A Quantitative Review of Cognitive Functioning in Homeless Adults

Abstract

Aims of the Study

Approximately 1.5 million people in the United States experience homelessness in a year’s time (AHAR, 2012). Cognitive impairment is well established as a core contributor to occupational, social, and instrumental functional disabilities in serious mental illnesses and substance abuse, and, accordingly, is now considered a primary treatment target for these illnesses (Harvey, 2010; Kalechstein & Van Gorp, 2011). A number of studies have suggested that cognitive impairment may be highly prevalent among homeless adults (Backer & Howard, 2007; Burra, Stergiopoulos, & Rourke, 2009; Seidman et al., 1997; Spence, Stevens, & Parks, 2004). It is unclear whether cognitive impairments predate and contribute to becoming homeless or whether homelessness causes cognitive impairment. Cognitive impairment in homeless people may stem from multiple causes, including neurodevelopmental problems, such as intellectual disabilities and psychiatric disorders, or acquired brain dysfunction due to traumatic brain injury (TBI), substance use disorders, stroke, dementia, epilepsy, or other neurological conditions. It is also likely that aspects of homelessness, such as stress, malnutrition, and environmental exposures may directly impair brain function. Regardless of etiology, it is likely that planning, organization, and recall deficits interfere with the tasks necessary to sustain income to support housing, as well as in navigating and participating in preventative services or rehabilitative efforts toward reintegration once homeless. Thus, cognitive impairment may be an underappreciated barrier to efforts toward ending homeless.

There have been prior critical literature reviews of the role of cognitive impairment in homelessness that identified a small pool of studies that generally recruited convenience samples from community shelters. (Backer & Howard, 2007; Burra et al., 2009; Spence et al., 2004). These reviews were in consensus that, given variation and limitations in sampling and methods, observed frequencies of cognitive impairment were higher than those in age-comparative groups. However, none of these reviews employed a quantitative approach, and so it is currently unknown from this body of literature what the estimated frequency of cognitive impairment are among homeless people, nor the nature of the samples with respect to demographic and clinical variables (e.g., age, duration of homelessness). Moreover, additional studies published since 2009 have not been incorporated into a review. We conducted a quantitative review of 24 identified studies that reported results from objective cognitive testing in samples of people identified to be homeless. In addition to estimating the frequency of cognitive impairment in this population, we also obtained estimates of sample characteristics and clinical factors to better understand the risk factors for cognitive impairment in the broader homeless population.

Materials and Methods

Results

Discussion

In this quantitative review of homelessness and cognitive impairment, we found evidence for elevated frequencies of cognitive problems in homeless samples. We also identified a number of important gaps in the literature. In the pooled sample of 2969 homeless adults, about 25% screened positive for cognitive impairment. The mean full-scale IQ score was about one standard deviation below average, which would be below the 15^th percentile. Furthermore, of the studies that used neuropsychological assessments, mean scores were almost universally below average and were more commonly in the impaired range, suggesting that there may be considerable cognitive impairment in this population that is missed by brief screening measures (e.g., the MMSE). These estimates confirm, in a larger pool of studies, the conclusions of earlier reviews that cognitive impairment is likely common among homeless adults (Backer & Howard, 2007; Burra, Stergiopoulos, & Rourke, 2009; Seidman et al., 1997; Spence, Stevens, & Parks, 2004), and should be an important element of routine screening and rehabilitation.

There was substantial variation across studies in the frequency of cognitive impairment, and this review identified some potential sources of this variation and areas of methodological improvement and need for greater representation of the homeless population. For one, variation in the assessment and quality of cognitive assessment was present. Understanding that comprehensive neuropsychological batteries are not compatible with routine care in most service settings, there are more reliable and valid screening measures than the MMSE for cognitive impairment (e.g., the Montreal Cognitive Assessment (Nasreddine et al., 2005). Additionally, performance-based measures of functional capacity (e.g., the UCSD Performance-based Skills Assessment [UPSA(Patterson, Goldman, McKibbin, Hughs, & Jeste, 2001)]) may be better able to capture impairments in instrumental living skills. Functional Capacity measures (e.g., direct observation of instrumental activities such as financial management) are highly associated with cognitive ability in a wide variety of illnesses; one study indicated that the UPSA was sensitive to housing status (Stergiopoulos, Burra, Rourke, & Hwang, 2011).

