An index of access to essential infrastructure to identify where physical distancing is impossible

The study addresses how limited household-level infrastructure in African countries constrains the ability to practice physical distancing during infectious disease outbreaks such as COVID-19. Despite early and stringent distancing policies across Africa due to weak health systems and limited medical capacity, many households lack private water, sanitation, adequate living space, transportation, and communication tools, impeding adherence to WHO distancing guidelines. In a context of low vaccination coverage, limited testing, and substantial socioeconomic costs (e.g., GDP contraction, rising poverty, school closures), the research aims to quantify and map the capability for physical distancing, arguing that missing essential private infrastructure undermines the effectiveness of governmental measures and contributes to disease spread. The purpose is to develop an index to inform targeted interventions and preparedness for current and future epidemics.

Existing country-level indices such as WHO’s International Health Regulations (IHR) Monitoring and Evaluation Framework and RAND’s Infectious Disease Vulnerability Index primarily assess national health system capacity and governance, not household capability to prevent spread via distancing. These indices show Africa generally lags in preparedness and is highly vulnerable, with hotspots in West Africa, but lack subnational detail. Brown et al. proposed a national “home environment for protection” (HEP) index using different indicators and equal-weighted aggregation. The present work differs by focusing on indicators directly linked to social interaction (private toilet, private water source, ICT, private transport, living space), exploiting geo-referenced DHS data to provide subnational and pixel-level estimates, using PCA to derive data-driven weights, and incorporating population density. Correlations between the PDI and HEP, IHR, and vulnerability indices are weak, indicating the PDI captures a complementary dimension of risk related to household capacity for distancing.

Data: The study uses nationally representative Demographic and Health Surveys (DHS) for 34 African countries (latest available, generally 2016–2018, none earlier than 2007), covering >700,000 households, and gridded population density from SEDAC/CIESIN GPW v4. Indicators: five components reflecting essential private infrastructure required for physical distancing: (1) lack of private toilet facilities (shared sanitation), (2) lack of a private drinking water source (use of public/open sources), (3) lack of ICT (no mobile phone in household), (4) lack of private transportation (no bike, motorbike, or car), and (5) lack of space (persons per room used for sleeping). Index construction: Principal component analysis (PCA) at the household level constructs a one-dimensional physical distancing index (first principal component). Sensitivity analyses show the index is robust to exclusion of any single indicator and stable over time for countries with multiple survey years. Aggregation: Household PDIs are aggregated to country and first administrative level (admin-1) using DHS sample weights and additionally weighted by log(1 + population density) to account for heightened interaction risk in denser areas. Population density (people/km²) weighting is applied in aggregation. High-resolution mapping: Bayesian distributional regression and simulation provide pixel-level PDI estimates at 10 × 10 km across the sample and 5 × 5 km for country-specific maps (e.g., Ghana, Ethiopia, Kenya, South Africa). The Bayesian framework accommodates non-linear covariate effects (e.g., P-splines) and spatial interactions based on coordinates. Normalization: Final indices are normalized between 0 (high access to essential home infrastructure) and 1 (lowest access), using min-max across relevant comparison units (countries or regions). Outputs: Country- and region-level PDI maps and pixel-level risk maps for selected countries; correlations with observed subnational COVID-19 cases where available. Data/code availability: DHS and GPW data are public; derived datasets and R code available upon request from the corresponding author.

