Predictors of differences in health services utilization for children in Nigerian communities

Preventive Medicine, Volume 96, Year 2017

Interpretation

Health service utilization is an important component of child health promotion. Evidence shows that two-thirds of child deaths in low and middle income countries could be prevented if current interventions were adequately utilized. Aim of this study was to identify determinants of variation in health services utilization for children in communities in Nigeria. Multivariable negative binomial regression model attempting to explain observed variability in health services usage in Nigerian communities was applied to the 2013 Nigeria Demographic and Health Survey data. We included the index of maternal deprivation, gender of child, community environmental factor index, and maternal health seeking behaviour, multiple childhood deprivation index and ethnicity diversity index as the independent variables. The outcome variable was under-fives’ hospital attendance rates for acute illness. Of the 7577 children from 896 communities in Nigeria that were sick 1936 (25.6%) were taken to the health care facilities for treatment. The final model revealed that both multiple childhood deprivation (incidence rate ratio [IRR] = 1.23, 95% confidence interval [CI] 1.12 to 1.35) and children living in communities with a high ethnic diversity were associated with higher rate of health service use. Maternal health seeking behaviour was associated with a significantly lower rate of health care service use. There are significant variations in health services utilization for sick children across Nigeria communities which appear to be more strongly determined by childhood deprivation factors and maternal health seeking behaviour than by health system functions. This study was based on secondary analyses of cross-sectional population-based data from the 2013 Nigeria Demographic and Health Survey (DHS). Nigeria covers a total area of about 923,768 km2. It is the thirty-second largest country in terms of land mass and the most populous country in Africa with a recent estimate of its population as 140,431,790 (NPC, ICF International, 2013). About 67.8% of the population live in rural areas and 32% in urban areas. There are 374 identifiable ethnic groups in Nigeria with varying languages, customs and cultures (NPC, ICF International, 2013). The largest ethnic groups are the Yoruba, Hausa/Fulani and Igbo which account for about 68% of the population (NPC, ICF International, 2013). Available statistics indicate that about 8% of the population are categorised as poor, 34% as lower class, 25% as lower middle class, 18% as upper middle class, 8% as lower upper class and 3% as upper class (Nigeria Population Distribution by Socioeconomic Class, 2015). The 2013 DHS (NPC, ICF International, 2013) was conducted in Nigeria to collect data on demographic, environmental, socioeconomic, and health issues (family planning, infertility, nutritional and health status of children, their mothers and the fathers) from a nationally representative sample of 39,902 women aged 15–49 years and 18,229 men aged 15–59 years in 38,904 households (NPC, ICF International, 2013). The survey used a three-stage cluster sampling technique. The country was stratified into 36 States and the Federal Capital Territory (FCT), Abuja making 37 districts in total. The primary sample unit (PSU) was based on 2006 General Population and Housing Census enumeration areas (EAs). The first stage involved selecting 896 localities (clusters). In the second stage, one EA was randomly selected from most localities. A total of 904 EAs were selected, with 372 in urban areas and 532 in rural areas (NPC, ICF International, 2013). The third stage involved random selection of a fixed number of 45 households in every urban and rural geographical area. Data collection procedures have been published elsewhere (NPC, ICF International, 2013). Data (on demographic characteristics, wealth, anthropometry, female genital cutting and awareness of HIV/AIDS, knowledge of HIV prevention, sexual behaviour, and domestic violence) were collected by conducting face-to-face interviews with women and men who met the eligibility criteria. Among all eligible individuals and households, participation rates were 98% for household, 98% for women and 95% for men (NPC, ICF International, 2013). Each woman was asked to provide a detailed history of all her live births in chronological order, including whether a birth was single or multiple, assigned sex of the child, date of birth, survival status, age of the child on the date of interview if alive and age at death of each live birth, if the child was not still alive. Hospital attendance rates for acute illness at a community level was the response variable. We focused on data for children under-five who had had an episode of diarrhea and/or fever/cough in the preceding 2 weeks before the survey and whose parents/carers sought consultation from a health care provider (either public or private). We included the following independent variables; gender of child, community environmental factor index, maternal health seeking behaviour, multiple childhood deprivation index and ethnicity diversity index. We used composite indices because they are easier to interpret than a battery of separate indicators and because they help to construct narratives for lay and literate audiences. In addition, they reduce the visible size of a set of indicators without dropping underlying information. Furthermore, multidimensional concepts like