Decomposing health inequality with population-based surveys: a case study in Rwanda

What explains the large disparity in child stunting in the Philippines? A decomposition analysis

OBJECTIVE

About one-third of under-five Filipino children are stunted, with significant socio-economic inequality. This study aims to quantify factors that explain the large gap in stunting between poor and non-poor Filipino children.

DESIGN

Using the 2015 Philippine National Nutrition Survey, we conducted a linear probability model to examine the determinants of child stunting and then an Oaxaca-Blinder decomposition to explain the factors contributing to the gap in stunting between poor and non-poor children.

SETTING

Philippines.

PARTICIPANTS

1881 children aged 6-23 months participated in this study.

RESULTS

The overall stunting prevalence was 38·5 % with a significant gap between poor and non-poor (45·0 % v. 32·0 %). Maternal height, education and maternal nutrition status account for 26 %, 18 % and 17 % of stunting inequality, respectively. These are followed by quality of prenatal care (12 %), dietary diversity (12 %) and iron supplementation in children (5 %).

CONCLUSIONS

Maternal factors account for more than 50 % of the gap in child stunting in the Philippines. This signifies the critical role of maternal biological and socio-economic circumstances in improving the linear growth of children.

Poor and non-poor gap in under-five child nutrition: a case from Nepal using Blinder-Oaxaca decomposition approach

INTRODUCTION

Many low-and middle-income countries (LMICs) have improved health indicators in the past decades, however, there is a differential in outcomes between socioeconomic groups. Systematic analysis of drivers of child nutrition gap between non-poor and poor groups has a policy relevance in Nepal and other countries to make progress towards universal health coverage (UHC). The objective of this paper was to estimate the mean height-for-age z scores (HAZ) gap between under-five children belonging to non-poor and poor groups, divide the gap into components (endowments, coefficients and interaction), and identify the factors that contributed most to each of the component.

METHODS

Information about 6277 under-five children was extracted from the most recent nationally representative Nepal Multiple Indicator Cluster Survey (MICS) 2019. HAZ was used to assess nutritional status of children. Wealth index was used to categorize children into non-poor and poor. Mean HAZ gap between groups was decomposed using Blinder-Oaxaca technique into components: endowments (group difference in levels of predictors), coefficients (group difference in effects of predictors), and interaction (group difference due to interaction between levels and effects of predictors). Detailed decomposition was carried out to identify the factors that contributed most to each component.

RESULTS

There was a significant non-poor and poor gap in nutrition outcome measured in HAZ (0.447; p < 0.001) among under-five children in Nepal. The between-group mean differences in the predictors of study participants (endowments) contributed 0.210 (47%) to the gap. Similarly, the between-group differences in effects of the predictors (coefficients) contributed 0.308 (68.8%) towards the gap. The interaction contributed -0.071 (15.8%) towards minimizing the gap. The predictors/variables that contributed most towards the gap due to (i) endowments were: maternal education, province (Karnali, Sudurpaschim, Madhesh), residence (rural/urban), type of toilet facility and ethnic group (Dalit and Muslim); (ii) coefficients were: number of under-five children in family, ethnic group (Dalit and Muslim), type of toilet facility, maternal age and education.

CONCLUSION

Decomposition of the child nutrition gap revealed that narrowing the inequality between wealth groups depends not only on improving the level of the predictors (endowments) in the poor group but also on reducing differential effects of the predictors (coefficients).

Socioeconomic inequality in premiums for a community-based health insurance scheme in Rwanda

Community-based health insurance (CBHI) has gained popularity in many low- and middle-income countries, partly as a policy response to calls for low-cost, pro-poor health financing solutions. In Africa, Rwanda has successfully implemented two types of CBHI systems since 2005, one of which with a flat rate premium (2005-10) and the other with a stratified premium (2011-present). Existing CBHI evaluations have, however, tended to ignore the potential distributional aspects of the household contributions made towards CBHI. In this paper, we investigate the pattern of socioeconomic inequality in CBHI household premium contributions in Rwanda within the implementation periods. We also assess gender differences in CBHI contributions. Using the 2010/11 and 2013/14 rounds of national survey data, we quantify the magnitude of inequality in CBHI payments, decompose the concentration index of inequality, calculate Kakwani indices and implement unconditional quantile regression decomposition to assess gender differences in CBHI expenditure. We find that the CBHI with stratified premiums is less regressive than CBHI with a flat rate premium system. Decomposition analysis indicates that income and CBHI stratification explain a large share of the inequality in CBHI payments. With respect to gender, female-headed households make lower contributions towards CBHI expenditure, compared with male-headed households. In terms of policy implications, the results suggest that there may be a need for increasing the premium bracket for the wealthier households, as well as for the provision of more subsidies to vulnerable households.

Nonadherence to tuberculosis treatment and associated factors among patients using directly observed treatment short-course in north-west Nigeria: A cross-sectional study

INTRODUCTION

Tuberculosis is a public health problem in Nigeria. One of the most effective ways of controlling tuberculosis is the directly observed treatment short-course. However, some factors influence tuberculosis patients' treatment nonadherence via directly observed treatment short-course. The study objective was to assess medication nonadherence and associated factors among tuberculosis patients in north-west Nigeria.

