Choice of the RDS sampling strategy
Former studies among drug users in Germany were conducted as convenience sample in drug consumption facilities, or among persons in OST-facilities or detoxification units [
4,
12‐
14,
54,
59]. We aimed to reach a more representative sample of the whole drug scene in a city and chose RDS, because this method claims to achieve representativeness. PWID are known to be a well-networked group, and RDS proved to be an effective method to rapidly recruit PWID [
46,
50,
60,
61]. A long recruitment process over months would not be feasible in most study sites. Furthermore, RDS makes it possible to recruit PWID who might not be reached by the existing low-threshold facilities in the study cities and thus can only be reached by their personal social network. Collecting information about the PWID currently not reached is vital for improving local prevention efforts as well as for the planning of future regular monitoring of HBV, HCV and HIV among PWID by using data collected through these facilities.
Asking questions on knowledge
Initially, in the pilot cities, the knowledge section of the questionnaire was designed questions on transmission and prevention of HBV, HCV and HIV with true and false response options that participants had to choose. Due to concerns about interviewer bias as well as for ethical considerations - realising that the knowledge questions presented an opportunity of learning for the participants - the design was changed. After discussion with experts during the kick-off-meeting of this study in May 2012, true statements concerning transmission, prevention, treatment and general characteristics of HBV, HCV and HIV are presented to participants. By providing true statements, and also introducing them as such, participants now have the opportunity to learn something while completing the interview. In case of knowledge gaps, participants are offered to see a counsellor, even if test results are not requested by the person. Learning facts about transmission and prevention of blood-borne viral diseases during the interview and being counselled in case of knowledge gaps is one of the interventional parts of the DRUCK-study. First experiences with this kind of knowledge statements prove feasible and well-accepted by both study staff and participants.
Pre- and post-test counselling in low-threshold drug facilities
Until now, voluntary counselling and testing (VCT) for blood-borne viruses in Germany is mainly offered by medical doctors, substitution therapy specialists, and by some local health departments (anonymously and free of charge). Due to the illicit nature of drug consumption, PWID are not well reached by these services. A pilot project offering anonymous HIV rapid testing and counselling in some German drop-in-facilities showed that this group was reached successfully [
62].
We conduct this survey in low-threshold drug facilities, where VCT for HBV, HCV and HIV was rarely available before. Before starting recruitment, staff is trained for all VCT procedures. Counselling follows international recommendations and the German recommendations provided by Deutsche AIDS-Hilfe [
63]. If well-accepted during the study by both, staff and injectors, implementation of a regular VCT offer and of rapid testing in the facility may be eased.
Challenges and limitations
There are a number of important challenges and limitations preexisting when designing the study. PWID are a stigmatized, hard-to reach population. Although all data is anonymised, and no personal information is collected from study participants, participants might have difficulties in reporting sensitive data such as unsafe use and sexual behaviours, imprisonment or testing history. We try to limit the reporting bias of sensitive data by choosing interviewers who are familiar to most of the local PWID (e.g. social workers or volunteers working at the local facilities). Behavioural surveys cannot take place without the informed consent of the respondent. However, signing a form might inhibit the feeling of anonymity. Thus, we offer participants to give oral consent to the study site manager, if written consent is refused.
Another issue concerns the question of true answers in self-reported behavioural data due to social desirability, and also of accuracy in items with 12 months recall periods.
Furthermore, potential selection bias might be a problem in our study population. Experience has shown that respondents are more likely to refuse to participate if they are asked to provide a specimen for testing [
41]. Reaching the targeted sub-sample size for the respective study cities depends on many factors and might not be possible for each city. To reach the targeted total sample size (2,033), it might be necessary to conduct the study in an additional city.
In chronic infections like HBV, HCV and HIV infections it is not possible to draw direct associations between behavioural items and time of acquisition of infection. We can only identify associations between the status “infected” or “not infected” and risk and preventive behaviours reported for a certain time period in the past. In case an infection was newly diagnosed in the study we may assume that the same behaviour caused infection. If an infection was known already by the participant, direction of cause and effect could be both ways.
Because of the cross-sectional nature of this survey, we will be unable to infer direct causal relationships between socio-demographic and behavioural factors or the impact of prevention programs on behaviour change.
Moreover, a change of laboratory methodologies could limit direct comparison of test results of the two pilot cities and the remaining six cities. The dilution of the original sample volume during elution from the filter disk was 1:10 for the first two cities and then increased to maximally 1:15 for the following six cities to allow repeated testing of samples. With respect to HIV serology, established HIV infections are detected at comparable sensitivity by the two assays, however, individuals in a very early seroconversion stage could be missed by third generation EIA [50% of samples in seroconversion (reactive EIA/indeterminate immunoblot] resulted in false-negative immunoblot results, but still showing reactive/indeterminate ELISA]. The prevalence of HIV could therefore be slightly underestimated in 6/8 cities, but seroconverters are expected to represent only a very small fraction of the study participants.
We furthermore had to modify the anti-HBV and anti-HCV antibodies test systems after the pilot study. All methodologies were validated, but direct comparison between study sites has to be done with caution. Assessment of all serological test systems for HBV and HCV shows good accordance between directly tested serum samples in comparison to DBS except for anti-HBs in weakly positive sera. Weakly anti-HBs positive samples could be missed with this procedure. Thus the prevalence of anti-HBs antibodies, e.g. of HBV vaccinated individuals, based on the DBS technique, must be considered as minimal estimate. However, the anti-HCV and anti-HBc results demonstrate high accuracy of our test systems and yielded comparable validation results.
This is the first sero-behavioural survey among current injectors in Germany aiming to collect representative data on blood-borne viral infections in several cities in over 20 years. Knowledge from behavioural, serological and molecular testing data will help to focus prevention strategies. Depending on the outcomes, we will recommend conducting similar surveys in regular intervals, to monitor changes in behaviour and prevalence over time and assess outcomes of adapted preventive measures.