welfare, well-being, human development, environmental sustainability, industrial competitiveness and so on cannot be adequately represented by individual indicators (OECD, 2008). We used a childhood deprivation index previously described in a study by Uthman (2009). The childhood deprivation index in this study was operationalized with a principal component comprised of the proportion of children with low birth weight, not breast fed, with short birth interval (< 24 months), high number of under-fives in the household and children with high birth order. A standardized score with mean 0 and standard deviation of 1 was generated from this index; with higher scores indicative of higher childhood deprivation (Uthman, 2009). Maternal deprivation comprised of the proportion of mothers who are non-literate, unemployed, residing in rural areas and living below the poverty level (asset index < 20% poorest quintile). This was derived using principal component analysis on 3 variables that included proportion of children in households with access to safe water, proper sanitation, and low pollution cooking fuel. A standardized score with mean 0 and standard deviation of 1 was generated with higher scores indicative of better and cleaner environmental status. This was operationalized with a principal component analysis comprised of the proportion of respondents: with a health card, who attended ante natal care clinic and received tetanus vaccine during pregnancy, with the child's delivery in the hospital and with child received at least one dose of vaccination. A standardized score with mean 0 and standard deviation of 1 was generated from this index; with higher scores indicative of better MHSBI. The ethnicity of the children was computed by using an ethnicity diversity index. This index was created using a formula (Eq. (1) below) which captures both the number of different ethnic groups in an area and the relative representation of each group (Vyas and Kumaranayake, 2006). where: xi = population of ethnic group i of the area, y = total population of the area, n = number of ethnic groups in the area. Scores can range from 0 to approximately 1. For clarity of interpretation, each diversity index is multiplied by 100; the higher the index score, the greater the diversity in the area. If an area's entire population belongs to one ethnic group, then the area has zero diversity. An area's diversity index increases to 100 when the population is evenly divided into ethnic groups. This study was based on secondary analysis of existing survey datasets from the archive of the DHS who granted us permission for use of anonymised data. The instruments and conduct of the 2013 Nigeria DHS was approved by the Institutional Review Board (IRB) of ICF Macro International in the United States and Nigeria Health Research Ethics Committee (NHREC) of the FMOH. This research is limited to the use of previously collected anonymised data. To determine the number of component included in the factor analyses, we used the criterion: eigenvalues ≥ 1 and we also inspected the scree plot. Factor loadings ≥ 0.4 were judged to be significant. The results of the PCA tests are included in the online Supplementary material (tables 4, 5 and 6). A negative binomial multivariable regression model due to over-dispersion of the outcome variable was developed to explain the observed variability in health services utilization described by Harrell et al. (1996) and Freemantle et al. (2009). Associations between health service utilization and all included independent variables were first examined at the univariable level. Gender and characteristics statistically associated with a P value of 0.1 at univariable level were subsequently fitted in the multivariable negative binomial regression model. Gender was included in the multivariable negative binomial regression model because we wanted to assess interaction effects of gender and other independent variables. We also fitted another multivariable negative binomial regression using backward stepwise model selection with a P value of 0.10 with the result similar to the previously described method (see Supplementary material table 3). P value of < 0.05 was used to define statistical significance. The final model was used to predict the hospital attendance rates for each community (unit of analysis derived by collapsing child level data at community level) and compared the predicted and the observed hospital attendance rates. Regression diagnostics were used to judge the goodness-of-fit of the model. They included the Likelihood Ratio test (LHR test), and the Hosmer-Lemeshow model fit test. The tolerance test, variance inflation factors (VIF), reciprocal of VIF (Tu et al., 2004, Tu et al., 2005), and estimates of adjusted R squared of the regression model were used to test for multicollinearity using the routine collin in Stata. A VIF > 10 or mean VIF > 6 represent severe multicollinearity (Hocking, 1996). There was no issue of concern regarding the regression diagnostics for model fit and multicollinearity tests. In addition, model validation evaluating potential over-fitting was carried out using bootstrapping. Briefly, 100 bootstrap samples were generated from the original datasets using a resampling technique. The original model was re-fitted in the testing datasets to estimates adjusted (corrected) estimate of the predictors. The corrected estimates of the predictors remained unchanged after bootstrapping. All statistical analyses were carried out using Stata statistical software for windows version 14 (StataCorp, 2015).

