A research study looking at services for people who are homeless in New York has been receiving a lot of press. In a nutshell, the program looked at randomizing homeless families to a new model of a centralized hub of services, or to usual care. You can see my comments on why a research ethics board might support this program here and here. Strong arguments against the project are here, here, or here. In a nutshell, the ethical question comes down to whether there is reasonable grounds to believe that we already know that the intervention (in this case a program called HomeBase) will work. If we know this, then it is unethical to withold it from anyone in the name of research. I would say that the jury is clear, and that this study is indeed unethical.
Here’s the issue, we only have limited funds to work with, so we want our programs and interventions to be effective as possible. One powerful way to determine effectiveness is to randomly assign people to different programs, and monitor the outcomes. This is the basis by which all medications are trialed, and this type of research is very good at answering certain questions. However, in the real world it’s tougher to justify a control group, as any service is likely to be more effective than usual care for people who have been marginalized in the most absolute sense, who are homeless.
It was with this mindset that I heard Martha Burt of the Urban Institute suggest at the “Rethinking Homelessness” conference in Montreal that randomized control trials are necessary to assess homelessness prevention programs. I approached her afterwards and asked her about the ethics of this, and I think she convinced me of one case in which randomization to services with homeless people is ethical.
Martha suggested that in the vast majority of cases for people who apply to homelessness prevention programs, without any assistance they would not become absolutely homeless. Most people find ways to avert homelessness on their own, and for only a small percentage, this doesn’t work. So, the challenge with homeless prevention is that it’s hard to know if you are simply preventing homelessness with wouldn’t have happened regardless. In this case, randomizing people to usual care and an intervention becomes ok ethically, as usual care is enough to meet the needs of most people.
However, I could be wrong, so argue with me on this one.