Achieving universal health coverage (UHC) is an important policy goal in both high- and low-income countries. Expanding health insurance to cover more people, increasing the scope of healthcare services and offering financial protection are among the policy tools used to advance this goal.^1,2 Many low- and middle-income countries (LMICs) are promoting health insurance schemes to shield individuals from the financial burden of healthcare and the risk of catastrophic health expenditures (CHEs) – out-of-pocket payments so high relative to income that they push families into poverty or crowd out basic needs.^3,4,5 Despite sustained efforts, however, enrolment in LMICs remains low.^6,7 This is concerning, as health insurance programmes are designed not only to shield families from CHEs but also to reduce their vulnerability to poverty – an overarching target of the Sustainable Development Goals (SDGs).

Prior work identifies drivers of low enrolment on both the demand and supply sides. On the demand side, inability to pay premiums and incomplete understanding of insurance as a risk management mechanism frequently depress participation.^7,8,9,10 On the supply side, the predominance of informal employment in many LMICs constrains contribution collection and risk pooling.^7,8,11 Another widely discussed factor is the poor quality of care in public facilities, which has been argued to deter enrolment.^12,13,14,15 Following Donabedian, we define quality as a combination of structure (resources and infrastructure), process (adherence to clinical protocols) and experience (patients’ perceptions and satisfaction).¹⁶ The hypothesis is straightforward: poor-quality care is thought to deter enrolment, as individuals may see little value in paying for insurance that entitles them to substandard services. Yet, despite the prominence of this assumption, the empirical evidence linking measured quality to insurance participation in LMICs remains scarce.

Most existing studies focus on the reverse relationship – namely, the impact of insurance on quality of care. In high-income countries, insurance expansion has been shown to improve access and outcomes: for example, Nabi et al. demonstrate that uninsured or publicly insured cancer patients in the United States are less likely to reach high-volume hospitals,¹⁷ while McWilliams summarises evidence that lack of coverage worsens access and health outcomes across multiple conditions.¹⁸ In LMICs, community-based schemes have shown similar effects: Michielsen et al. find that insurance in India improved accountability and strengthened patient bargaining power at the point of care, although challenges remained in ensuring service quality.¹⁹ Far fewer studies test whether quality influences enrolment. Evidence to date is mixed: in Ghana, lower perceived quality correlated with reduced enrolment, particularly among women,²⁰ while in Ethiopia, improvements in technical quality and patient satisfaction under community-based health insurance (CBHI) did not translate into higher enrolment or renewal.²¹The lack of robust, comparable measures of quality helps explain why the role of quality in insurance enrolment remains underexplored in LMICs.^22,23

In this study, we assess whether ANC quality is associated with women’s enrolment in health insurance in Namibia. We draw health facility information from Namibia’s 2009 Service Provision Assessment (SPA) – notably a census of all health facilities – and household information from the 2013 Namibia Demographic and Health Survey (NDHS). Using geospatial data, we link household clusters to nearby facilities and analyse the relationship between enrolment and ANC quality in a regression framework that adjusts for key socioeconomic and demographic covariates.

We chose ANC as a proxy for maternal healthcare quality for three reasons. Firstly, while maternal or infant mortality and morbidity could in principle reflect quality, these outcomes are lagging and shaped by multiple determinants beyond facility care (e.g. nutrition, transport access, emergency obstetric care, postnatal practices).²⁴ As such, they are not well suited for isolating facility performance in DHS household-linked analyses. In contrast, ANC provides contemporaneous, service-specific indicators across structure, process and experience. Secondly, the ANC module in Namibia’s 2009 SPA survey is among the most comprehensive and complete sections, with detailed measures of structural readiness, provider adherence and client feedback. Thirdly, this combination of completeness, relevance to women’s health and inclusion of patient experience allows the construction of a multidimensional quality index, making ANC a particularly robust and policy-relevant proxy for examining whether measured care quality influences insurance enrolment.

Our contributions are twofold. Substantively, this is the first national analysis in Namibia to assess whether measured service quality shapes health insurance enrolment. Methodologically, the study advances prior work by constructing a multidimensional ANC quality index (structure, process and experience) and linking a facility census to a nationally representative household survey via geospatial methods. This approach moves beyond perceptions, aligns local service environments with population data and provides a replicable framework for other LMIC setting.