AI Digest

Study Justification:

– Health service utilization is crucial for promoting child health.
– Evidence suggests that many child deaths in low and middle income countries could be prevented with better utilization of interventions.
– This study aimed to identify factors that contribute to variations in health service utilization for children in Nigerian communities.

Study Highlights:

– The study used data from the 2013 Nigeria Demographic and Health Survey.
– A multivariable negative binomial regression model was applied to explain the variability in health service usage.
– The study found that childhood deprivation and living in communities with high ethnic diversity were associated with higher rates of health service use.
– Maternal health seeking behavior was associated with lower rates of health care service use.
– The study revealed significant variations in health service utilization for sick children across Nigerian communities, with childhood deprivation and maternal health seeking behavior playing a stronger role than health system functions.

Study Recommendations:

– Policies and interventions should focus on addressing childhood deprivation factors and improving maternal health seeking behavior to increase health service utilization for children.
– Efforts should be made to reduce disparities in health service utilization across different communities in Nigeria.
– Strategies should be developed to improve access to and quality of health care services for children in communities with high ethnic diversity.

Key Role Players:

– Ministry of Health: Responsible for developing and implementing policies to improve health service utilization for children.
– Community Health Workers: Involved in providing health care services and promoting health seeking behavior in communities.
– Non-Governmental Organizations: Can support the implementation of interventions and programs aimed at improving health service utilization for children.
– Health Care Providers: Play a crucial role in delivering quality health care services to children.

Cost Items for Planning Recommendations:

– Training and capacity building for health care providers and community health workers.
– Development and implementation of awareness campaigns to promote maternal health seeking behavior.
– Improvement of health care infrastructure and facilities in communities with low health service utilization.
– Monitoring and evaluation of interventions to assess their effectiveness and make necessary adjustments.

Strength of Evidence

The strength of evidence for this abstract is 7 out of 10.
The evidence in the abstract is moderately strong. The study is based on secondary analyses of cross-sectional population-based data from the 2013 Nigeria Demographic and Health Survey, which provides a large sample size and representative data. The study also uses a multivariable negative binomial regression model to analyze the data. However, the abstract does not provide information on the limitations of the study or potential biases. To improve the strength of the evidence, the authors could include a discussion of the limitations and potential biases in the abstract.

