.281) and assisted delivery by a medical professional (OR = 1.994; CI = 1.631–2.438)], and have a higher number of ANC visits (IRR = 1.158; CI = 1.110–1.208) than their counterparts who are not enrolled, urban women who are enrolled in the NHIS on the other hand, recorded statistically insignificant results compared to their counterparts not enrolled. The PSM results corroborated the rural-urban differences in effects. Conclusion: The rural-urban differences in delivery and antenatal care utilisation are in favour of rural women enrolled in the NHIS. Given that poverty is endemic in rural Ghana, this positions the NHIS as a potential social equaliser in maternal health care utilisation especially in the context of developing countries by increasing access to delivery care services and the number of ANC visits. We deployed the 2014 Ghana Demographic and Health Survey (GDHS) which is a nationally representative survey administered by the Ghana Statistical Service (GSS). The 2014 GDHS employed a two-staged stratified sample frame where systematic sampling with probability proportional to size was used to identify enumeration areas from which households were selected based on 2010 Population and Housing Census. The GDHS covered 9396 eligible women aged 15–49 out of 9656 registering a response rate of 97.3%. Our focus group was women with birth histories within the past five years preceding the survey. This group constitutes 4294 women. However, after managing the data and accounting for missing observations across the three dependent variables and twelve independent variables in the inferential analyses, our total comparable sample size reduced to 4169 registering an attrition rate of 2.9%. The rural and urban sub-samples considered for the analyses are 2457 and 1712 women respectively. It is worth mentioning that the nonproportional allocation of the women sample to different regions and to their urban and rural areas using the GDHS can cause differences in probability of selection and response rates in our sample distribution. The study adjusted for these concerns by applying individual weight for women using analytic weight for the descriptive statistics and by declaring our survey design to include the individual weight variable for women divided by 1,000,000 in the case of the inferential analyses. Three main variables were used to measure delivery and antenatal care utilisation, namely place of delivery, assistance at delivery and the number of ANC visits. The place of delivery is a binary dependent variable which measures whether the delivery took place at a health facility or otherwise. Deliveries that took place at a health facility were recoded as one (1), otherwise zero (0). Assistance at delivery measures whether the birth attendant is a trained medical professional or otherwise. Birth attendants in the categories of doctor, nurse, midwife and community health officers were recoded as one (1), otherwise zero. The number of ANC visits, on the other hand, is a count variable measuring the number of antenatal care visits made during pregnancy. Whereas the first two dependent variables depend on the availability of health facilities and skilled medical professionals, the third depends on the medical condition and needs of the specific woman. However, with WHO’s current recommended number of visits of at least eight (8) as of December 2017, regular visits are encouraged for expectant mothers, than otherwise. The leading independent variable is enrolment in Ghana’s NHIS program. Respondents who are enrolled in the scheme were coded as one, and zero for those who are not enrolled. The study also controlled for demographic, socio-economic and locational factors that influence maternal health care utilisation. The demographic variables include age which was measured as current age in completed years, marital status recoded as (1 = never married; 2 = currently married; 3 = Formerly married), ethnicity dummies (1 = Akan; 2 = Ga; 3 = Ewe; 4 = Northern), and religion dummies (1 = Christian; 2 = Moslem; 3 = Traditional; 4 = No religion). The socio-economic variables include mothers’ level of education recoded (1 = no level of schooling; 2 = primary education; 3 = secondary school and beyond), employment status of mothers was recoded (0 = not employed; 1 = employed), wealth quintile which is a composite index constructed from household asset data and dwelling characteristics using principal component analyses was coded as (1 = poorest; 2 = poorer; 3 = middle; 4 = rich/richest). Two locational factors were used in the analyses, namely residential dummy (0 = rural; 1 = urban) and regional dummies (1 = Western; 2 = Central; 3 = Greater Accra; 4 = Volta; 5 = Eastern; 6 = Ashanti; 7 = Brong Ahafo; 8 = Northern; 9 = Upper East; 10 = Upper West). The PCA was used to create a continuous variable from barriers to seeking medical care (getting permission to go for treatment; getting the money needed for treatment; distance to health facility; not wanting to go alone). Each of the mentioned variables was recoded as one (1) in the case of a big problem, and zero (0) otherwise. Hence the PCA is imposed on these dummies to derive a continuous variable representing the barriers to medical care with Kaiser-Meyer-Olkin measure (KMO) of 0.65. The study deployed two main estimation techniques, the binary logistic and the negative binomial estimation techniques. The choice of the two variant estimation techniques was underscored by the six hypotheses of the study, measurement of the dependent variables and the need to correct for biases associated with overdispersion in the data. To suggest attributions, the results from the mentioned estimation techniques were verified using a quasi-experimental approach in the Propensity Score Matching. Subsequent sub-sections provide a brief description of the analytical tools deployed. The odds ratio variant of the logistic estimation technique was used to examine the rural-urban effects of NHIS enrolment on the place of delivery and the delivery care provided. This is because the two dependent variables are binary outcomes variables. The two models are specified as: Where λi1−λi is the odds that a pregnant woman delivers in a health facility, and πi1−πi is the odds that the pregnant woman received a delivery care from a medical professional, NHIS represents the NHIS enrolment, WQ is the wealth quintile, EDUC is the level of education, EMP is the employment status, MAR is the marital status, Age denotes the age, REL is the religious affiliation, ETHNi is the ethnicity variable, RES is the area of residence, REG represents the regional dummies, BTA denotes barriers to access and FTV is the frequency of watching television. The Negative Poisson estimation technique was used to analyse the third outcome variable “number of antenatal visits during pregnancy”. The choice of this estimation technique is underscored by the observation that the mentioned variable is not only a count variable, but preliminary diagnostic indicates that the variance exceeds the mean by 1.681. This cumulated into a problem of overdispersion which potentially biases the standard errors and the parameters of interest. The negative Binomial estimation technique is presented below: E(ANC) is the expected log count of the number of ANC visits, whereas the other covariates are in the case of Eqs. (1) and (2). Finally, the Incident Rate Ratio (IRR) was imposed on the expected log count of the number of ANC visits for the ease of interpretation and policy advocacy. This estimation technique enables the study to adjust for confounding effects and match women who are enrolled in the NHIS with those who are not enrolled. Given the binary nature of two of our dependent variables (place of delivery and assistance at delivery), we imposed a Linear Probability Model (LPM) assumption on their distributions to produce meaningful and intuitive corroborative results of the PSM. The PSM model for this study is stated as: Where Y1 and Y0 are the potential outcomes (delivery care and ANC visits) corresponding to women who are enrolled in the NHIS and otherwise; πi is the average treatment effect of a pregnant women enrolled in NHIS on delivery care and the number of ANC visits; H is the NHIS enrolment which is equal to 1; and X include women with similar propensities to be included in either the treated (enrolment) or the control group (non-enrolment) . The study adopted three main matching techniques, namely common support, nearest neighbour, and kernel in estimating Eq. (4). The bootstrap standard errors over 100 iterations were used to ensure robust results, whereas a seed of 1001 was used to guarantee the replicability of our PSM findings.