Few studies were longitudinal and it is likely that the pathways to cognitive impairment in this population are as diverse as are those to becoming homeless. Neuropsychiatric risk factors were highly prevalent in the population, as one-fifth of sampled patients had a history of psychotic disorder and one-third had a history of affective disorder. Over half had a history of substance abuse. These estimates may underestimate lifetime psychiatric burden, given that some studies reported only current diagnosis. In addition, over half of individuals sampled reported a lifetime history of a head injury, which was variously defined. It is unclear which of these risk factors was associated with the greatest elevated risk for cognitive impairment, as studies rarely assessed the relative strength of association between putative risk factors and cognitive impairment. It is also unclear if these risk factors predated the onset of homelessness. Longitudinal research on the temporal sequelae of psychiatric, neurological, and substance abuse and homelessness would aid in determining the course of these risk factors in relationship to the onset of homelessness, duration of homelessness, and response to rehabilitative efforts. It may be likely, given the diversity of risk factors, that subgrouping based on risks (e.g., neurodevelopmental risks such as schizophrenia or intellectual disability vs. acquired ones such as traumatic brain injury or substance use disorder) may yield a more clinically informative platform for risk stratification.

Future work should also broaden the representation of homeless adults. Although varied in terms of nation of origin of the study, the samples were nearly all from residents in sheltered settings. Thus, unsheltered adults were not typically represented in these studies, likely due to the problems in recruitment that arise in such persons. Moreover, about 28% of approached persons did not participate in data collection, so the cognitive profile of people who elect not to participate in research or cognitive screening is unknown. Importantly, greater than 83% of the samples were men, yet recent estimates indicate that 38% of sheltered homeless people are women (AHAR, 2012).

Additional potential risk factors that were not well addressed by the literature include learning/intellectual disabilities and other diagnoses such as PTSD. It was notable, in the few studies that estimated pre-morbid IQ, the mean score was below average (m=90) and near that of full-scale IQ. This finding hints at the presence of pre-morbid and perhaps developmental deficits in a portion of the homeless population. Moreover, most of the studies in the review reported on data collected in the 1990s and early 2000s, prior to the current expansion of focus on PTSD. Given the recent wave of Veterans returning from foreign conflicts, additional surveillance systematically capturing the co-occurrence of history of traumatic injuries, homelessness, and cognitive impairment would be particularly timely.

This review was limited in providing a comprehensive picture cognitive impairment in homeless people due to the relatively small number of studies and the inconsistencies across studies in data collection strategies, definitions, and reporting of homelessness, diagnoses and cognitive ability. As such, we were not able to examine systematically sources of variation of estimates of cognition and were not able to weight our estimate by study quality. Moreover, given the preponderance of cross-sectional studies, we could not determine the direction of the association between homelessness and cognition. Nonetheless, our review indicates that cognitive impairment is a common aspect of the complex clinical picture of homeless adults. The identified average prevalence of cognitive impairment was about 5 to 8 times greater than the rate of cognitive impairment in the U.S. population older than age 70 (the mean age across the samples was 46.1 years) (Sheffield & Peek, 2011). Future research would need to identify the extent to which homelessness exerts additional risk to cognition beyond psychiatric illnesses (e.g., psychosis or substance abuse), and whether cognitive impairment reduces the impact of housing-focused rehabilitation as it does in other rehabilitation targets (Kurtz, 2011; McGurk & Mueser, 2013; Sofuoglu, DeVito, Waters, & Carroll, 2013). As such, we hope that this review can stimulate research on the role and rehabilitation of cognitive impairment in the persistent problem of homelessness.

Acknowledgments

Footnotes

References