Abstract

This study was based on secondary analyses of cross-sectional population-based data from the 2013 Nigeria Demographic and Health Survey (DHS). Nigeria covers a total area of about 923,768 km2. It is the thirty-second largest country in terms of land mass and the most populous country in Africa with a recent estimate of its population as 140,431,790 (NPC, ICF International, 2013). About 67.8% of the population live in rural areas and 32% in urban areas. There are 374 identifiable ethnic groups in Nigeria with varying languages, customs and cultures (NPC, ICF International, 2013). The largest ethnic groups are the Yoruba, Hausa/Fulani and Igbo which account for about 68% of the population (NPC, ICF International, 2013). Available statistics indicate that about 8% of the population are categorised as poor, 34% as lower class, 25% as lower middle class, 18% as upper middle class, 8% as lower upper class and 3% as upper class (Nigeria Population Distribution by Socioeconomic Class, 2015). The 2013 DHS (NPC, ICF International, 2013) was conducted in Nigeria to collect data on demographic, environmental, socioeconomic, and health issues (family planning, infertility, nutritional and health status of children, their mothers and the fathers) from a nationally representative sample of 39,902 women aged 15–49 years and 18,229 men aged 15–59 years in 38,904 households (NPC, ICF International, 2013). The survey used a three-stage cluster sampling technique. The country was stratified into 36 States and the Federal Capital Territory (FCT), Abuja making 37 districts in total. The primary sample unit (PSU) was based on 2006 General Population and Housing Census enumeration areas (EAs). The first stage involved selecting 896 localities (clusters). In the second stage, one EA was randomly selected from most localities. A total of 904 EAs were selected, with 372 in urban areas and 532 in rural areas (NPC, ICF International, 2013). The third stage involved random selection of a fixed number of 45 households in every urban and rural geographical area. Data collection procedures have been published elsewhere (NPC, ICF International, 2013). Data (on demographic characteristics, wealth, anthropometry, female genital cutting and awareness of HIV/AIDS, knowledge of HIV prevention, sexual behaviour, and domestic violence) were collected by conducting face-to-face interviews with women and men who met the eligibility criteria. Among all eligible individuals and households, participation rates were 98% for household, 98% for women and 95% for men (NPC, ICF International, 2013). Each woman was asked to provide a detailed history of all her live births in chronological order, including whether a birth was single or multiple, assigned sex of the child, date of birth, survival status, age of the child on the date of interview if alive and age at death of each live birth, if the child was not still alive. Hospital attendance rates for acute illness at a community level was the response variable. We focused on data for children under-five who had had an episode of diarrhea and/or fever/cough in the preceding 2 weeks before the survey and whose parents/carers sought consultation from a health care provider (either public or private). We included the following independent variables; gender of child, community environmental factor index, maternal health seeking behaviour, multiple childhood deprivation index and ethnicity diversity index. We used composite indices because they are easier to interpret than a battery of separate indicators and because they help to construct narratives for lay and literate audiences. In addition, they reduce the visible size of a set of indicators without dropping underlying information. Furthermore, multidimensional concepts like welfare, well-being, human development, environmental sustainability, industrial competitiveness and so on cannot be adequately represented by individual indicators (OECD, 2008). We used a childhood deprivation index previously described in a study by Uthman (2009). The childhood deprivation index in this study was operationalized with a principal component comprised of the proportion of children with low birth weight, not breast fed, with short birth interval (< 24 months), high number of under-fives in the household and children with high birth order. A standardized score with mean 0 and standard deviation of 1 was generated from this index; with higher scores indicative of higher childhood deprivation (Uthman, 2009). Maternal deprivation comprised of the proportion of mothers who are non-literate, unemployed, residing in rural areas and living below the poverty level (asset index < 20% poorest quintile). This was derived using principal component analysis on 3 variables that included proportion of children in households with access to safe water, proper sanitation, and low pollution cooking fuel. A standardized score with mean 0 and standard deviation of 1 was generated with higher scores indicative of better and cleaner environmental status. This was operationalized with a principal component analysis comprised of the proportion of respondents: with a health card, who attended ante natal care clinic and received tetanus vaccine during pregnancy, with the child's delivery in the hospital and with child received at least one dose of vaccination. A standardized score with mean 0 and standard deviation of 1 was generated from this index; with higher scores indicative of better MHSBI. The ethnicity of the children was computed by using an ethnicity diversity index. This index was created using a formula (Eq. (1) below) which captures both the number of different ethnic groups in an area and the relative representation of each group (Vyas and Kumaranayake, 2006). where: xi = population of ethnic group i of the area, y = total population of the area, n = number of ethnic groups in the area. Scores can range from 0 to approximately 1. For clarity of interpretation, each diversity index is multiplied by 100; the higher the index score, the greater the diversity in the area. If an area's entire population belongs to one ethnic group, then the area has zero diversity. An area's diversity index increases to 100 when the population is evenly divided into ethnic groups. This study was based on secondary analysis of existing survey datasets from the archive of the DHS who granted us permission for use of anonymised data. The instruments and conduct of the 2013 Nigeria DHS was approved by the Institutional Review Board (IRB) of ICF Macro International in the United States and Nigeria Health Research Ethics Committee (NHREC) of the FMOH. This research is limited to the use of previously collected anonymised data. To determine the number of component included in the factor analyses, we used the criterion: eigenvalues ≥ 1 and we also inspected the scree plot. Factor loadings ≥ 0.4 were judged to be significant. The results of the PCA tests are included in the online Supplementary material (tables 4, 5 and 6). A negative binomial multivariable regression model due to over-dispersion of the outcome variable was developed to explain the observed variability in health services utilization described by Harrell et al. (1996) and Freemantle et al. (2009). Associations between health service utilization and all included independent variables were first examined at the univariable level. Gender and characteristics statistically associated with a P value of 0.1 at univariable level were subsequently fitted in the multivariable negative binomial regression model. Gender was included in the multivariable negative binomial regression model because we wanted to assess interaction effects of gender and other independent variables. We also fitted another multivariable negative binomial regression using backward stepwise model selection with a P value of 0.10 with the result similar to the previously described method (see Supplementary material table 3). P value of 10 or mean VIF > 6 represent severe multicollinearity (Hocking, 1996). There was no issue of concern regarding the regression diagnostics for model fit and multicollinearity tests. In addition, model validation evaluating potential over-fitting was carried out using bootstrapping. Briefly, 100 bootstrap samples were generated from the original datasets using a resampling technique. The original model was re-fitted in the testing datasets to estimates adjusted (corrected) estimate of the predictors. The corrected estimates of the predictors remained unchanged after bootstrapping. All statistical analyses were carried out using Stata statistical software for windows version 14 (StataCorp, 2015).

