Individuals experiencing homelessness are among the most medically, socially, economically, and politically vulnerable in our society. Because of this, patterns of service use by these individuals can often serve as indicators of the safety net capacity and overall wellness of the systems of care we have in place – like “canaries in the coal mine.” According to the latest estimates, about half a million people experience homelessness on a given evening, and they are substantially more likely to interact with the criminal justice and healthcare systems [PDF].
Homelessness is enormously disruptive to health and quality of life. Chronic health conditions are more prevalent, with premature onset; there is greater exposure to infectious diseases and higher rates of multiple risk behaviors. This leads to a higher risk of mortality, with average life expectancy estimated to be around 30 years shorter [PDF] relative to housed comparison groups.
In an article published in the January 2019 issue of Medical Care, Dr. Rishi Wadhera, of the Richard and Susan Smith Center for Outcomes Research at Beth Israel Deaconess Medical, and colleagues characterized the healthcare system interactions of individuals experiencing homelessness using hospital data from three states over 7 years (2007-2013). They analyzed the trends in hospitalizations, the primary causes, and the characteristics of hospital admissions.
Going Back in Time
To get a better idea of what they were looking at, let’s travel back in time to just before the Affordable Care Act (ACA) was implemented. After all, we can’t know how to interpret the effect of that policy without knowing the landscape into which it was introduced. The economy in 2007 was, to put it mildly, on the rocks. The bottom had dropped out of the housing market and banking system, and the government was opening up the coffers to bail out big companies and communities alike. The HEARTH Act was passed in 2009 in an effort to reform federal housing programs and help care for those hit hardest by the recession. And in terms of health reform, Massachusetts led the nation in passing universal health coverage legislation, which went into effect in 2006. Modeled on that legislation, the ACA was signed in 2010 and fully implemented in 2014.
Wadhera et al. obtained hospital admission records (from 2007 to 2013 for Massachusetts and Florida, and 2007-2011 for California) from the State Inpatient Databases (SID). They included 185,292 admissions by individuals (aged 18 and over) experiencing homelessness and 32,322,569 admissions by non-homeless individuals matched on demographics and payer status. Matching was used in order to isolate the relationship between homelessness and hospitalizations from the influence of those characteristics.
What they found was a system in a deep state of distress. Among people experiencing homelessness, rates climbed by 23-49% over the 5 year observation period. The largest causes of admissions were problems of either mental health or substance abuse. In fact, these issues occurred nearly 3 times(!) as often (52 vs 18%) relative to matched non-homeless admissions.
The differences in the nature of these hospital stays were curious as well. Hospitalizations for individuals documented as homeless had lower mortality rates, lower costs of care per day and per admission, and yet they had longer lengths of stay. Fundamental to understanding all of this, almost 42% of the cohort experiencing homelessness (and also the matched, housed cohort) were uninsured during this period.
There were a few limitations involved with the findings, which the authors were open about. They couldn’t do much to characterize the nature or extent of homelessness episodes in their cohort. Nor could they determine veteran status, which is a major driver of access to healthcare. Hospitalization rates and cost-of-care estimates also excluded utilization behaviors previously documented in homeless cohorts such as over-use of ED visits and EMS transports [PDF] and the associated increased costs of care.
This research reflects the situation in just 3 states using just 5-7 years of data per state. It’s hard to know if the changes observed were part of any longer term trends or cyclical, capturing normal peaks and valleys. It’s also unclear how well the homelessness variable captured the population of interest. In such a complex system, the issue of attribution to specific external factors is extraordinarily challenging. That said, I for one look forward to seeing how circumstances have changed since the ACA began to seriously influence the landscape (starting in 2014).
Explaining the Findings
The dynamic relationships that drive healthcare utilization by those in homelessness are convoluted beyond belief. The unique utilization patterns associated with homelessness are a result of limited access to healthcare, health insurance options, heightened medical complexity and need, and competing priorities related to their environment and living conditions (just to scratch the surface).
One explanation for the increased hospitalization rates may be that barriers to accessing inpatient care were already lowering over the period before the ACA. Particularly in Massachusetts, it’s possible that increased health insurance coverage and safety net systems led to the higher treatment rates seen over the course of this study. It also seems possible that primary care shortages pushed individuals into tertiary care. Either way, hospitalizations among those experiencing homelessness were clearly on the rise.
Mental health and substance abuse drove a lot of hospitalizations. Hospital admissions for these kinds of problems are largely avoidable with access to consistent, reliable outpatient behavioral health services. But the supply of behavioral health care in the US was greatly outmatched by demand [PDF] during this same time. These access barriers and under-treatment of behavioral health conditions likely increased unnecessary admissions even higher.
The lower acuity of patient presentations (as evidenced by lower inpatient mortality rates) likely led to lower costs per day and per admission. However, the socioeconomic determinants and medical complexity of these patients and the lack of community resources to which they could be discharged led to longer stays. This effect in homeless hospital patients has been previously reported by Hwang et al. in the April 2011 issue of Medical Care.
While it’s not clear what effect health reforms had on the trends observed by this study, let’s hope that policy can help relieve some of the pressures which this research brings to light.
Ben King