We focus on Namibia for three reasons. Firstly, the 2009 SPA is a facility census, enabling comprehensive national mapping of service readiness and ANC quality – rare in LMICs.²⁵ Secondly, Namibia has relatively high but unequal health insurance coverage in the region, providing informative variation for studying determinants of participation.²⁶ Thirdly, the temporal ordering (SPA preceding NDHS) aligns with our hypothesis that facility quality can influence subsequent household enrolment decisions. These features make Namibia an instructive setting for examining how measured quality relates to insurance enrolment and for developing an approach that can be replicated elsewhere. Geographically linking household survey clusters to health facility characteristics creates an opportunity to analyse how local service environments – including quality – relate to care-seeking and financial protection decisions.^27,28,29

The remainder of the paper is organised as follows. We begin with an overview of the Namibian healthcare system, followed by a description of the data and the geospatial linkage of household insurance information to SPA facility data on ANC quality. Next, we outline the empirical strategies, encompassing both national and subnational analyses. We then present the key findings and conclude with a discussion of their implications, including study limitations, policy relevance and directions for future research.

The Namibian Ministry of Health and Social Services (MoHSS) regulates, organises and delivers healthcare to approximately 2.1 million people through 14 regional directorates and 34 health districts.^30,31 Specifically, service delivery is pluralistic: the public sector provides 70% – 75% of care, faith-based institutions, 15% – 20% and the private sector, roughly 5%.³² Public facilities primarily serve unemployed and lower-income groups,³³ while private providers are concentrated in the 13 cities and 26 towns; faith-based providers are especially important in rural areas.

Despite this mix, insurance enrolment remains low: > 80% of women and men aged 15–49 years remain uninsured.^31,34 Public services are provided through 30 district hospitals, 44 health centres and 269 clinics, supplemented by roughly 1150 outreach clinics. Namibia is sparsely populated (2.4 persons per km²), with one in five residents living in the capital, Windhoek.³⁵ While low density complicates service delivery in remote areas, coverage disparities are driven more by socioeconomic than purely geographic factors.

Government health spending constituted 13% of the national budget in 2015 (near the Abuja 15% target).^36,37 Financing shares of total health expenditure (THE) were 64% government, 20% employers, 10% households and 6% donors.^36,38 Household spending is largely out of pocket, and approximately 71% of government health expenditure is directed to secondary or tertiary care.³⁸

The private sector is financed mainly through medical-aid funds – either open (available to all) or closed (restricted to firms or industries).^34,38 The government operates the Public Service Medical-Aid Scheme (PSEMAS), a large, closed scheme. In this voluntary programme, employees contribute a flat rate and the government covers shortfalls; because contributions are not income related and benefits accrue disproportionately to higher earners, PSEMAS functions as a regressive scheme and imposes high fiscal costs.³⁹

Policy debates have focused on stronger risk pooling and equity, including proposals for a Namibia Medical Benefits Fund (NMBF) as a social health insurance (SHI) scheme, alongside broader hybrid models to enhance efficiency and financial protection.^32,40,41 Recent developments underline the centrality of public-sector quality and confidence: in 2025, authorities announced that beginning in April 2026, senior officials would be required to use public hospitals and clinics, a confidence-building measure paired with planned upgrades in staffing, medicines and infrastructure.⁴²

Health insurance enrolment is highly unequal. Only 5% of individuals in the poorest consumption quintile are covered versus 70% in the wealthiest quintile.^43,44 Coverage is also higher among urban residents and those with higher education.^31,45 In 2014, private insurance holders (including dependents and pensioners) contributed roughly Namibian (N) $2.5 billion ($220m) in premiums.⁴⁶ High unemployment and a large informal sector (about 1.53 million people were unemployed, economically inactive or under 15 in 2015⁴⁵) limit participation in voluntary schemes and heighten vulnerability to catastrophic health costs.