Materials

https://efashare.b-cdn.net/materials/ebb42e04ee7526b20f741a8634b9409b2c7c51620bb70df73a91b5e03dfa21a8/mmc1.doc

Innovations

Based on the information provided, here are some potential innovations that could be used to improve access to maternal health:

1. Mobile Health (mHealth) Solutions: Develop mobile applications or text messaging services that provide pregnant women and new mothers with information and reminders about prenatal care, vaccinations, and postnatal care. These tools can also help connect women to healthcare providers and facilitate appointment scheduling.

2. Community Health Workers: Train and deploy community health workers to provide education, support, and basic healthcare services to pregnant women and new mothers in rural areas. These workers can help bridge the gap between communities and healthcare facilities, ensuring that women receive the care they need.

3. Telemedicine: Establish telemedicine programs that allow pregnant women in remote areas to consult with healthcare providers through video conferencing or phone calls. This can help overcome geographical barriers and provide access to specialized care and advice.

4. Transportation Solutions: Develop innovative transportation solutions, such as mobile clinics or ambulances, to improve access to healthcare facilities for pregnant women in remote areas. This can help overcome transportation challenges and ensure timely access to maternal health services.

5. Financial Incentives: Implement financial incentive programs to encourage pregnant women to seek prenatal and postnatal care. This could include providing vouchers or cash transfers to cover transportation costs or offering discounts on healthcare services.

6. Maternal Health Education: Develop culturally appropriate educational materials and programs to raise awareness about the importance of maternal health and encourage women to seek care. These materials can be distributed through community centers, schools, and religious institutions.

7. Public-Private Partnerships: Foster collaborations between the government, private sector, and non-profit organizations to improve access to maternal health services. This could involve leveraging private sector resources and expertise to expand healthcare infrastructure and services in underserved areas.

8. Quality Improvement Initiatives: Implement quality improvement initiatives in healthcare facilities to ensure that pregnant women receive high-quality care. This could involve training healthcare providers, improving infrastructure and equipment, and implementing standardized protocols for maternal health services.

9. Maternal Health Insurance: Establish or expand health insurance programs that specifically cover maternal health services. This can help reduce financial barriers and ensure that women can access the care they need without facing significant out-of-pocket expenses.

10. Data-driven Decision Making: Use data from surveys, health records, and other sources to identify gaps in access to maternal health services and target interventions accordingly. This can help prioritize resources and interventions in areas with the greatest need.

It is important to note that the specific implementation of these innovations would require careful consideration of local context, resources, and stakeholder involvement.

