Impoverishing effects of out-of-pocket health payments in Malawi
A methodological comparison of spatial multilevel models, standard multilevel and logistic regression models using survey data
Bibliographic Data
| ID | 21835594 |
|---|---|
| Authors | Atupele N Mulaga (0000-0002-9172-0366, University of Malawi, corresponding author), Mphatso Kamndaya (0000-0002-7597-3339, University of Malawi), Mphatso S Kamndaya (Malawi University of Business and Applied Sciences), Salule J Masangwi (0000-0001-7895-6780, University of Malawi) |
| Year | 2025 |
| Volume | 44 |
| Issue | 5 |
| Pages | 465-478 |
| Publication date | 2025-07-29 |
| Peer Reviewed | Yes |
| Open Access | Yes |
| Type | ARTICLE |
| Venue | African Geographical Review (JOURNAL) |
| Journal identifiers | ISSN: 1937-6812 • E-ISSN: 2163-2642 |
| Publisher | Informa UK Limited (PUBLISHER • GB) |
| DOI | 10.1080/19376812.2025.2452557 |
| OpenAlex | W4406420289 |
| Language | EN |
| References cited | 31 |
Out-of-pocket health payments are payments made by households at a point of use of health services. Researchers often use single-level logistic models to examine factors associated with impoverishment due to health payments. However, single-level models fail to account for neighborhood correlation which exist in complex survey data. The paper compares spatial multilevel to standard multilevel and single-level logistic models in terms of performance of model estimates and fit. It uses data from Malawi integrated household survey collected from 12,447 households and simulated data. Mean squared error and percentage bias were used to compare performance. Deviance Information Criterion was used to assess model fit. The results show that spatial multilevel and standard multilevel models provide similar fixed parameter estimates when both within and between neighborhood correlation exist in data while single-level model provides biased estimates and poor fit. Households with at least one chronically ill member, at least one hospitalized member or located in rural areas were significantly more likely to face impoverishment. Researchers using complex survey data should be cautious as both within and between neighborhood correlation may exist in data and failure to account for spatial correlation may lead to biased estimates consequently wrong conclusions
Econometrics · Economics · Geography · Logistic regression · Multilevel model · Payment · Statistics · Survey data collection · Computer Science · Global Maternal and Child Health · Healthcare Systems and Reforms · HIV/AIDS Impact and Responses · Mathematics
Household catastrophic health expenditure
Bayesian image restoration, with two applications in spatial statistics
Comparison of a Spatial Perspective with the Multilevel Analytical Approach in Neighborhood Studies
Bayesian Measures of Model Complexity and Fit
Assessing the household financial burden associated with the chronic non-communicable diseases in a rural district of Vietnam
Spatial disparities in impoverishing effects of out-of-pocket health payments in Malawi
Assessing medical impoverishment and associated factors in health care in Ethiopia
Catastrophic healthcare expenditure and impoverishment in tropical deltas
Bayesian random effects modelling with application to childhood anaemia in Malawi
Spatial analysis of factors associated with HIV infection in Malawi
Area variations in health
Health financing at district level in Malawi
Comparison of a spatial approach with the multilevel approach for investigating place effects on health
Comparing Spatial and Multilevel Regression Models for Binary Outcomes in Neighborhood Studies
Financial burden of household out-of pocket health expenditure in Viet Nam
Inequality in Beijing
| Citation velocity | historical |
|---|---|
| Highly cited | No |