Factors associated with hepatitis B vaccine series completion in a randomized trial for injection drug users reached through syringe exchange programs in three US cities

An in-depth description of the study sample and procedures has been published elsewhere [15, 16]. Participants were enrolled between May 2003 and March 2006 at SEP locations in the three cities. Individuals were eligible to receive the vaccine if they could demonstrate evidence of having injected within the past 30 days (injection stigmata), were 18 years of age or older, were screened for and found susceptible to HBV, and were deemed able to provide informed consent. Oral consent for screening and subsequent participation based on HBV vaccination and infection status was obtained orally prior to screening.

To determine eligibility for vaccination, participants were screened by serological testing for antibodies to core and surface antigens (HBcAb and HBsAb, respectively) and for surface antigen (HBsAg). Individuals who tested negative for all three tests were informed that they were susceptible to HBV infection and invited to receive vaccination through the study. Individuals who tested HBcAb-positive, regardless of the test results for HBsAb were informed that they had been previously infected. Individuals who tested HBsAb-positive but negative on the other tests were informed that they had been previously successfully vaccinated. Individuals who tested HBsAg-positive, regardless of either of the other two HBV results, were informed that they were likely to be actively infected, instructed when they returned for their results to seek medical care, and advised on how to avoid transmitting the virus to others.Those whose serologic tests results indicated that they were susceptible to HBV infection were administered the first vaccine dose when they returned for their test results and then instructed to return for the second dose one month later. A random numbers assignment log, maintained at each study site, was used to assign participants to either the standard (0, 1, and 6 months) or accelerated (0, 1, and 2 months) schedule at the Dose 2 visit. All participants were invited to return for a final visit seven months after their first dose to assess acquisition of protective immunity, and in Chicago a fourth dose was given to those in the accelerated arm. The recruitment, intervention, and analysis scheme is shown in Figure 1.

The outcome of interest was administration of the final vaccine dose (Dose 3) among those receiving Dose 1. A set of covariates was included to test our hypotheses and identify confounders. These included treatment group, SEP usage variables, city of enrolment, sociodemographic variables, current health status variables, injection practices, and history of either engaging in or receiving commercial sex work services. Hepatitis knowledge and psychosocial factors were also assessed. The hepatitis knowledge instrument was an adaptation of the National Institute on Drug Abuse Risk Behavior Assessment questionnaire and included 22 questions using a True/False/Don’t Know response format to assess knowledge of routes of infection, disease detection and treatment, and prevention options [17, 18]. The psychosocial instrument, developed specifically for this study and grounded in Protection Motivation Theory, contained 21 items that assessed the six constructs in the theoretical model [19, 20]. These included respondents’ perceptions about (1) personal vulnerability to HBV infection, (2) disease severity for HBV infection, (3) response efficacy of specific prevention behaviors (e.g., using SEPs, refusing to share injection equipment) in reducing risk of HBV infection, (4) self-efficacy for engaging in specific prevention behaviors, (5) social peers’ approval of the respondent engaging in HBV risk reduction behaviors (i.e., social outcome expectancy), and (6) the importance of social approval from their peers (i.e., social outcome value). Items were phrased as an “I- statements” and used a 5-point Likert response format ranging from Strongly Agree to Strongly Disagree. The number of items included in each sub-scale ranged from one (vulnerability and severity subscales) to seven (response efficacy) items. Inter-item correlations (Cronbach’s alpha) for the multiple-item subscales were calculated and ranged from 0.62 to 0.86. Data obtained from these items were converted to z-scores, and the mean score for each subscale was calculated. A series of questions were included concerning motives for participation that included remaining healthy, protecting the health of others, participating in research, and receiving compensation.

Statistical analyses were conducted using SAS version 9.1. Logistic regression was used to calculate bivariate associations between all co-variates and the outcome. Covariates that were significantly associated with vaccine completion at the p ≤ 0.10 level in the bivariate analyses were subsequently entered into a multivariate logistic regression model with backwards elimination of any covariate that did not remain significant at the p < 0.05 level or did not change other coefficients by >10%.

We also explored several areas of potential collinearity. To test the hypothesis that those earning more money were more likely to have had a recent medical visit and less likely to use SEPs [21, 22], chi-square tests were used to test for associations between income and doctor visits during the past year and income and SEP utilization. The laws governing syringe exchange varied with there being no limit to the number of syringes that could be received at each visit in Chicago in contrast to a cap of 30 syringes that could be received on a one-for-one basis in Hartford and Bridgeport. Therefore, to test if there was an interaction between participants’ main source of syringes and their city of enrollment, an interaction term was created and entered into a multivariate logistic regression including the interaction term and main effects with receiving Dose 3 as the outcome.

