Police encounters are recurrent for public drunkenness.^19,20 In one study,³ 70% of homeless alcoholic men had a history of imprisonment.

Although treatment with detoxification and abstention (“detox”) is the best option from a health perspective, the likelihood of rehabilitation among among people both alcoholic and homeless is low.^6,21–25 Obstacles to sobriety include psychiatric illness, poor social support, lack of stable housing, duration of addiction and refusal of treatment.

Harm reduction is a policy to reduce the adverse health, social and economic consequences of substance use without requiring abstinence. Methadone maintenance treatment of opioid dependence, for example, is superior to detox in reducing heroin use and behaviours that increase the risk of HIV infection.²⁶ After an inquest into the freezing deaths of homeless alcoholic men, a pattern was noted of heavy alcohol consumption before shelter entry to achieve in-shelter abstinence, followed by early-morning alcohol-seeking to avoid the symptoms of withdrawal.²⁷ Despite the high incidence of people with chronic alcoholism dying homeless in cities worldwide, this population has been underrecognized in program development and in the clinical literature. In response, a managed alcohol harm-reduction program was developed for people with long-term homelessness and refractory alcoholism.

Study subjects were housed at the shelter in an area designated for MAP and were provided with beds and meals. The program employed a client care worker to supervise the participants, give aid with activities of daily living, help fill out applications for social benefits, accompany them to medical appointments and dispense regular medications. Participants were given up to a maximum of 5 ounces (140 mL) of wine or 3 ounces (90 mL) of sherry hourly, on demand, from 0700– 2200, 7 days per week. Medical care was provided 24 hours per day by nurses and 2 physicians associated with the project, with daily nurse and weekly physician visits. Medical records were kept on a secured online system developed by the Ottawa Inner City Health Project.²⁹

Program participants were enrolled into MAP and included for analysis with approval from the Ottawa Hospital research ethics board and police services. Inclusion required continuous program participation for at least 5 months by July 2002. Data for all 17 eligible subjects were included for analysis; no one left the program before 5 months or was excluded from the study. A consent and confidentiality statement was read to each person at program entry, and written consent obtained to access hospital and police records. The project was analyzed as a before-and-after study design. Charts from all 5 area hospitals were reviewed for 3 years before program entry and while participants were in the program for number of ED visits, ambulance use and diagnoses of trauma, seizures or intoxication at presentation. Hospital admissions and lengths of stay were recorded, as were blood-test markers of alcohol use. The police services' computerized database was accessed for each participant by name, date of birth and all aliases, and the number of police encounters recorded for the same period as for hospital records.

Descriptive statistics using the mean and standard deviation were used for normally distributed continuous outcomes. Average monthly rates of use of ambulance services, visits to hospital EDs, diagnoses, hospital admissions and police encounters were calculated for a 36-month period before and up to 24 months after program enrolment. Differences in outcome measures before and after program entry were assessed with the paired Student's t test. Average values for blood-test results were calculated for the 24 months before and during program enrolment for each participant, and analyzed for statistical differences.

For the 3 years before entry into the program, the total mean monthly number of ED visits by all participants was 13.5, a monthly mean per participant of 0.79 (Table 2). During the program, this number decreased to a group monthly total of 8.1 (mean 0.51 visits per participant), with a decreased mean monthly paired difference per subject of 0.28 (standard deviation 0.35, p = 0.004; Fig. 1). Use of ambulance services, hospital admissions and ED visits all showed a decreasing trend, as did diagnoses of intoxication, trauma and convulsion, although statistical significance was not attained. Police encounters decreased from a monthly mean of 18.1 for the group to 8.8 (p = 0.018).

When concentrations of blood markers of alcoholism recorded during the 2 years before enrolment were compared with those obtained during the program, differences in group averages were nonsignificant, as were individual paired differences (results not shown).

