We have known for years that homelessness is associated with premature death. Now a study commissioned by the Centre for Homelessness Impact, for the first time on anything like this scale, gives us a much clearer measure of how great that risk is, where it is greatest and what is driving it.

It is a systematic review and meta-analysis, which brings together 116 studies involving more than 130 million people in total, including very large comparison populations. For the analysis of deaths from all causes alone, it draws on data from more than 2.5 million people who had experienced homelessness and more than 109 million people in comparison populations. Across that evidence, people who had experienced homelessness faced more than twice the risk of death, and more than nine in ten of the individual estimates pointed in the same direction.

The timing could hardly be more relevant. Andy Burnham, the new Prime Minister, has made ending long-term rough sleeping across the UK at the earliest opportunity one of his first commitments in office. This study strongly supports that ambition: people who have slept rough have the highest mortality risk of any group examined. But it also shows why the response cannot stop there, because the mortality inequality does not begin and end on the street.

Rough sleeping carries a pooled mortality risk more than seven times that of comparison populations, but elevated mortality is also found among people using shelters and low-cost hotels, while the association is similarly severe among men and women. That wider picture matters because the homelessness we see most readily is not the only homelessness carrying profound risks to health and life.

I have seen before how a single statistic can change the way an issue is understood. While I worked at the homelessness charity Crisis, I commissioned the research that became Homelessness Kills, led by Bethan Thomas at the University of Sheffield.

Published in 2012, the study found that the average age at death among the people experiencing homelessness identified in the research was 47, falling to 43 for women. Those figures became some of the most widely quoted statistics in homelessness, and for good reason: they made the human consequences of homelessness impossible to ignore.

But they are also often asked to do more than they can. They describe the ages at which people in that study died; they are not estimates of life expectancy, and they do not tell us how much greater the risk of death is for someone experiencing homelessness than for someone who is not.

This systematic review and meta-analysis gives us a much stronger basis for answering that question. By bringing together studies with comparison populations and synthesising the most comprehensively adjusted estimates, it allows us to quantify the excess mortality associated with homelessness, compare different experiences of homelessness and examine where the greatest cause-specific inequalities lie.

So the figures for the average age of death of 47 for men and 43 for women still matter as part of the history of the evidence, but they should no longer carry the whole argument. This review gives us a broader and more rigorous account of the risk.

The cause-specific findings make that especially clear. Mortality is elevated for 33 of the 36 causes of death examined, with some of the largest inequalities associated with psychoactive-substance-use disorders, drug overdose, accidental injury and alcohol-related causes. For psychoactive-substance-use disorder the pooled risk is more than 21 times higher, for other accidental injury more than 13 times higher and for drug overdose more than 11 times higher. The authors note that some of the associations they found are of a magnitude rarely seen in modern epidemiology.

Those findings should change the questions we ask. Ending long-term rough sleeping is an essential ambition and this study gives us another compelling reason to pursue it, but success cannot mean waiting until homelessness has become prolonged and visible before the risks to health command our attention. We also need to ask whether those risks are being recognised earlier, whether people experiencing less visible forms of homelessness can get the treatment and care they need, and whether housing, health, mental-health and substance-use services are working together before people reach crisis.

The review cannot tell us that one intervention will prevent every premature death, nor should we ask it to. What it does give us is a much stronger foundation from which to act. The paper points to areas where there is already evidence to draw on, including opioid substitution treatment, naloxone and psychosocial interventions, alongside the need for better integration of health, housing and social services and an adequate supply of affordable housing.

When Homelessness Kills was commissioned, the aim was to make visible a reality that had received far too little attention. More than a decade later, this study allows us to see that reality with much greater clarity: how large the mortality inequality is, how widely it reaches and where some of the greatest risks lie.

The evidence has moved on. Our response now needs to move with it.

• Ligia Teixeira is Chief Executive of the Centre for Homelessness Impact 

‍