Perth cohort exposes a stark mortality gap among people who have experienced homelessness

A dynamic cohort study from Perth, Western Australia, documents a mortality burden that is both measurable and deeply troubling. Among 8,753 people who experienced homelessness and engaged with local services between 2016 and 2022, investigators identified 360 deaths over a three-year period from 2020 through 2022. The median age at death was 50 years, more than three decades younger than the current Australian median age at death of 82 years.

The numbers are not just poor. They point to a sustained failure of health and social systems to keep pace with the risks faced by people who cycle through homelessness services, shelter, and unstable housing. A mortality profile this young usually signals a dense cluster of untreated chronic disease, mental illness, substance use, exposure-related injury, infectious disease, and delayed access to care. In this cohort, those pressures are visible in the age at death alone.

Multiple data sources were used to make deaths visible

This study was designed around a problem that often keeps homeless mortality out of public view: deaths are easy to miss when people move frequently, lack stable records, and interact with fragmented services. Investigators used multiple data sources in line with the Homeless Mortality Data Toolkit approach, then routinely captured, verified, and reported deaths among people who had experienced homelessness in Perth.

That methodological choice matters because a single registry or service dataset would almost certainly miss people who die outside a hospital, outside a known service setting, or after losing contact with support systems. The study therefore reflects a practical epidemiologic response to a known surveillance blind spot. Even with that approach, the 360 deaths identified over three years were described as likely an undercount.

This is the kind of undercount that weakens policy response. If mortality is incomplete, the size of the problem is easier to dismiss, and accountability becomes diffuse.

Premature death was concentrated far below the national norm

The comparison with the national benchmark is striking. A median age at death of 50 years in this cohort sits 32 years below Australia’s current median age at death of 82 years. For clinicians, that gap suggests not an isolated excess of deaths, but a structural pattern of premature mortality that begins decades earlier than in the general population.

The burden is unlikely to be driven by one diagnosis alone. Homelessness amplifies risk across nearly every organ system and every care setting: cardiovascular disease becomes harder to monitor, diabetes management becomes unstable, mental health care is interrupted, and acute illness often presents late. When a person lacks housing, even basic prevention becomes difficult. Medication storage, wound care, nutrition, sleep, and follow-up all become unstable variables.

A 50-year median age at death reflects those cumulative failures in a single statistic.

Aboriginal people accounted for nearly one-third of deaths

Aboriginal people accounted for 30% of the deaths identified in the cohort. In the context of homelessness, that proportion signals the intersection of two long-standing inequities: the overrepresentation of Aboriginal people in housing instability and the broader health gap driven by colonization, disadvantage, and barriers to culturally safe care.

The study does not provide more granular cause-of-death data in the abstract, so the specific mechanisms behind these deaths cannot be inferred from the report alone. Even so, the overrepresentation is clinically and socially important. Mortality surveillance that does not capture Aboriginal status accurately will understate need, obscure service gaps, and weaken targeted responses.

In practical terms, better mortality tracking is not just a data exercise. It is part of identifying where culturally safe outreach, continuity of care, and coordinated housing support are falling short.

Why contemporaneous mortality tracking matters for policy and care

The study’s central argument is straightforward: homeless deaths must be tracked in real time if systems are expected to reduce them. Delayed data collection makes a preventable death look like a historical footnote instead of an actionable event. Timely verification allows health services, homelessness organizations, and government agencies to monitor trends, identify clusters, and measure whether interventions are working.

The biological and social logic are aligned here. Premature death among people experiencing homelessness is rarely the result of a single catastrophic event alone. More often, it emerges from accumulated exposure to harsh environments, poor access to primary care, missed diagnoses, delayed treatment, and the erosion of protective routines. Mortality surveillance is one of the few tools capable of capturing that accumulated harm at a population level.

The study also frames death registration as a form of public accountability. Without verified counts, a system can claim concern while remaining blind to the scale of preventable loss.

The undercount problem remains central

Even with triangulated data sources, the authors judged the 360 deaths to be likely underestimated. That caveat is important because it means the true mortality burden may be higher than the already alarming estimate. Underascertainment is common when people are disconnected from conventional administrative systems, and it is especially likely in mobile populations with intermittent service contact.

For clinicians and policymakers, the implication is blunt. If current systems cannot reliably count deaths among people who have experienced homelessness, they cannot credibly claim to be monitoring equity or measuring improvement.

The study does not offer a cause-specific mortality breakdown in the abstract, so the exact composition of these deaths cannot be stated here. What is clear is the scale of premature mortality, the concentration of death well before old age, and the disproportionate impact on Aboriginal people.

A concrete benchmark for future action

The most useful aspect of this cohort is not simply that it counted deaths. It established a repeatable framework for tracking them over time. That creates a baseline against which future mortality trends can be compared, and it gives public health agencies a target for intervention.

Targets should be set to reduce the observed three-decade life expectancy chasm. Without numeric goals, mortality in homeless populations can remain a persistent background harm rather than a measurable failure demanding response. The study offers a clear standard for surveillance, but it also leaves a hard clinical and ethical question in place: whether health systems and governments will use these data to change service access, housing policy, and continuity of care before more lives are lost.


Reporting by Dr. A.R. Athira, MBBS for Medicine World Foundation, with additional data provided by BMJ Open.