Does community health care cut hospital use? Australian evidence says yes

One of the most persistent promises in health policy is that investing in community health services will keep people out of hospitals. If primary and community-based care can catch illness early, manage chronic conditions properly, and support vulnerable populations before they deteriorate, the argument goes, then emergency departments should see fewer walk-ins and hospital wards should admit fewer patients. A new study from Australia puts that promise under the microscope, and the results are more complicated, and in some respects more unsettling, than the conventional wisdom suggests.

Researchers from the Melbourne Institute of Applied Economic and Social Research at the University of Melbourne, together with colleagues from the Victorian Department of Treasury and Finance and Monash University’s Centre for Health Economics, have conducted one of the most rigorous assessments to date of how community health services affect hospital utilisation in Australia. Their analysis, published in BMC Health Services Research, focuses on the Victorian Community Health Program, a state-run network of services designed to reach priority populations at risk of poorer health under a distinctly social model of care. Unlike systems aimed at the general population, Victoria’s program deliberately targets people facing disadvantage, offering a broad mix of nursing, allied health, dental, counselling and health promotion services in community settings.

The central question the team asked is deceptively simple: do people who use community health services end up using hospitals less? The answer, drawn from an unusually rich trove of linked administrative data, turns out to be no, at least not in the way policymakers might hope. Community health users did experience fewer unplanned hospital readmissions, a genuinely positive signal that suggests continuity of community-based care may help some patients avoid the revolving door of returning to hospital shortly after discharge. But on almost every other measure of hospital use, community health users fared no better, and on several they fared measurably worse.

Compared with matched non-users, people who accessed the Community Health Program had longer average lengths of stay in hospital, more potentially preventable admissions, more admissions involving hospital-acquired complications, and more presentations to emergency departments. There was no significant difference between users and non-users in the total number of hospital admissions. In other words, the widely assumed substitution effect, in which community care simply displaces hospital care, did not emerge in this setting.

The methodological architecture of the study is what makes these findings worth taking seriously. Because people who use community health services are not randomly selected, they tend to be sicker, poorer, and more medically complex than the general population, a naive comparison of users and non-users would be hopelessly confounded. To address this, the researchers employed Propensity Score Matching, a statistical technique that estimates, for each community health user, the probability of being a user given observable characteristics, and then pairs each user with a non-user who had a comparable probability of using the program. This approach effectively constructs a comparison group that resembles the user population on observed dimensions, allowing the analysts to isolate the association between program access and subsequent hospital utilisation with greater confidence.

The data infrastructure underpinning the analysis is equally notable. The team extracted and linked three separate administrative datasets: the Community Health Minimum Dataset, which records community health service contacts; the Victorian Admitted Episodes Dataset, capturing details of hospital admissions; and the Victorian Emergency Minimum Dataset, which logs emergency department presentations. Linking these sources at the individual level allowed the researchers to follow people across the interface between community care and hospital care, a boundary where data fragmentation has historically made evaluation extremely difficult. The study emerged from research commissioned by the Victorian Department of Treasury and Finance specifically to test whether such individual-level administrative data could feasibly be used to evaluate the effects of community health on hospital utilisation and health outcomes.

International evidence on the health benefits of community health services is generally positive, but the effect on reducing hospitalisations has been mixed, often depending heavily on specific settings, institutions, and the populations served. Prior Australian evidence has been limited, which makes the new study a meaningful addition to a sparse literature. The authors are careful, however, about what their results can and cannot support. The mixed effects they observe may reflect previously unmet health needs among community health users rather than a failure of the program itself. People who finally gain access to community services after long periods of neglect may be sicker at baseline than their matched counterparts appear on paper, and increased hospital contact could represent appropriate care finally reaching people who needed it all along.

This interpretation carries important implications for how the findings should be read. A longer length of stay among community health users could indicate more complex clinical presentations, not inefficient care. More potentially preventable admissions might signal unmanaged conditions that the program is only beginning to address, or conversely, that community services are successfully identifying problems that hospitals previously absorbed silently. More emergency department presentations could reflect better health literacy and a greater propensity to seek help, or genuine unmet need. The data alone cannot adjudicate among these explanations, and the authors are candid about this uncertainty.

Indeed, the study’s limitations are as instructive as its findings. The researchers identify the lack of data on primary care use as a key constraint. Without knowing how often community health users visit general practitioners, it is impossible to distinguish whether community health services complement, substitute for, or operate independently of mainstream primary care. Similarly, the absence of data on health improvements that allow hospital avoidance, and on changes in patient well-being, means the analysis captures only the utilisation side of the ledger. A program could reduce hospital use substantially while leaving patients worse off, or increase hospital contact while dramatically improving quality of life. Measuring utilisation alone cannot tell which story applies. The authors state plainly that access to additional linked data is necessary before these results can meaningfully inform policy decisions.

The ethical and procedural context of the work is also worth noting. The study involved secondary analysis of de-identified, pre-existing data and received ethics approval from the Melbourne University Human Research Ethics Committee. It was conducted in accordance with the National Statement on Ethical Conduct in Human Research issued by Australia’s National Health and Medical Research Council. The funding agency, the Victorian Department of Treasury and Finance, played no role in the design of the study, the analysis of the data, or the interpretation of the results, according to the authors’ competing interests declaration.

For health systems around the world grappling with aging populations, rising chronic disease burdens, and strained hospital budgets, the Australian findings land at an uncomfortable moment. Many governments have pinned cost-containment hopes on shifting care out of hospitals and into communities. The Victorian evidence suggests that this shift, at least as implemented under a social model of care targeting disadvantaged populations, does not mechanically translate into reduced hospital utilisation. That does not mean community health programs are failing; it may mean their value lies elsewhere, in equity of access, in early intervention, in improved well-being, in fewer readmissions, dimensions that hospital utilisation statistics alone cannot capture.

The finding of reduced unplanned readmissions deserves particular emphasis. Readmissions are among the most scrutinised quality indicators in health services research, often signalling failures in discharge planning and post-acute support. If community health services genuinely help patients stay out of hospital after discharge, that is a concrete, measurable benefit with both clinical and financial significance, even if other utilisation metrics move in the opposite direction.

What the study ultimately demonstrates is the power and the peril of rigorous evaluation. For decades, community health programs have been funded on the strength of intuitive arguments and selective evidence. When researchers finally link comprehensive administrative data and apply careful matching methods, the picture that emerges resists easy narratives. The Victorian Community Health Program appears to deliver some benefits, notably fewer unplanned readmissions, while being associated with greater hospital contact along other dimensions, plausibly because it serves a population with substantial unmet need. Disentangling whether those associations reflect the program’s effects, the population’s underlying health trajectory, or the interaction between the two is precisely the research agenda the authors say must come next.

Until that research is done, the study stands as a caution against assuming that community investment automatically pays for itself through hospital savings, and as a demonstration that linked administrative data, used carefully, can force uncomfortable questions into the open. For Victoria, and for any health system betting on community care to relieve hospital pressure, the message is clear: the relationship between community health and hospital utilisation is not a simple trade, and the evidence to guide these high-stakes decisions is only beginning to be built.