The study, published in PLOS Mental Health by Gallen Mlenge, Odass Bilame, and Lutengano Mwinuka, draws on the fifth wave of the Tanzania National Panel Survey, conducted in 2020 and 2021. The panel survey is a household-based instrument designed to be representative of the national population, which makes it a rare and valuable window into health behaviour in a low-income setting where routine administrative data on care-seeking is often incomplete. From the survey, the researchers analysed a sample of 11,891 individuals aged 15 years and above who had non-missing outcome data, making it one of the largest attempts to characterise healthcare utilisation patterns in the Tanzanian population.

One of the central methodological challenges the authors faced is familiar to anyone working on mental health in low- and middle-income countries: population surveys rarely include clinical diagnostic instruments. Instead, the team used a proxy measure. Mental health need was identified through the Washington Group Short Set item on difficulty remembering or concentrating, applied at the severe threshold. This item is a recognised proxy associated with common mental disorders, and while it cannot substitute for a clinical interview, it provides a consistent, internationally comparable indicator that can be embedded in large multipurpose surveys. Using this definition, 54 individuals in the sample were classified as having proxy-identified mental health need, corresponding to a weighted prevalence of 0.45 percent of the adult population.

The outcome measure was deliberately broad. The researchers counted any visit to a formal provider in the four weeks preceding the interview, regardless of the reason for the visit. This captures the overall propensity to engage with the formal health system rather than care-seeking for a specific condition. Overall, weighted formal healthcare utilisation stood at just 19.2 percent, meaning that fewer than one in five Tanzanians aged 15 and above had contacted a formal provider in the previous month. In a country where the burden of disease includes both communicable and increasingly non-communicable conditions, that figure underscores how much of the population remains outside the reach of routine formal care.

To explain who used care and who did not, the authors turned to the Andersen Behavioural Model, a long-standing framework in health services research that organises determinants of utilisation into predisposing characteristics, enabling resources, and need factors. They estimated survey-weighted logistic regressions producing adjusted odds ratios and predicted probabilities. The results revealed a system that sorts people by demography and geography rather than by need. Female sex, older age, specifically 55 years and above, and residence in the Southern Highlands zone were each independently associated with higher utilisation, suggesting that both biological and structural factors shape contact with the health system.

The most striking finding, however, concerned household wealth. Utilisation rose in a strong dose-response gradient across wealth quintiles: the richer the household, the more likely its members were to have visited a formal provider. Individuals in the richest quintile had 2.71 times higher adjusted odds of utilisation compared with those in the poorest quintile. In other words, the ability to pay, and the enabling resources that come with it, such as transport, proximity to facilities, and the capacity to absorb the indirect costs of seeking care, appear to be among the dominant gatekeepers to formal healthcare in Tanzania. This gradient is a direct challenge to the principle of equitable access that underpins universal health coverage ambitions.

Against this backdrop, the finding on mental health need is quietly devastating. Proxy-identified mental health need was not significantly associated with formal healthcare utilisation. The adjusted odds ratio was 1.27, with a 95 percent confidence interval spanning 0.53 to 3.04 and a p-value of 0.588. While the point estimate leans in the direction of slightly higher utilisation among those with identified need, the interval is wide and crosses unity, meaning the data cannot distinguish the effect from no effect at all. People whose survey responses signalled severe difficulty remembering or concentrating were, statistically, no more likely to have seen a formal provider in the previous month than anyone else.

The authors were careful about the limits of that subgroup. Only 54 individuals carried the proxy-identified need flag, and missing outcome data affected 505 individuals in a pattern classified as Missing Not At Random, meaning that whether someone’s outcome was missing was related to the outcome itself. To probe how fragile the subgroup estimates might be, the team conducted an extreme bounds sensitivity analysis. The result was a stark illustration of statistical uncertainty: true utilisation in the subgroup could plausibly range from 2.0 percent to 92.7 percent. The authors therefore describe their findings for the mental health need subgroup as exploratory, a candid caveat that is rare and welcome in this literature.

That honesty matters, because the temptation in global mental health research is often to overread thin data. Here, the study does two things at once. It establishes with reasonable confidence the broad structural determinants of care-seeking in Tanzania, since the full sample of nearly 12,000 people provides substantial statistical power for the wealth, sex, age, and zone effects. At the same time, it demonstrates precisely why mental health need remains nearly invisible in household survey data: the affected group is small, the proxy is coarse, and the missingness is informative. The study is as much a methodological lesson as a substantive one, showing what national panels can and cannot reveal about the health-seeking behaviour of people with mental health conditions.

The policy implications, nonetheless, point in a clear direction. If wealth is the strongest predictor of who reaches a doctor, then removing financial barriers, through mechanisms such as fee exemptions, insurance expansion, or support for indirect costs, should sit at the centre of efforts to make healthcare utilisation more equitable in Tanzania. And if residents of some geographic zones systematically lag behind, then strengthening service availability in lower-utilisation regions becomes an equally urgent priority. For the mental health community specifically, the study suggests that integrating mental health need into general health system planning will require better measurement, larger samples, and data systems that do not lose the very people whose outcomes are hardest to capture. Until then, the quiet finding that mental health need does not move the needle on care-seeking in Tanzania stands as both a data point and a warning about how much the health system, and the surveys that measure it, still miss.

Subject of Research: Determinants of formal healthcare utilisation among adults with and without proxy-identified mental health need in Tanzania

Article Title: Patterns and determinants of formal healthcare utilisation among individuals with and without proxy-identified mental health need in Tanzania: A cross-sectional analysis of the National Panel Survey

Article References: Mlenge, G., Bilame, O., & Mwinuka, L. (2026). Patterns and determinants of formal healthcare utilisation among individuals with and without proxy-identified mental health need in Tanzania: A cross-sectional analysis of the National Panel Survey. PLOS Mental Health, 3(9), e0000644. https://doi.org/10.1371/journal.pmen.0000644

Image Credits: AI Generated

DOI: 10.1371/journal.pmen.0000644

Keywords: Tanzania, healthcare utilisation, mental health, health equity, household wealth, National Panel Survey, Washington Group Short Set, Andersen Behavioural Model, low- and middle-income countries, cross-sectional analysis, health policy, survey methods