- Geographic heterogeneity: Substantial variation in PDI across and within countries. Highest risk areas (low capability to distance) cluster in West Africa, including Ghana, The Gambia, Togo, Sierra Leone, Benin, Liberia, Senegal, and Côte d'Ivoire, driven by relatively high population density and limited private infrastructure. Countries with lower density and comparatively better infrastructure (e.g., Namibia, Gabon, Mozambique, South Africa) show lower PDI relative to others. - Within-country hotspots: Regional hotspots are pronounced and masked by national averages, e.g., western Kenya (Kisumu, Mombasa, Nairobi), southern/central Côte d'Ivoire (Abidjan, Bas-Sassandra, Yamoussoukro), north-western Tanzania (Geita, Shinyanga, Simiyu, Tabora), and north-east South Africa (KwaZulu-Natal, Gauteng). - Pixel-level maps: 5 × 5 km PDI maps for Ghana, Ethiopia, Kenya, and South Africa reveal high subnational heterogeneity and areas where distancing is practically impossible, informing targeted interventions. - Correlation with COVID-19 cases: Across nine countries with subnational data (DRC, Ethiopia, Mozambique, Namibia, Nigeria, Niger, Senegal, South Africa, Togo), regional PDI correlates with detected cases (correlation coefficients 0.4 to 0.9), indicating utility for identifying potential transmission hotspots. - Indicator prevalence and heterogeneity: 45% of households share toilets (mean sharing with two other households; range: 1.32 in Mozambique to 6.17 in Ghana). Average 3.2 people per sleeping room; up to five in Senegal and The Gambia. 40% rely on public/open water sources. Households without a mobile phone range from 56% (Madagascar) to 3% (Senegal). Ownership of transport (bike/motorbike/car) ranges from 5% (Ethiopia) to 94% (Burkina Faso). Shared sanitation and crowding have the highest PCA weights. - Comparisons with other indices and GDP: Weak negative associations between PDI and HEP, IHR, and vulnerability indices (reported correlations around −0.15 to −0.37). Weak negative association with GDP per capita; relationship appears non-linear with substantial heterogeneity among poorer countries, implying that low income does not uniformly determine poor distancing capacity. - Double burden: Some countries (e.g., Benin, The Gambia, Sierra Leone, Togo) exhibit both high PDI and weak health system capacity. Others (e.g., Ghana) have high PDI but comparatively better health system readiness; Rwanda, South Africa, and Namibia show lower PDI and stronger health systems.

The findings demonstrate that lack of essential private infrastructure substantially limits the ability of households to adhere to physical distancing, undermining the effectiveness of public health measures. The strong regional correlations between PDI and COVID-19 cases suggest the PDI can help identify subnational hotspots where transmission may accelerate once introduced. Given the economic and social costs of broad distancing policies, the PDI supports more precise, spatially targeted strategies such as prioritizing vaccine distribution, expanded testing, and provision of protective supplies (e.g., masks, soap, and hygiene facilities at community toilets). The PDI complements health system preparedness indices by capturing a household-level dimension of vulnerability not reflected in national readiness metrics. However, infrastructure alone does not determine outcomes; adherence, testing intensity, importation risk, and demographics (e.g., older populations) also influence caseloads, as illustrated by South Africa. Overall, the PDI offers a practical tool for surveillance-response planning and resource allocation in LMICs.

Short-term containment of current and future SARS-CoV-2 variants depends on accelerating vaccination, especially for populations unable to distance due to infrastructural constraints. Long-term resilience requires investments in basic domestic infrastructure—private sanitation, water, adequate living space, transport, and communication—to reduce infection risk and improve well-being beyond pandemics. The study reveals widespread infrastructural deficits across African countries and significant within-country hotspots, implying that targeted interventions should prioritize these vulnerable regions alongside high-risk demographic groups. Addressing this infrastructure gap will enhance the effectiveness of public health measures in future epidemics and elevate quality of life.

- Scope of indicators: The PDI includes only private water, sanitation, living space, communication, transport, and population density; other relevant factors are omitted. - Behavioral and policy data: No direct measures of individual adherence, mobility, or effectiveness of government policies; no modeling of importation risk. - Predictive scope: The PDI is not a standalone predictor of national-level outbreaks; it is better suited for identifying subnational areas where spread may be faster once introduced. - Spatial resolution limits: DHS cluster coordinates are randomly displaced, limiting assessment of very fine-scale heterogeneity (e.g., 1 × 1 km, intra-urban differences). - Data recency and coverage: Some of the poorest or fragile states (e.g., South Sudan, Somalia, Central African Republic) lack recent DHS data; surveys span 2007–2018 for many countries, with limited temporal change in PDI over the last 5–10 years. - Measurement weighting: While PCA provides data-driven weights, it reflects the variance structure of included variables and may not perfectly capture epidemiological risk contributions of each component.