Social Solidarity

BUSIA UHC/POPULATION HEALTH PROGRAM BASELINE SURVEY IN BUNYALA SUB-COUNTY, BUSIA

Contributors:

Jeremiah Laktabai, Co-Principal Investigator Associate Professor, Department of Family Medicine, School of Medicine, College of Health Sciences, Moi University

Laura Ruhl, Co-Principal Investigator Assistant Professor, Department of Medicine, Indiana University School of Medicine Benjamin Andama, Co-Investigator Health Financing Manager, Academic Model Providing Access to Healthcare (AMPATH)

Caitrin Kelly, Co-investigator Assistant Professor Department of Medicine Indiana University Department of Medicine

Matthew Turissini, Co-Investigator Assistant Professor, Department of Medicine, Indiana University School of Medicine Jamil Said, Co-Investigator Assistant Lecturer, Department of Human Anatomy and Department of Medicine, School of Medicine, College of Health Sciences, Moi University

Becky Genberg, Co-Investigator Assistant Professor, Department of Epidemiology, Johns Hopkins School of Public Health

Bishnu Thapa, PhD Student Department of Health Services, Policy and Practice, Brown University, School of Public Health

Beryl Maritim, Co-Investigator Program Manager, Population Health Initiative, Academic Model Providing Access to Healthcare (AMPATH)

Michael Scanlon, Co-Investigator Assistant Director of Research, Indiana University Center for Global Health

Cornelius Lagat, Co-Investigator M&E/Data Assistant, Population Health Initiative, Academic Model Providing Access to Healthcare (AMPATH)

Allan Kimaina, Co-Investigator Statistician, Population Health Initiative, Academic Model Providing Access to Healthcare (AMPATH)

Elvirah Riungu Population Health Initiative, Academic Model Providing Access to Healthcare (AMPATH)

Michael Kibiwott Population Health Initiative, Academic Model Providing Access to Healthcare (AMPATH)

1 Backround

The Population Health Initiative is an AMPATH/MOH partnership initiated in 2016 working with county governments to develop a model for Universal Health Coverage (UHC) in Kenya. Its goal is to integrate siloed vertical services into a more comprehensive integrated system of providing care by focusing on three key strategies: (1) support economic empowerment by maximizing the power of community groups, (2) create a seamless public health care system, and (3) fully partner with NHIF to increase enrollment and retention in the insurance program. In 2019, AMPATH signed an MOU with Busia County government to design and implement the Busia UHC program. In order to provide household baseline data to inform the design and the monitoring and evaluation (M&E) of the program, we are conducting a cross-sectional mixed method baseline survey of households in Bunyala sub-County, Busia.

2 Objective

The objective of this study is to provide baseline data on UHC to inform the Busia UHC pilot. We plan to repeat this survey in future as funds become available in order to measure the impact of the UHC pilot in Bunyala sub-County on health service delivery, healthcare utilization, perceptions of healthcare quality in the public sector and household health spending.

Objective 1. To estimate UHC in Bunyala sub-County, Busia using an established UHC measurement framework

  • Objective 1.1. To estimate the health service coverage
  • Objective 1.2. To determine equity in access to preventative and curative health services
  • Objective 1.3. To estimate the prevalence of catastrophic health expenditure and the proportion of the population that is impoverished by out of pocket spending

Objective 2. To describe the correlates of NHIF health insurance enrolment among the informal sector

  • Objective 2.1. To evaluate the relationship between social solidarity, affordability of premiums, citizen awareness and empowerment and other factors and NHIF enrolment among the informal sector households.
  • Objective 2.2. To measure participation in economic empowerment interventions and its association with enrolment in NHIF

3 Population GIS

3.1 Location

drawing

3.2 Busia

drawing

4 Socio-solidarity | Bivariate Analysis

drawing

drawing

4.1 Table 1: Insurance status and household characteristics of the respondents

4.2 Table 2: Willingness to prepay and tolerance of risk cross-subsidies among the respondents

4.3 Plots

4.3.1 Respondents who agreed most with statement (can be presented in a bar graph plot)

4.3.2 Others

5 Socio-solidarity | Multivariate Analysis

5.1 Model Validation Checks

Before model interpretation, let’s check on model accuracies.

## 
## Call:
## roc.formula(formula = have_health_insurance ~ final.logit$fitted.values,     data = model.df %>% select(c(pat.y, pat.x)) %>% na.omit(),     plot = TRUE, grid = TRUE, print.auc = TRUE, show.thres = TRUE,     ci = TRUE, boot.n = 100, ci.alpha = 0.9, stratified = FALSE,     main = "ROC Curve", col = "blue")
## 
## Data: final.logit$fitted.values in 1484 controls (have_health_insurance No) < 203 cases (have_health_insurance Yes).
## Area under the curve: 0.7655
## 95% CI: 0.7289-0.8022 (DeLong)

An AUC of > 70% generally indicates a good model fit. In this case, we have an AUC of over 95%, with very precise CI

5.2 Model Tabulation

In general, all the design variables were statistically significant in unadjusted regression. After adjustment for social-demographic, and economic covariates we see some variables becoming insignificant.