Three people declined to be interviewed about their drinking history, health and life satisfaction, and 3 others died before being interviewed. The remaining 11 participants all reported a markedly decreased consumption of beverage and nonbeverage alcohol, and most reported improved sleep, hygiene, nutrition and health. Paired data were available for 10 participants to compare the amount of alcohol they were consuming before program entry with the amount consumed during MAP (Table 3). Subjects noted that a typical day's consumption was difficult to estimate; most described drinking all the alcohol that was available and would drink until they lost consciousness. For all participants, the absolute amount of alcohol consumed was found to decrease, from an average of 46 drinks per day to 8 (p = 0.002). Ten participants who agreed to be administered the DSWL Scale scored a median of 22, consistent with feeling “slightly satisfied” with life.³⁰

The client care workers interviewed all noted improved hygiene and nutrition for all participants during the program. Compliance with medication, defined as taking it as prescribed at least 80% of the time, was noted for 88% of subjects. The majority were reported to attend scheduled medical appointments.

Police encounters decreased by 51% and ED visits by 36%, which, given the associated “unit encounter” costs ($93 and $270, respectively), offset a portion of the costs of MAP. Police encounters and ED visits were seen to increase for 2 subjects (Fig. 1), but both had been in jail or living in another province during the 2 years before MAP enrolment and their reports were not captured in the Ottawa system. Blood-test markers of alcohol use remained stable, and participants and client care workers reported improvements in health, nutrition and hygiene. Compliance with prescribed medications and attendance at medical appointments was excellent compared with what might be predicted for alcoholic individuals living without homes. Three participants died of causes and at ages that have previously been described among homeless people;^15,16,18 they died of intracerebral hemorrhage, cardiac arrest and acute alcoholic hepatitis, respectively. It must be noted that MAP is intended as a program with no stop date per admitted individual; participants would be expected to die of causes that are consequences of life-long addiction.

This study had limitations. Although it may have been preferable to compare 2 such groups in a randomized controlled trial, logistical, population and financial constraints made such methodology unfeasible. Potential biases identified with the one-group pretest–post-test study design include biases of history, maturation, testing and instrumentation, as well as statistical regression to the mean. However, there has been no change in ED, police or social policies to account for the decreased use of ED and police services. Maturation or biologic changes in the participants over time would tend to bias against MAP, with expected declines in health. Pre-and post-program hospital and police encounters would not be subjected to testing bias, since external databases were used. Observations were repeated over time with no instrument decay or regression to the mean. Clinical regression, in which participants might enter MAP when addictive consequences were at their worst and therefore appear to improve, is another possible source of bias; but the addiction in this group was of a severe and long-standing nature, and severity at program entry was likely representative of overall severity.

Continuity of care among homeless people has been found to be exceptionally difficult. Shelter operators already having demonstrated cultural competence in caring for the homeless were integrated into a shelter-based medical model of care to address previously unmet needs. This served to treat vulnerable individuals in a timely manner and coordinate their care, which allowed timely discharge from hospital. Police in frequent contact with people repeatedly inebriated in public have the opportunity to refer potential program participants to MAP and address a need within a system otherwise obliged to repeatedly process minor offences and bring people in for overnight detox in a police cell. Program development is ongoing for preventive care against infections such as tuberculosis and hepatitis and for administration of HIV tests and immunizations. For people whose drinking pattern has stabilized in MAP, psychiatric evaluations and follow-up have been successful.³¹ Finally, the option to detoxify from alcohol is always presented; once stabilized in the program, a few participants have successfully been medically detoxified and received housing, a formidable accomplishment considering the severity of an on-average 35-year addiction in which subjects drank daily to unconsciousness. This appears attributable to tempering alcohol consumption in a safe environment, which makes alterations of behaviour, including detoxification, possible.

In one large study,^32,33 mentally ill homeless people in supportive housing had decreased shelter use, incarcerations, admissions to hospital and lengths of hospital stay. In another study,²⁴ only 20% of people with case__managed alcoholism were able to maintain housing. Although housing is immensely beneficial for health, it is difficult to maintain without appropriate skills. Part of the success of MAP has likely been due to the supportive housing provided, but housing alone would not have prevented alcohol-seeking, consumption and the harm therefrom.

MAP is an innovative program based on a harm-reduction model that, when evaluated in a small group, appeared to be effective in decreasing alcohol consumption and the use of crisis services. Those responsible for the well-being of homeless people should consider the implementation and prospective evaluation of programs that integrate health services within shelters using a harm-reduction strategy.