ANC service coverage in Namibia is high – nearly universal for a first antenatal visit and roughly two-thirds for the recommended fourth visit – yet quality is uneven.³¹ Analyses of Namibia’s SPA data by Do et al. show that while structural readiness (equipment, diagnostics and staffing) is generally adequate, there is wide variation in process quality and client experience, with inconsistent counselling and protocol adherence.⁴⁷ Global benchmarks, similarly, highlight that across LMICs, high ANC utilisation often fails to translate into consistent quality, with deficits in provider training, diagnostic supplies and counselling services.²⁴

Regionally, Namibia’s coverage of the first ANC visit is among the highest (~99%).⁴⁸ Malawi, Zambia and Zimbabwe also report high coverage, generally in the mid- to high-90% range.⁴⁹ Yet across these settings, quality gaps are evident: fewer women receive key evidence-based components such as blood pressure checks, tetanus vaccination or nutrition counselling.^47,50,51 Facility assessments in selected districts of Malawi, Zambia and Zimbabwe reinforce this point, showing that fewer than half of facilities met basic ANC infrastructure standards and that process quality (e.g. counselling, blood or urine tests) was frequently suboptimal.⁵⁰ Analyses of SPA data in Namibia, Kenya and Senegal similarly reveal deficiencies in infrastructure and supplies, prolonged waiting times and inconsistent adherence to ANC protocols. Importantly, higher-level facilities do not consistently exceed clinics in process quality, and waiting time emerges as a robust predictor of lower client satisfaction.⁴⁹ Thus, Namibia pairs unusually high ANC coverage with persistent process and counselling gaps – a pattern broadly seen across the region – making it a particularly informative setting to test whether measured ANC quality is associated with women’s insurance enrolment.

Health insurance affordability remains a salient challenge. From 2009 to 2013, average monthly medical-aid premiums increased to approximately N$1633.00, including a 17.2% rise in 2010 relative to 2009.^33,52 During this period, the average household earned about N$7500.00 per month, implying that premiums represented roughly 22% of income, with an even heavier burden on poorer households.⁵³ By 2023, premiums ranged from N$756.00 for an entry-level plan to N$3742.00 for a top plan.⁵⁴ With gross national income (GNI) per capita estimated at the United States (US) $4870.00 (about N$6087.00 per month), the top plan accounted for roughly 61.5% of average monthly income, while the entry plan still required 12.4%.⁵⁵ This financial reality demonstrates that despite economic changes, healthcare insurance premiums continue to impose a significant financial strain on many households, particularly those earning below the national average, and especially those in poverty. Although maternal and child health services are subsidised in public facilities, women face indirect costs (transport, supplies and accommodation near facilities). In rural areas, maternity waiting homes mitigate travel but charge daily fees.⁵⁶ Accordingly, even where services are subsidised, insurance can protect against indirect and ancillary costs.^5,57

Understanding the broad determinants of insurance enrolment – including ANC quality – is therefore of high policy importance.

In Table 2, we report multivariate estimates from the main specification with individual- and community-level covariates. Coefficients (×100) are interpreted as percentage point (pp) changes in the probability of having insurance per one-unit increase in the ANC quality index. For the 15 km buffer (Panel A), a one-unit increase in ANC quality is associated with a 0.02 pp change in the probability of enrolment. Estimates are of similar magnitude for the 10 km buffer, and in both cases, confidence intervals include zero (i.e. not statistically significant at conventional levels).

To probe access-constrained settings, we re-estimated the model using a 30 km buffer (Table 3). Here, a one-unit increase in the ANC quality index is associated with a 0.07 pp change in enrolment; this estimate is also not statistically significant.

A policy-relevant way to examine the association is to invert the coefficient: using the 15 km model in Table 2, an increase of roughly 50 units in the ANC quality index would be required to raise insurance coverage by 1 pp.

We also estimated a categorical specification that groups ANC quality into quintiles (Q1–Q5). Results from this sensitivity analysis (Online Appendix – Table 2-A1) mirror the continuous-index findings: relative to Q1, higher-quality quintiles show no statistically significant and no monotonic differences in enrolment at the national level.

As a further sensitivity analysis for GPS displacement, we re-estimated the 10 km and 15 km models restricting the sample to urban clusters, where DHS displaces coordinates by at most 2 km (Table 4). In these urban-only models, the ANC quality coefficients remain small and statistically non-significant, whereas education and wealth gradients are similar to those in the full sample, suggesting that the null association between ANC quality and insurance uptake is not driven by coordinate displacement.

Sensitivity analyses (Table 4) confirmed the robustness of these findings across multiple specifications, including buffer sizes, functional forms (quintiles) and urban-only subsamples.

We further examined whether the association between ANC quality and insurance differed by urban–rural residence using interaction models at the 15 km buffer. Predicted probabilities from these models (Figure 3) show relatively flat quality–insurance profiles in both rural and urban areas, with only modest level differences between locations across the observed quality range. This indicates no strong evidence that ANC quality is more strongly associated with insurance uptake in either urban or rural settings.