AI Digest

Study Justification:

This study aims to examine the impact of the National Health Insurance Scheme (NHIS) on the use of delivery and antenatal care services among rural and urban women in Ghana. Despite the efforts of the NHIS to improve healthcare access, maternal deaths still occur predominantly in rural areas. By investigating the rural-urban differences in the effects of NHIS enrollment, this study provides valuable insights into the potential of the NHIS to bridge the healthcare utilization gap and reduce maternal mortality.

Study Highlights:

– The study used data from the 2014 Ghana Demographic and Health Survey, which included a nationally representative sample of 4169 women.
– The results showed that NHIS enrollment increased delivery care utilization and the number of antenatal care visits at the national level.
– However, significant rural-urban differences were observed. Rural women enrolled in the NHIS were more likely to utilize delivery care and have a higher number of ANC visits compared to their non-enrolled counterparts. In contrast, urban women enrolled in the NHIS did not show significant differences compared to non-enrolled urban women.
– The findings suggest that the NHIS has the potential to be a social equalizer in maternal healthcare utilization, particularly for rural women in poverty.

Study Recommendations:

– The NHIS should continue to prioritize efforts to improve healthcare access and utilization among rural women, as they face greater challenges in accessing delivery and antenatal care services.
– Strategies should be implemented to address the barriers to healthcare access faced by urban women enrolled in the NHIS, as they did not show significant improvements in healthcare utilization compared to non-enrolled urban women.
– Further research is needed to explore the specific factors contributing to the rural-urban differences in healthcare utilization and to identify targeted interventions to address these disparities.

Key Role Players:

– Ministry of Health: Responsible for overseeing and implementing healthcare policies and programs, including the NHIS.
– National Health Insurance Authority: Responsible for managing and administering the NHIS.
– Ghana Statistical Service: Provides data collection and analysis support for research studies like the Ghana Demographic and Health Survey.
– Healthcare Providers: Play a crucial role in delivering quality delivery and antenatal care services to women enrolled in the NHIS.