AI Innovations Description

The study titled “Predictors of differences in health services utilization for children in Nigerian communities” aimed to identify determinants of variation in health services utilization for children in communities in Nigeria. The study used data from the 2013 Nigeria Demographic and Health Survey and applied a multivariable negative binomial regression model to explain the observed variability in health services usage.

The study found that both multiple childhood deprivation and children living in communities with high ethnic diversity were associated with higher rates of health service use. On the other hand, maternal health seeking behavior was associated with a significantly lower rate of health care service use. These findings suggest that childhood deprivation factors and maternal health seeking behavior play a stronger role in determining health services utilization for sick children in Nigeria communities than health system functions.

The study used a three-stage cluster sampling technique to select a nationally representative sample of households in Nigeria. Data on demographic characteristics, wealth, anthropometry, female genital cutting, awareness of HIV/AIDS, knowledge of HIV prevention, sexual behavior, and domestic violence were collected through face-to-face interviews with eligible women and men.

The study used composite indices to measure childhood deprivation, maternal deprivation, maternal health seeking behavior, and ethnicity diversity. These indices were generated using principal component analysis and standardized scores. Higher scores on the childhood deprivation index and ethnicity diversity index indicated higher levels of childhood deprivation and greater diversity in the area, respectively.

A negative binomial multivariable regression model was developed to explain the observed variability in health services utilization. Associations between health service utilization and independent variables were examined at the univariable level, and variables with a P value of 0.1 or less were included in the multivariable regression model. The final model was used to predict hospital attendance rates for each community and compared to the observed rates.

The study conducted regression diagnostics to assess the goodness-of-fit of the model and test for multicollinearity. No issues of concern were found regarding model fit and multicollinearity. Model validation was also performed using bootstrapping, which confirmed the stability of the model’s estimates.

In summary, the study identified childhood deprivation factors and maternal health seeking behavior as important determinants of health services utilization for sick children in Nigerian communities. These findings can inform the development of innovative interventions to improve access to maternal health, such as targeted programs addressing childhood deprivation and promoting positive maternal health seeking behavior.

AI Innovations Methodology

Based on the information provided, the study aimed to identify determinants of variation in health services utilization for children in communities in Nigeria. The methodology used was a multivariable negative binomial regression model applied to the 2013 Nigeria Demographic and Health Survey data. The independent variables included the index of maternal deprivation, gender of child, community environmental factor index, maternal health seeking behavior, multiple childhood deprivation index, and ethnicity diversity index. The outcome variable was under-fives’ hospital attendance rates for acute illness.

To simulate the impact of recommendations on improving access to maternal health, a similar methodology could be applied. First, potential recommendations could be identified based on the findings of the study and other relevant research. These recommendations could include interventions to address childhood deprivation factors, improve maternal health seeking behavior, and promote community-level interventions for better access to maternal health services.

Once the recommendations are identified, the next step would be to simulate the impact of these recommendations on improving access to maternal health. This could be done by collecting data on relevant variables, such as maternal health service utilization rates, community-level factors, and socioeconomic indicators. A regression model, such as a negative binomial regression, could then be developed to analyze the relationship between these variables and access to maternal health services.

The model could be used to predict the impact of implementing the recommendations on improving access to maternal health. This could be done by comparing the predicted hospital attendance rates for each community with and without the implementation of the recommendations. The model could also be validated using techniques such as bootstrapping to ensure its accuracy and reliability.

Overall, the methodology to simulate the impact of recommendations on improving access to maternal health would involve identifying potential recommendations, collecting relevant data, developing a regression model, and using the model to predict the impact of the recommendations.

Author

Dima Health

Statistics:

Citations: 10

Identifiers:

DOI: 10.1016/j.ypmed.2016.12.035

ISSN: 00917435

Research Areas:

Community Interventions, Disparities, Environmental, Health System and Policy, Infectious Diseases, Maternal Access, Maternal and Child Health, Quality of Care, Sexual and Reproductive Health, Social Determinants, Violence and Injury

Study Countries:

Nigeria

Study Design:

Cohort Study, Cross Sectional Study, Grounded Theory

Study Approach:

Quantitative

Participants Gender:

Male & Female