Data Sources
The framework for the Vancouver SIF evaluation was designed prior to the SIF's opening and involved a number of methodological approaches. In light of the lack of existing quantitative efficacy data [17], the existence of ethical concerns [22], and an awareness that a non-randomized studies may be vulnerable to substantial selection biases [23], the Vancouver SIF evaluation is primarily structured around a prospective cohort design that involves the longitudinal measurement of a number of outcomes including blood-borne infection and overdose incidence, risk behavior, drug use practices, such as public drug use, and health services use.
The Vancouver SIF evaluation is somewhat unique because of the availability of a number of pre-existing data sources. These data sources include the community health and safety evaluation (CHASE) cohort, which is a community recruited virtual cohort of Downtown Eastside residents that prospectively and retrospectively examines health service use in the community by linking to administrative health record databases. In addition, the Vancouver Injection Drug Users Study (VIDUS) is an ongoing prospective cohort study of injection drug users that involves semi-annual serology of HIV and HCV as well as a semi-annual questionnaire [24]. VIDUS and CHASE allow for the description of IDUs in the community who are using Insite and a comparison between those that are and are not using the service.
In addition, in order to augment these data sources and to allow for close examination of the characteristics of Insite clients over time, a prospective cohort of Insite users has also been established. The Scientific Evaluation of Supervised Injecting (SEOSI) cohort is based on a representative sample of Insite users. The sample is derived through random recruitment of Insite users who are offered an informed consent to enroll into the study. Random recruitment involves attending the SIF at times of the day that are randomly selected using a random number generation program in SPSS, and inviting all users who use the SIF at this time to enroll in the study. As with VIDUS, participants provide a blood sample and conduct an interviewer-administered questionnaire. The SEOSI questionnaire deals with items that are particularly relevant to Insite, such as risk behaviours, public drug use, satisfaction with Insite, and access to medical care and addiction treatment services. All SEOSI participants provide informed consent to link to the Insite database so that SIF use can be tracked, as well as informed consent to access administrative health record databases in the community. As of September 1, 2004 over 900 Insite users have been enrolled into SEOSI and comparisons of socio-demographic variables (age, gender, etc) has shown that the SEOSI cohort is statistically similar to the overall cohort of insight users (all p > 0.05).
Client Satisfaction
Measures of client satisfaction are compiled as part of the SEOSI questionnaire. Through ratings of service quality in terms of the 5 SERVQUAL dimensions: Tangibles (e.g., the appearance of the physical facilities); Reliability (e.g., the ability of staff to perform the service dependably); Responsiveness (e.g., the willingness of staff to help clients and provide prompt service); Assurance (e.g., security, credibility and courtesy); and Empathy (e.g., ease of access, approachability and effort taken to understand clients' requirements). Similarly, reasons for avoiding the service are measured among IDUs in VIDUS who have not used Insite.
Additional Data Sources
These above prospective cohort data will be augmented by a number of other data sources including: process indicators, measures of community satisfaction and perceived impact, standardized measures of public order, and qualitative and quantitative measures of the health of the target population. The collection of each of these data sources is described below.
Process Measures
In order to track service use in the database at an individual level, while allowing for participant anonymity, each client must select a unique client 'handle' or nickname. The SIF database has a search function that allows for rapid searches based on demographic information, such as birth date, if an individual forgets their handle. Similar anonymous tracking of individual clients is commonly used at needle exchanges and other services for illicit injection drug users [25].
A primary purpose of the evaluation is to measure process indicators related to service uptake within the SIF, and this is enabled through the Insite database. The database tracks what drugs participants are consuming (heroin, cocaine, etc) and what services, such as nursing care and counseling services, are accessed by each client. For instance, in the month of May 2004, over 1300 unique visits were logged into the database.
Community and Staff Satisfaction
Community satisfaction and the perceived impact of the SIF on business persons are measured through a community survey that is performed in person among street recruited residents and at street-level businesses. The survey is similar to surveys being used in the Sydney SIF trial, and examines perceived changes in the neighborhood after the SIF's opening. In addition, staff satisfaction with the operation of the facility is measured through focus groups and qualitative interviews with staff persons. These interviews focus on how service delivery can be improved and on what measures can be taken to ensure staff safety and satisfaction.
Public Order
Standardized measures of public order were undertaken to examine the impact of the SIF on several indicators of public injection drug use. In brief, the survey protocol involves measuring specified public order indicators within an a priori defined geographical area in the neighborhood and at a priori defined times of the week. Data collection times are spread evenly throughout the week and involved walking through the study zone in the same pattern. Measures of discarded syringes, injection-related litter, and public injection drug use are all measured prospectively. An evaluation of these indicators has recently been described in detail [26].