Cost Items for Planning Recommendations:

– Training and Capacity Building: Budget for training healthcare providers to improve the quality of delivery and antenatal care services.
– Infrastructure Development: Allocate funds for the construction and maintenance of health facilities, particularly in rural areas.
– Outreach and Awareness Campaigns: Allocate resources for community outreach programs to raise awareness about the NHIS and the importance of utilizing delivery and antenatal care services.
– Monitoring and Evaluation: Budget for monitoring and evaluation activities to assess the effectiveness of interventions and identify areas for improvement.
– Research and Data Collection: Allocate funds for future research studies to gather more evidence on healthcare utilization and identify additional strategies for improvement.
Please note that the cost items provided are general categories and not specific cost estimates. Actual cost planning would require a detailed analysis and budgeting process.

Strength of Evidence

The strength of evidence for this abstract is 8 out of 10.
The evidence in the abstract is strong, as it is based on a nationally representative sample of 4169 women from the 2014 Ghana Demographic and Health Survey. The study used multivariate logistic and negative binomial models, as well as the Propensity Score Matching technique, to analyze the data. The results show significant rural-urban differences in the effects of NHIS enrollment on delivery care utilization and antenatal care services. To improve the evidence, the study could consider addressing the potential biases associated with the nonproportional allocation of the women sample to different regions and urban/rural areas. Additionally, providing more details on the specific variables used in the models and the statistical significance of the results would enhance the clarity and transparency of the evidence.

