The essential methodology is to recruit a group of controls who have seemingly been at risk for a long period without becoming infected and to compare it to an appropriate comparison group of cases who have been infected. An important component of this research is thus to find study sites and subjects for whom having remained uninfected is indeed an accomplishment; in practice, this means looking at data on infection markers by years of potential exposure.
Since the focus of this method is to explore ways in which those at risk run their lives so as to remain uninfected, it is important to minimize the probability that the negative controls have remained uninfected for biological reasons such as having immunity of a kind that does not show up on standard assays. One way to do this is to focus on more than one infection at a time. Thus, in New York we are studying how IDUs remain uninfected with both HIV and HCV by comparing these IDUs to those who are infected with both and those who are infected only with HCV. (The proportion that is infected only with HIV is quite small, and is more likely to reflect biological causation.) Although the fact that HIV prevalence among IDUs at this time is "only" 19% [27] reduces the value of HIV negativity as a way to rule out low biological susceptibility to hepatitis C seroconversion, Aitken et al., (2004) [28] have found a very low probability of falsely classifying subjects' hepatitis C status due to clearing prior infection (but remaining antibody-positive) or natural immunity. Nonetheless, there may be a small proportion of injection drug users who become exposed to hepatitis C but clear the virus without becoming antibody positive [29, 30], and this does pose some risk of misclassification–and supports the importance of focusing on more than one infection where possible.
Figure 1 provides an overview of the processes that this method addresses. Long-term risk avoidance involves minimizing being in situations and engaging in behaviors that pose a high risk of infection, and, when one is in a high-risk situation, having a repertoire of methods to avoid risk anyway. This means that the research focus is on practices that are socially-embedded in one's social network, as well as on how a person or small group of people avoid or resist social pressures to take risk. We want to learn from the experts–those who have avoided infection–what they have done to stay safe, what they have done for other reasons that has also helped them to stay safe, how they learned to act in these ways, and how they managed to maintain relatively safe behavior over a period of many years while engaging in behaviors (drug injection, selling sex) that, on the surface, seem to pose a high risk of leading to infection.
Recruiting the participants and eligibility issues
The logic of the project requires that participants will have been at risk (in this case, injecting drugs) long enough for most of their cohort to have become infected with one or both of the infectious agents being studied. A second component to this logic may be less obvious–the value of an upper bound on the exposure time for eligibility. This upper bound serves two primary substantive functions. The first is that we are generally going to try to understand how people stay safe under more or less current conditions, so we want to restrict the time period to a period in which conditions have been more or less similar. For New York, syringe exchange on a large scale began about 15 years before the project began gathering data, and this greatly reduced the difficulties of obtaining sterile injection equipment. The second reason an upper bound is useful is that it may restrict the effects of mortality upon the sample of infected participants. People who are infected with HIV (particularly if they are also infected with hepatitis C) had a very high mortality rate before the late 1990s–but this mainly became true more than five years after they became infected. Thus, for IDUs in New York, our criterion of recruiting IDUs with 8 – 15 years of injection experience (with interviews beginning in 2005) meant that they had all begun injecting after syringe exchange opened; had mainly injected after the subsequent great decline in incidence rates [31], and would have been very unlikely to have become infected early enough to have died before HAART became widely used. Finally, these criteria would recruit a sample of double-negatives who were positive deviants since only 25% of New York City IDUs in a reasonably representative detox treatment entry sample who had been injecting for 8 – 15 years remained both HIV and HCV negative [13].
Similar patterns of reasoning motivated colleagues in Australia and in England to propose 8 – 15 years of injecting as their sampling frame for studies of how IDUs there had remained uninfected with hepatitis C [13].
Since we are deliberately seeking out positive deviants (a relatively rare group) as half of our overall sample, there is considerable cost saving in studying subjects who are referred to us by research or other sites who themselves do the appropriate testing for HIV and hepatitis C antibody and also screen subjects by how long they have been injecting drugs. On the other hand, this has led to slow recruitment when these sites have other priorities, and also limits our sample to those who use the sites.
Two sources recruited the subjects included in this paper: First, the Risk Factors project at Beth Israel Medical Center detoxification center, which has been the subject of many research papers [27, 31] was the referral source for five subjects who were negative for both viruses, one who was positive for both viruses, and two who were positive only for hepatitis C. No one they approached to take part in the Staying Safe Project told them that they would not participate or refused to provide them with information about how we could reach them. However, since some time elapsed before antibody test results became available, it was not always possible to reach potential subjects to schedule an interview. This may have resulted in our sample under-recruiting IDUs with unstable lives. HIV testing was conducted at the New York City Department of Health Laboratory using repeated enzyme-linked immunosorbent assays (ELISA) testing with Western blot confirmation. Hepatitis C testing there used Abbott HCV EIA 2.0 (Hepatitis C Virus encoded Antigen (Recombinant c100-3, HC-31, andHC-34); confirmatory testing (RIBA HCV 3.0 SIA) was performed when absorbance values were equal to or less than 3.8: Chiron RIBA HCV 3.0 SIA (Hepatitis C Virus encoded antigen (Recombinant c33c and NS5 antigens;Synthetic 5-1-1, c100 and c22 peptides).
Second, the Etiology of Bloodborne Viral Iinfections project used respondent-driven sampling to recruit a relatively representative sample of injectors, and referred those meeting our criteria to us for interview. All eligible subjects whom they referred were interviewed. It was the referral source for twelve subjects who were negative for both viruses, two who were positive for both viruses, and three who were positive only for hepatitis C. Their sera were screened for anti-HCV with a third generation enzyme immunoassay (Abbott Laboratories, Chicago, Illinois); and anti-HIV testing was performed using licensed ELISA screening and western blot confirmatory tests.