Model Interpretation | Patient Loyalty
variable_name interpretation Significant
age18-24 The odds of Patient Loyalty in age18-24 is 0.08 ( 0 , 0.39 ) times lower than the odds of Patient Loyalty in the reference level, fixing all else constant. Yes
age25-44 The odds of Patient Loyalty in age25-44 is 0.54 ( 0.3 , 1 ) times lower than the odds of Patient Loyalty in the reference level, fixing all else constant. Yes
age45-64 The odds of Patient Loyalty in age45-64 is 0.99 ( 0.57 , 1.76 ) times lower than the odds of Patient Loyalty in the reference level, fixing all else constant. No
sexMale The odds of Patient Loyalty in sexmale is 1.01 ( 0.7 , 1.44 ) times higher than the odds of Patient Loyalty in the reference level, fixing all else constant. No
household size The odds of Patient Loyalty in household size is 1.01 ( 0.96 , 1.05 ) times higher than the odds of Patient Loyalty in the reference level, fixing all else constant. No
ever attended schoolYes The odds of Patient Loyalty in ever attended schoolyes is 2.28 ( 1.32 , 4.12 ) times higher than the odds of Patient Loyalty in the reference level, fixing all else constant. Yes
occupationHomemakers(Stay at home) The odds of Patient Loyalty in occupationhomemakers(stay at home) is 0.05 ( 0.02 , 0.14 ) times lower than the odds of Patient Loyalty in the reference level, fixing all else constant. Yes
occupationOthers (Specify) The odds of Patient Loyalty in occupationothers (specify) is 0.21 ( 0.02 , 1.3 ) times lower than the odds of Patient Loyalty in the reference level, fixing all else constant. No
occupationStudents The odds of Patient Loyalty in occupationstudents is 0 ( NA , 6.68021732071485e+22 ) times lower than the odds of Patient Loyalty in the reference level, fixing all else constant. No
occupationUnemployed/Seeking work The odds of Patient Loyalty in occupationunemployed/seeking work is 0.08 ( 0.02 , 0.24 ) times lower than the odds of Patient Loyalty in the reference level, fixing all else constant. Yes
occupationWorking in informal employment(eg farmers,artisans,juakali,business etc) The odds of Patient Loyalty in occupationworking in informal employment(eg farmers,artisans,juakali,business etc) is 0.07 ( 0.03 , 0.16 ) times lower than the odds of Patient Loyalty in the reference level, fixing all else constant. Yes
household income per monthKsh 1001-2000 The odds of Patient Loyalty in household income per monthksh 1001-2000 is 0.82 ( 0.44 , 1.49 ) times lower than the odds of Patient Loyalty in the reference level, fixing all else constant. No
household income per monthKsh 2001-3000 The odds of Patient Loyalty in household income per monthksh 2001-3000 is 0.64 ( 0.33 , 1.21 ) times lower than the odds of Patient Loyalty in the reference level, fixing all else constant. No
household income per monthKsh 3001-5000 The odds of Patient Loyalty in household income per monthksh 3001-5000 is 1.13 ( 0.64 , 1.99 ) times higher than the odds of Patient Loyalty in the reference level, fixing all else constant. No
household income per monthKsh 5001-10000 The odds of Patient Loyalty in household income per monthksh 5001-10000 is 1.71 ( 0.94 , 3.08 ) times higher than the odds of Patient Loyalty in the reference level, fixing all else constant. No
household income per monthKsh 501-1000 The odds of Patient Loyalty in household income per monthksh 501-1000 is 0.64 ( 0.35 , 1.15 ) times lower than the odds of Patient Loyalty in the reference level, fixing all else constant. No
household income per monthKsh 10000 and above The odds of Patient Loyalty in household income per monthksh 10000 and above is 4.99 ( 2.48 , 10.05 ) times higher than the odds of Patient Loyalty in the reference level, fixing all else constant. Yes
member financial group chamaYes The odds of Patient Loyalty in member financial group chamayes is 1.63 ( 1.15 , 2.3 ) times higher than the odds of Patient Loyalty in the reference level, fixing all else constant. Yes
Same flat amount The odds of Patient Loyalty in same flat amount is 1.44 ( 0.58 , 3.29 ) times higher than the odds of Patient Loyalty in the reference level, fixing all else constant. No
Progressive but the poor don’t pay at all The odds of Patient Loyalty in progressive but the poor don’t pay at all is 0.94 ( 0.62 , 1.46 ) times lower than the odds of Patient Loyalty in the reference level, fixing all else constant. No
Proportional The odds of Patient Loyalty in proportional is 0.97 ( 0.62 , 1.52 ) times lower than the odds of Patient Loyalty in the reference level, fixing all else constant. No
Progressive The odds of Patient Loyalty in progressive is NA ( NA , NA ) times NA than the odds of Patient Loyalty in the reference level, fixing all else constant. NA
Prepay The odds of Patient Loyalty in prepay is 1.3 ( 0.55 , 3.62 ) times higher than the odds of Patient Loyalty in the reference level, fixing all else constant. No
Wealth cross-subsidies The odds of Patient Loyalty in wealth cross-subsidies is 2.42 ( 1.09 , 6.43 ) times higher than the odds of Patient Loyalty in the reference level, fixing all else constant. Yes
Risk cross-subsidies The odds of Patient Loyalty in risk cross-subsidies is 2.86 ( 1.17 , 8.12 ) times higher than the odds of Patient Loyalty in the reference level, fixing all else constant. Yes

6 Appendix

6.1 Ordinal Data analisis

As we know that Wilcoxon Rank-Sum test can often be used provided the two independent samples are drawn from populations with an ordinal distribution.

For ordinal data, we are assuming a significance level of α = 0.05. We have a paired data over here.So,We can do hypothesis testing (two Tailed Test) to identify the given research proposition:

Hypothesis:

  • Null Hypothesis, Ho, MedianDifference = 0

  • Alternative Hypothesis, Ha, MedianDifference ≠ 0

  • Wilcoxon Rank-Sum Test For paired data set with ties,

  • WilcoxonRankSumtest was used for ordinal cohttps://rpubs.com/mominulislam2329/WilcoxonRankSumtest