Abstract

We deployed the 2014 Ghana Demographic and Health Survey (GDHS) which is a nationally representative survey administered by the Ghana Statistical Service (GSS). The 2014 GDHS employed a two-staged stratified sample frame where systematic sampling with probability proportional to size was used to identify enumeration areas from which households were selected based on 2010 Population and Housing Census. The GDHS covered 9396 eligible women aged 15–49 out of 9656 registering a response rate of 97.3%. Our focus group was women with birth histories within the past five years preceding the survey. This group constitutes 4294 women. However, after managing the data and accounting for missing observations across the three dependent variables and twelve independent variables in the inferential analyses, our total comparable sample size reduced to 4169 registering an attrition rate of 2.9%. The rural and urban sub-samples considered for the analyses are 2457 and 1712 women respectively. It is worth mentioning that the nonproportional allocation of the women sample to different regions and to their urban and rural areas using the GDHS can cause differences in probability of selection and response rates in our sample distribution. The study adjusted for these concerns by applying individual weight for women using analytic weight for the descriptive statistics and by declaring our survey design to include the individual weight variable for women divided by 1,000,000 in the case of the inferential analyses. Three main variables were used to measure delivery and antenatal care utilisation, namely place of delivery, assistance at delivery and the number of ANC visits. The place of delivery is a binary dependent variable which measures whether the delivery took place at a health facility or otherwise. Deliveries that took place at a health facility were recoded as one (1), otherwise zero (0). Assistance at delivery measures whether the birth attendant is a trained medical professional or otherwise. Birth attendants in the categories of doctor, nurse, midwife and community health officers were recoded as one (1), otherwise zero. The number of ANC visits, on the other hand, is a count variable measuring the number of antenatal care visits made during pregnancy. Whereas the first two dependent variables depend on the availability of health facilities and skilled medical professionals, the third depends on the medical condition and needs of the specific woman. However, with WHO’s current recommended number of visits of at least eight (8) as of December 2017, regular visits are encouraged for expectant mothers, than otherwise. The leading independent variable is enrolment in Ghana’s NHIS program. Respondents who are enrolled in the scheme were coded as one, and zero for those who are not enrolled. The study also controlled for demographic, socio-economic and locational factors that influence maternal health care utilisation. The demographic variables include age which was measured as current age in completed years, marital status recoded as (1 = never married; 2 = currently married; 3 = Formerly married), ethnicity dummies (1 = Akan; 2 = Ga; 3 = Ewe; 4 = Northern), and religion dummies (1 = Christian; 2 = Moslem; 3 = Traditional; 4 = No religion). The socio-economic variables include mothers’ level of education recoded (1 = no level of schooling; 2 = primary education; 3 = secondary school and beyond), employment status of mothers was recoded (0 = not employed; 1 = employed), wealth quintile which is a composite index constructed from household asset data and dwelling characteristics using principal component analyses was coded as (1 = poorest; 2 = poorer; 3 = middle; 4 = rich/richest). Two locational factors were used in the analyses, namely residential dummy (0 = rural; 1 = urban) and regional dummies (1 = Western; 2 = Central; 3 = Greater Accra; 4 = Volta; 5 = Eastern; 6 = Ashanti; 7 = Brong Ahafo; 8 = Northern; 9 = Upper East; 10 = Upper West). The PCA was used to create a continuous variable from barriers to seeking medical care (getting permission to go for treatment; getting the money needed for treatment; distance to health facility; not wanting to go alone). Each of the mentioned variables was recoded as one (1) in the case of a big problem, and zero (0) otherwise. Hence the PCA is imposed on these dummies to derive a continuous variable representing the barriers to medical care with Kaiser-Meyer-Olkin measure (KMO) of 0.65. The study deployed two main estimation techniques, the binary logistic and the negative binomial estimation techniques. The choice of the two variant estimation techniques was underscored by the six hypotheses of the study, measurement of the dependent variables and the need to correct for biases associated with overdispersion in the data. To suggest attributions, the results from the mentioned estimation techniques were verified using a quasi-experimental approach in the Propensity Score Matching. Subsequent sub-sections provide a brief description of the analytical tools deployed. The odds ratio variant of the logistic estimation technique was used to examine the rural-urban effects of NHIS enrolment on the place of delivery and the delivery care provided. This is because the two dependent variables are binary outcomes variables. The two models are specified as: Where λi1−λi is the odds that a pregnant woman delivers in a health facility, and πi1−πi is the odds that the pregnant woman received a delivery care from a medical professional, NHIS represents the NHIS enrolment, WQ is the wealth quintile, EDUC is the level of education, EMP is the employment status, MAR is the marital status, Age denotes the age, REL is the religious affiliation, ETHNi is the ethnicity variable, RES is the area of residence, REG represents the regional dummies, BTA denotes barriers to access and FTV is the frequency of watching television. The Negative Poisson estimation technique was used to analyse the third outcome variable “number of antenatal visits during pregnancy”. The choice of this estimation technique is underscored by the observation that the mentioned variable is not only a count variable, but preliminary diagnostic indicates that the variance exceeds the mean by 1.681. This cumulated into a problem of overdispersion which potentially biases the standard errors and the parameters of interest. The negative Binomial estimation technique is presented below: E(ANC) is the expected log count of the number of ANC visits, whereas the other covariates are in the case of Eqs. (1) and (2). Finally, the Incident Rate Ratio (IRR) was imposed on the expected log count of the number of ANC visits for the ease of interpretation and policy advocacy. This estimation technique enables the study to adjust for confounding effects and match women who are enrolled in the NHIS with those who are not enrolled. Given the binary nature of two of our dependent variables (place of delivery and assistance at delivery), we imposed a Linear Probability Model (LPM) assumption on their distributions to produce meaningful and intuitive corroborative results of the PSM. The PSM model for this study is stated as: Where Y1 and Y0 are the potential outcomes (delivery care and ANC visits) corresponding to women who are enrolled in the NHIS and otherwise; πi is the average treatment effect of a pregnant women enrolled in NHIS on delivery care and the number of ANC visits; H is the NHIS enrolment which is equal to 1; and X include women with similar propensities to be included in either the treated (enrolment) or the control group (non-enrolment) . The study adopted three main matching techniques, namely common support, nearest neighbour, and kernel in estimating Eq. (4). The bootstrap standard errors over 100 iterations were used to ensure robust results, whereas a seed of 1001 was used to guarantee the replicability of our PSM findings.

Materials

N/A

Innovations

Based on the study, the recommendation to improve access to maternal health is to focus on expanding the National Health Insurance Scheme (NHIS) in rural areas of Ghana. This is because the study found that rural women enrolled in the NHIS were more likely to utilize delivery care services and have a higher number of antenatal care visits compared to their counterparts who were not enrolled. To implement this recommendation, the following steps can be taken:

1. Increase awareness: Conduct targeted awareness campaigns in rural areas to educate women about the benefits of enrolling in the NHIS and the importance of accessing maternal healthcare services.

2. Improve accessibility: Expand the coverage of NHIS facilities in rural areas by establishing more health facilities and ensuring they are adequately staffed with trained medical professionals.

3. Enhance transportation: Address transportation challenges by providing affordable and reliable transportation options for pregnant women in rural areas to access healthcare facilities.

4. Community engagement: Engage with local communities and traditional leaders to promote the importance of maternal healthcare and encourage women to enroll in the NHIS.