Eliciting interview data
To elicit such data, we engage in a combination of questions that help us understand their overall biography and the threats and resources this has involved; their history of risk and risk avoidance; and detailed elicitation of some of the key concepts presented in Figure 1.
Eliciting biographical information
Life histories are elicited in a two-step process. First, we work with the participant to draw a visual timeline to serve us as a schematic reminder of the participant's overall life and, equally or more important, to serve as an easily-accessible reference during the remainder of the interview. (We enter this into TimeLine software after the interview since doing it during the interview is cumbersome and disrupts the natural flow of a person's describing her or his life.) An example of such a timeline appears as Figure 2.
These timeline diagrams provide a valuable resource in analysis that helps us identify patterns in the life history data such as the relatively large number of years it took some subjects between starting to inject and beginning to use needle exchange.
After this, we interview the subject with a more detailed biographically-structured interview guide (see Additional file 1) that elicits narrative about their history and experiences with drug use, access to drugs, sexual experiences and relationships, institutional experience (including medical history, hospitalization, incarceration and similar issues), knowledge and practices of HIV and hepatitis C and related issues, and strategies and tactics to avoid stigma, high risk situations, and infection.
Eliciting additional information about protective tactics, strategies and environments
During this interview, and particularly at the end of it, as well as in follow-up interviews, we try to understand the social processes that lead to these tactics, strategies and behaviors. Issues here include:
1.Practices versus behaviors: Probes about the extent to which particular behaviors or behavioral patterns are socially-embedded practices or are more individualized behaviors. This is not easy interviewing, because there is a strong tendency in the drug culture as well as in the society as a whole [32] to describe one's behaviors as expressions of one's personality and choices. One approach we take to probing this is to ask them about how they learned to do things that way–since one perspective on culture is that it consists largely of learned behavior. Another approach is to ask them about how others would react if they behaved in other ways, which gets at the processes by which norms are maintained [33–37]
2. Indigenous prevention tactics: One of our goals is to identify indigenous prevention tactics–that is, behavioral rules of thumb or practices which help subjects control their risk that have immediate face validity as possible prevention messages. These differ from trajectories, resources, and strategies in that they are immediate and do not necessarily address the issue of long-term maintenance of safer practices (including maintaining indigenous prevention tactics).
3. Strategic action: To determine whether a subject's particular pattern of action is strategic (i.e., planned and intentional), we probe about why they engaged in the pattern of action; the extent to which they planned it out; what they had in mind; whether it emerged out of group discussions, and if so, what the group was trying to accomplish if anything. Given the frequently-changing situations IDUs find themselves in, particularly in those periods of their lives when they are street users, they sometimes have trouble recognizing the extent to which their actions indeed are based on planning and intentionality rather than being overwhelmingly circumstantial and reactive. We have uncovered some of their strategies in these circumstances by asking them how they apply their principles of infection avoidance, for example, to different circumstances they face; and asking them to comment of how and why they sometimes act differently in various circumstances from how those with them act.
Personality data about traits
One possible explanation for why some IDUs might be able to remain uninfected would be if they share some common personality trait or traits that remain reasonably constant over the many years they have been injecting drugs. To test for this, we use selected parts of the NEO PI-R questionnaire. This questionnaire is based on Costa & McCrae's Five Factor Model of Personality (FFM), which is a highly influential theory of personality. It provides a relatively comprehensive taxonomy of individual differences. The five domains are Neuroticism, the tendency to experience negative affect (e.g., anxiety, depression, and hostility); Extraversion, the quantity and intensity of interpersonal interactions; Openness to Experience, seeking and appreciation of new experiences; Agreeableness, the quality of interpersonal interactions along a continuum from compassion to antagonism; and Conscientiousness, the amount of persistence, organization, and motivation in goal-directed behaviors [38, 39]. These factors show extraordinary stability over time in normal populations, with little change over decades [39, 40] (The NEO PI-R contains 240 items. It can be completed in 30 – 40 minutes).
Interview Contexts
Interviews were conducted under confidentiality and other protocols approved by the institutional review board of NDRI at a storefront field site located within a few blocks of one referral site; another source referred participants to us from several locations (including one in which they rented space in a separate part of our storefront). The different sections of the interviews took, on average: 55 minutes for the Time Line; 81 minutes for the detailed interview about protective strategies and related matters in the context of their life history; and, in seven cases when follow-up interviews have been conducted to elicit more information about these topics, with special attention to strategies and to the sharing of equipment other than syringes, approximately another 80 minutes. Participants are reimbursed $40 for their time and trouble for the first interview and $30 for any subsequent interviews.
Analyzing the data to develop grounded hypotheses
Data analysis follows fairly standard techniques from grounded theory and life history methodologies. All interviews are taped and transcribed. The transcripts are then coded by both of the field staff using a combination of theory-based and emergent coding categories [17], who also discussed and resolved differences in coding. The Principal Investigator read selected transcripts as well. Discussions during project meetings helped new concepts to emerge; there were often concepts that were more abstract or process-oriented than those that emerged during the coding itself. As staff developed ideas about important processes, categories, or concepts, they wrote and circulated theory notes for further discussion by project members.
Due to difficulty recruiting infected subjects during the early months of the project, the first set of emergent codes were dominated by the materials from subjects who had remained uninfected–that is, from those who had successfully "stayed safe." Later, as we interviewed IDUs who had been infected with HCV and in some cases also with HIV, categories began to emerge that reflect their lives and, usefully, the comparisons between the positives and the negatives. Additional analyses may be conducted comparing those who are infected with both HIV and HCV with those who are infected with HCV but not HIV.
These categories and concepts are partially reflected in the hypotheses presented in the Results section below. Since both data collection and analysis are ongoing in this project, we anticipate that additional grounded hypotheses will be developed and those presented here may be modified.