In this study, we examined whether measured quality of antenatal care (ANC) is associated with women’s enrolment in health insurance in Namibia, linking a facility census (SPA) to a nationally representative household survey (DHS). Three consistent findings emerge. Firstly, nationally, the association between ANC quality and insurance enrolment is small in magnitude and statistically non-significant across multiple buffers (10 km, 15 km and 30 km) and functional forms (continuous index and quintiles). Secondly, there is meaningful regional heterogeneity: in the south – where facility density is low – higher ANC quality is modestly and marginally associated with greater insurance uptake, while no association is detected in the more densely served north. Thirdly, socioeconomic gradients dominate: wealth and education are strong, monotonic correlates of enrolment irrespective of local ANC quality levels.

These results complement and extend the limited empirical work on whether service quality drives insurance demand in LMICs. Prior studies more often assess the reverse pathway – insurance improving resources, processes and patient satisfaction – than the demand pathway from quality to enrolment. Evidence on the influence of quality on enrolment is mixed: perception-based measures in Ghana suggest that lower perceived quality depresses participation, whereas Ethiopian CBHI evaluations indicate that observed improvements in technical quality and satisfaction did not translate into higher enrolment or renewal.^21,71 The Namibia findings align more closely with the latter: within the ANC domain and at national scale, higher measured quality does not appear to be accompanied by higher enrolment in a substantive way.

The null findings at national level should be interpreted in light of two features of the Namibian context. Firstly, financial barriers are substantial. Premiums represent a large share of monthly income for many households, particularly in the lower quintiles; high unemployment and informality further constrain ability to pay. Where affordability is the binding constraint, marginal quality differences – especially in a single service domain – may not shift enrolment. Secondly, choice sets differ across space. In the north, options are more numerous; proximity and convenience may dominate decision-making, dampening any incremental contribution of quality to enrolment decisions. In the south, where facility options are fewer, the observable quality of available facilities plausibly becomes more salient for households weighing the value of coverage. This heterogeneity suggests that quality weighs more heavily where density is low, while in higher-density contexts, convenience and choice play a greater role. Across both settings, however, socioeconomic factors – particularly wealth and education – remain the most consistent predictors of enrolment.

Our findings reveal how Namibia’s structural inequities shape these gradients. While ANC quality shows no national association with enrolment, the stark 5% versus 70% coverage gap between poorest or richest quintiles persists across all quality levels (Figure 5) – reflecting Namibia’s formal or informal divide where educated urbanites access employer-sponsored schemes such as PSEMAS, but the ≥ 60% informal workforce faces premiums consuming 22% of income. Education likely signals both financial literacy for scheme navigation and formal job access. This dominance of socioeconomic position over service quality underscores a core tension in Namibia’s dual economy: public facilities serve the structurally excluded, yet insurance remains stratified by formal employment and wealth.

Although our analysis focused on ANC as a proxy for maternal healthcare quality, it represents only one dimension of service delivery. ANC provides contemporaneous, measurable indicators of structure, process and client experience, offering a consistent window into the service environment most directly relevant to surveyed women. The null national association should therefore be interpreted as evidence that improvements in this slice of quality do not, on average, coincide with higher insurance uptake – though they may do so in contexts where facility choice is limited.

Improving the quality of care is essential for strengthening health outcomes within health systems; however, on its own, it is unlikely to drive health insurance uptake at the national level in Namibia. While modest associations between quality and enrolment are observed in underserved regions with lower facility density, the overall findings indicate that socioeconomic inequalities—particularly differences in wealth and education—remain the primary determinants of insurance participation. These results suggest that decisions to enroll in health insurance are shaped less by marginal improvements in service quality and more by households’ ability to afford coverage and their level of knowledge about insurance. Consequently, efforts to expand insurance coverage should prioritize reducing financial barriers through targeted subsidies and pro-poor financing mechanisms, alongside efforts to improve knowledge of health insurance and its potential benefits. At the same time, continued investments in quality of care remain critical, particularly in underserved areas where quality may play a more influential role in shaping demand. Taken together, these findings underscore the need for integrated policy approaches that align quality improvements with affordability and access to achieve more equitable progress towards universal health coverage.