5. Mobile health services: Explore the use of mobile health services, such as telemedicine and mobile clinics, to reach women in remote areas and provide them with access to maternal healthcare services.

6. Financial incentives: Consider providing financial incentives, such as subsidies or discounts, to encourage pregnant women in rural areas to enroll in the NHIS and utilize maternal healthcare services.

By implementing these recommendations, it is possible to improve access to maternal health services for women in rural areas of Ghana and reduce the disparities in healthcare utilization between rural and urban areas.

AI Innovations Description

The recommendation that can be developed into an innovation to improve access to maternal health based on the study is to focus on expanding the National Health Insurance Scheme (NHIS) in rural areas of Ghana. The study found that rural women enrolled in the NHIS were more likely to utilize delivery care services and have a higher number of antenatal care visits compared to their counterparts who were not enrolled. This suggests that the NHIS has the potential to be a social equalizer in maternal healthcare utilization, particularly in rural areas where poverty is prevalent.

To implement this recommendation, the following steps can be taken:

1. Increase awareness: Conduct targeted awareness campaigns in rural areas to educate women about the benefits of enrolling in the NHIS and the importance of accessing maternal healthcare services.

2. Improve accessibility: Expand the coverage of NHIS facilities in rural areas by establishing more health facilities and ensuring they are adequately staffed with trained medical professionals.

3. Enhance transportation: Address transportation challenges by providing affordable and reliable transportation options for pregnant women in rural areas to access healthcare facilities.

4. Community engagement: Engage with local communities and traditional leaders to promote the importance of maternal healthcare and encourage women to enroll in the NHIS.

5. Mobile health services: Explore the use of mobile health services, such as telemedicine and mobile clinics, to reach women in remote areas and provide them with access to maternal healthcare services.

6. Financial incentives: Consider providing financial incentives, such as subsidies or discounts, to encourage pregnant women in rural areas to enroll in the NHIS and utilize maternal healthcare services.

By implementing these recommendations, it is possible to improve access to maternal health services for women in rural areas of Ghana and reduce the disparities in healthcare utilization between rural and urban areas.

AI Innovations Methodology

To simulate the impact of the main recommendations on improving access to maternal health, the following methodology can be used:

1. Data collection: Collect data on the current status of maternal health access in rural areas of Ghana, including information on the number of women enrolled in the NHIS, utilization of delivery care services, and number of antenatal care visits.

2. Define the simulation parameters: Determine the specific parameters to be simulated, such as the increase in NHIS enrollment, expansion of health facilities, improvement in transportation options, and implementation of mobile health services.

3. Model development: Develop a simulation model that incorporates the collected data and the defined parameters. The model should consider the relationships between NHIS enrollment, utilization of delivery care services, and number of antenatal care visits.

4. Scenario creation: Create different scenarios based on the defined parameters. For example, simulate the impact of a 10% increase in NHIS enrollment, the establishment of additional health facilities in rural areas, and the introduction of mobile health services.

5. Run the simulation: Implement the simulation model using the defined scenarios and observe the predicted outcomes. The model should provide estimates of the potential increase in NHIS enrollment, utilization of delivery care services, and number of antenatal care visits based on the implemented recommendations.

6. Analysis of results: Analyze the simulation results to assess the potential impact of the recommendations on improving access to maternal health. Compare the outcomes of different scenarios to identify the most effective strategies.

7. Sensitivity analysis: Conduct sensitivity analysis to test the robustness of the simulation results. Vary the parameters within a reasonable range to assess the potential impact on the outcomes.

8. Interpretation and reporting: Interpret the simulation results and provide a clear and concise report summarizing the findings. Highlight the potential benefits of implementing the recommendations and any limitations or uncertainties associated with the simulation.

By following this methodology, policymakers and stakeholders can gain insights into the potential impact of the recommendations on improving access to maternal health in rural areas of Ghana. This information can guide decision-making and resource allocation to effectively address the challenges and disparities in maternal healthcare utilization.

Author

Dima Health

Statistics:

Citations: 3

Identifiers:

DOI: 10.1186/s13561-022-00357-z

Research Areas:

Community Interventions, Health System and Policy, Maternal Access, Maternal and Child Health, Quality of Care, Sexual and Reproductive Health, Social Determinants

Study Countries:

Ghana

Study Design:

Case-Control Study, Cross Sectional Study, Quasi Experimental Study, randomised-control-trial

Study Approach:

Mixed-methods, Qualitative, Quantitative

Participants Gender:

Female