Optimal understanding of the longitudinal course of homelessness requires research methodologies for successful tracking of this difficult-to-follow population (McKenzie et al., 1999). It is important to enroll and retain participants in research populations to be able to generalize findings to the populations they represent. Challenges of tracking marginalized populations over time are well recognized (Bindman, Gmmbach, Keane, & Lurie, 1993; Conover et al., 1997; Kleschinsky, Bosworth, Nelson, Walsh, & Shaffer, 2009; Pollio, Thompson, & North, 2000), and tracking is complicated by substance use, psychiatric illness, and socioeconomic deprivation that are inherent in this population (North, Eyrich, Pollio, & Spitznagel, 2004).
A recent review by McKenzie, Tulsky, Long, Chesney, and Moss (1999) described 10 strategies for successful tracking of marginalized populations, emphasizing multimodal tracking methods and team organization and training. Three noteworthy studies have obtained successful tracking rates of 75% over 12 months (Wright, Allen, & Devine, 1995), 82% over 24 months (Hough, Tarke, Renker, Shields, & Glatstein, 1996), and 83% over 12 months (Cohen et al., 1993). Conover's group (Conover et al., 1997), following an ethnographic approach, achieved virtually complete follow-up of homeless mentally ill men over 18 months, but this effort was part of a long-term intervention project designed for continuous participant engagement.
Most studies have focused on details of tracking methods to improve tracking success. Few have examined individual characteristics that predict tracking difficulties, which might help identify people who need more extensive tracking efforts. Almost nothing is known about characteristics that predict the likelihood of tracking members of the homeless population. The identification of who might be more difficult to track, as well as how they might best be tracked, can substantially improve the ability to track more effectively and more efficiently.
This article describes a longitudinal study of a sample of 400 homeless people who were followed over 24 months using a multimodal tracking methodology as part of a comprehensive longitudinal study of homelessness and service utilization. Baseline information on this sample, reported elsewhere, found that the majority have lifetime substance use disorders (84% among men and 58% among women), which were far more prevalent than any other psychiatric disorder (North et al., 2004). This article's goals are to further explicate the necessity and practicality of extensive tracking methods and to elaborate on the specific factors that predict successful tracking and methods of tracking.
An epidemiologic sample of homeless individuals participated in a National Institute on Drug Abuse (NIDA)--funded longitudinal study designed to examine homelessness and prospective housing, service use, and other outcomes over two years. This project was approved by the Washington University School of Medicine Institutional Review Board. All participants provided written informed consent before entering the study.
Individuals who were literally homeless were eligible to participate. For this study, homelessness was defined as having no current fixed address of one's own and having spent the previous 14 nights in a public shelter or in some other unsheltered location without a personal mailing address such as a park, an abandoned building, a car, or on the streets. In addition, those who had spent the previous 14 days in very inexpensive transient lodging were included if they had been there for 30 days. Individuals who had stayed in a public shelter or unsheltered location for most of the previous 14 days were included if they had stayed less than one-half of those days temporarily with friends or relatives or in temporary single-room-occupancy facilities. Recruitment to the study was conducted over approximately one year (February 2000 through March 2001) using methods previously developed by this research team to provide representative sampling (80% from shelters and 20% from street locations) of the St. Louis homeless population (North, Eyrich-Garg, Pollio, & Thirthalli, 2010; Smith, North, & Spitznagel, 1992, 1993).
To recruit 400 participants, we invited 435 eligible homeless individuals to participate (92% participation). The baseline demographics of the study sample are presented in Table 1. Participants were contacted quarterly and were asked to check in with the project monthly (by calling project staff or stopping by the project offices) over 24 months. Participants were offered financial remuneration of $25 for annual interviews, $1 per monthly check-in, and $5 per quarterly interview. Individuals who completed their appointed contacts were entered in monthly and annual lotteries of $20 and $100, respectively.
To maximize opportunities for project staff to contact participants at a later time, we anchored the tracking process with a baseline contacts form, which was used to collect extensive locating information. It was updated quarterly and annually to provide the full name along with any aliases or nicknames; birth date and place; best mailing address; usual sleeping and eating locations; favorite hang-outs; employer information if applicable; full contact information for respondents' nuclear family members, significant others, friends and acquaintances, professional contacts such as primary doctor or other health care provider, lawyer, caseworker, and probation or parole officer; self-help participation; other service involvement; and driver's license and Social Security number. Study participants provided written consent for the release of their personal contact information to allow project staff to contact the individuals who they had named on their contacts form as being able to assist with locating them. All study participants received a calendar card with follow-up information marked in red and contact numbers and addresses for the research project offices. A photo identification card...