METHODS

A cross-sectional study enrolled tuberculosis patients using directly observed treatment short-course in public health facilities in Kano and Kaduna States from January 2015 to June 2016. The sample selection was conducted via a multistage sampling procedure. Data were collected using tuberculosis patients' demographic and clinical characteristics forms, well-validated structured instruments, and medical records. SPSS version 20 was used for data analysis. Logistic and multivariable logistic regression analyses to determine factors of medication nonadherence (P < 0.05).

RESULTS

Complete responses from 390 patients out of the 460 patients recruited for the study were used for data analyses. The mean age was 51.5 (standard deviation = ±13.8) years. The mean tuberculosis medication adherence questionnaire score was 4.35 ± 1.12. The prevalence of nonadherence to tuberculosis medication was 30.5%. Multivariable logistic regression analysis showed that having a monthly income between #100,000 and #199, 000 (adjusted odds ratio = 0.01; 95% confidence interval: 0.00-0.13), being widowed (adjusted odds ratio = 26.74, 95% confidence interval: 2.92-232.9), being married (adjusted odds ratio = 120.49, 95% confidence interval: 5.38-271.1), having a distance <5 km to directly observed treatment short-course center from home (adjusted odds ratio = 0.06, 95% confidence interval: 0.00-0.01), having a tuberculosis/HIV co-infection (adjusted odds ratio = 0.01, 95% confidence interval: 0.12-0.35), use of antiretroviral treatment and cotrimoxazole prophylaxis therapy medications (adjusted odds ratio = 24.9, 95% confidence interval: 19.6-304.3) were associated with tuberculosis medication nonadherence.

CONCLUSION

Tuberculosis medication nonadherence was high among the patients. Thus, patient-specific adherence education, attenuation of potential factors for tuberculosis medication nonadherence, and continual resource support for tuberculosis patients are needed to improve treatment outcomes.

Strongyloides stercoralis Infection at Different Altitudes of the Cusco Region in Peru

Strongyloides stercoralis affects 30-100 million people worldwide. The burden is underestimated because of the paucity of studies, limited geographical areas surveyed, and poor quality of diagnostic tests. This study aimed at determining the epidemiology of strongyloidiasis using sensitive microscopy testing in rural populations living at different altitudes in Cusco, Peru. Data were collected from subjects aged > 3 years living in Quellouno (elevation 2,600 ft) and Limatambo (elevation 8,379 ft) districts. Subjects provided one fresh stool sample and answer a standardized questionnaire. Fresh stool was tested on site using the Baermann's test and agar plate culture. Formalin-preserved stool was tested by rapid sedimentation. Eighty percent (585/715) of eligible subjects consented to participate; after excluding subjects with missing data, 65% (462/715) were included. Fifty-five percentage were female; the median age was 33 years (interquartile range 13-52), and 72% had government health insurance. Half had intestinal parasites, and Strongyloides was the most common (24.5%) followed by Giardia (15.5%), Blastocystis (14.9%), and hookworm (11.5%). The agar plate culture detected more cases of Strongyloides than Baermann's or sedimentation tests. Strongyloides infection was more common at low altitude (26.4%) than at high altitude (18.6%), but the difference was not statistically significant (P = 0.08). Older age, walking barefoot, bathing in rivers/streams, and using municipal sewage were associated with strongyloidiasis. Strongyloides was the most prevalent parasite in the areas studied and was associated with demographic, socioeconomic, and sanitary factors.

Assessing national and subnational inequalities in medical care utilization and financial risk protection in Rwanda

Background

Ensuring equitable access to medical care with financial risk protection has been at the center of achieving universal health coverage. In this paper, we assess the levels and trends of inequalities in medical care utilization and household catastrophic health spending (HCHS) at the national and sub-national levels in Rwanda.

Methods

Using the Rwanda Integrated Living Conditions Surveys of 2005, 2010, 2014, and 2016, we applied multivariable logit models to generate the levels and trends of adjusted inequalities in medical care utilization and HCHS across the four survey years by four socio-demographic dimensions: poverty, gender, education, and residence. We measured the national- and district-level inequalities in both absolute and relative terms.

Results

At the national level, after controlling for other factors, we found significant inequalities in medical care utilization by poverty and education and -in HCHS by poverty in all four years. From 2005 to 2016, inequalities in medical care utilization by the four dimensions did not change significantly, while the inequality in HCHS by poverty was reduced significantly. At the district level, inequalities in both medical care utilization and HCHS were larger than zero in all four years and decreased over time.

Conclusions

Poverty and poor education were significant contributors to inequalities in medical care utilization and HCHS in Rwanda. Policies or interventions targeting poor households or households headed by persons receiving no education are needed in order to effectively reduce inequalities in medical care utilization and HCHS.

Electronic supplementary material

The online version of this article (10.1186/s12939-019-0953-y) contains supplementary material, which is available to authorized users.