This analysis, designed to compare completion rates between two dosing schedules, was constructed on the hypothesis that completion rates would be higher among the accelerated group. A second manuscript comparing efficacy between those randomized to the accelerated versus the standard schedule is in preparation. As anticipated, vaccination completion was significantly more likely among participants in the accelerated treatment group, suggesting that shortening the vaccine schedule reduces the risk of encountering barriers to participation common among IDUs, which can include incarceration or competing health care and personal needs.

Low rates of completion of the three-dose HBV vaccine series have previously been documented among IDUs especially in, but not limited to, the United States [9, 23‐25]. Accelerated HBV vaccine dosing schedules that have been implemented among hard-to-reach populations including active IDUs have been shown to increase completion rates [8, 26‐29]. However, none of these prior studies has used SEPs as the access point for identifying and vaccinating active injectors. In our present study, despite offering financial incentives for vaccination at SEPs, only 58% of Dose 1 recipients completed the vaccine series. However, several studies have established a consensus that paying individuals do increase vaccination completion rates in drug injecting populations [10, 30, 31].

There has been discussion regarding whether or not a booster dose is necessary for the accelerated HBV vaccine series. Although benefits of a booster dose for the accelerated HBV vaccine schedule have been suggested, several studies have concluded that booster doses are neither necessary nor is maintenance of an antibody level ≥10 mIU/mL essential for protection because an anamnestic response has been detected up to 22 years post-vaccination [32‐34]. Further research would be necessary to evaluate the feasibility of offering a 12-month booster dose at SEPs. Although this might not seem a priority in light of the above findings, the lower success rate for vaccination among people who inject drugs suggests that a booster might be appropriate in this population [24, 35‐39].

Completion was slightly more likely among older participants, consistent with earlier findings of low vaccination rates among young injectors and among younger people in general [8, 11, 40]. Thus, special attention should be paid to recruiting and retaining younger participants, especially since they are less likely to have already been exposed to and infected with HBV (albeit increasingly more likely in the future to have been vaccinated against HBV infection in early childhood). Multifaceted interventions, targeting youth within their community deserve further consideration. A review of the literature on services and interventions for runaway and homeless youth concluded that interventions addressing the varied and interconnected needs of youth are more successful than those targeting one problem at a time [41, 42]. Further research is needed to specifically understand barriers to vaccine program completion among young injectors.

Participants with a poorer self-rated health score were more likely to complete the vaccine series. This runs counter to expectations from previous work that found competing needs served as a barrier to health care acquisition and preventive health care, particularly among vulnerable populations including homeless adults and IDUs [43‐45]. This may, instead, be one example of the potential for collinearity in the variables in our dataset. Older individuals were more likely to report poorer health (data not shown) and were also more likely to complete the vaccine series.

Compared to direct SEP customers, a significantly lower completion rate was observed among people who engaged in secondary exchange. The benefit of providing health services at SEPs has been repeatedly demonstrated [3, 9, 12, 23, 46, 47], but this advantage may be of limited benefit to non-customers. Our findings suggest that recruitment of non-SEP customers should draw on peer networks in addition to more traditional recruitment strategies (e.g., outreach, posted fliers). SEP customers who distribute SEP syringes to non-customers should be encouraged to promote participation through their own social networks, word of mouth, or easy-to-distribute promotional cards. The underlying reasons that some IDUs do not go directly to SEPs were not assessed within the current study. Further research is needed to identify barriers to SEP utilization and alternative methods for targeting this hidden population of IDUs.

This study has a number of limitations. First, participants were predominantly direct or indirect SEP customers, so the findings have limited generalizability to the full IDU population in these communities. Second, as with any study that relies on self-reported behavioral data, this evaluation may be influenced by self-report and recall bias. Third, recruitment methods were intended to simulate what would be realistic and feasible for a typical SEP with limited resources and staffing to implement a vaccination intervention; the recruitment strategy may have resulted in a non-random sampling that may have introduced unmeasured bias into the study population. Fourth, lack of adequate specificity in some of the variables collected such as sources of income, a full medical history, and the precise nature of social relationships limited a full assessment of competing subsistence or health needs. Finally, enrolment was slow. It took three years to recruit sufficient participants to power the analysis. We have previously reported that an improved study design would have involved getting rid of the two-week waiting period for test result to identify those eligible for vaccination and giving the first dose of vaccine at enrolment [15]. This approach was also found to be most cost-effective.