Design of the Indian NCA study (Indian national collaboration on AIDS): a cluster randomized trial to evaluate the effectiveness of integrated care centers to improve HIV outcomes among men who have sex with men and persons who inject drugs in India

Before initiating the baseline assessment, we conducted ethnographic research and community preparedness in 27 candidate study sites to: (1) identify potential “seeds” for RDS; (2) assess potential for contamination across study sites by exploring mobility; and (3) understand regional variation in existing HIV prevention and treatment services. On average, two focus group discussions (FGDs) and 6-8 in-depth interviews (IDIs) were conducted in each potential study site.

During this preparatory phase, we also conducted community meetings for a 3-month period with peer leaders and outreach workers from the MSM/PWID NGOs in the community to inform them of the study. RDS is a peer-driven chain referral strategy that hinges on the ability of participants to recruit peers. Injection drug use and same-sex behavior are both stigmatized in India and punishable by law. Thus, a key goal of these community meetings was to make the target populations aware that they might receive coupons from their peers/friends/sexual or injection partners to participate in a study and that this study was not a ploy to harm or arrest MSM or PWID but rather to evaluate their access to HIV services and understand the needs of the communities.

The goal of the baseline assessment, conducted between September 2012 and December 2013, was to establish baseline prevalence of study outcomes for sites in the trial. We accrued samples of ~1000 eligible participants in each candidate study site where we partnered with one or more NGOs working with the target population. Sampling was conducted using RDS, a chain-referral sampling method [50, 51] which approximates a probability sample of populations when sampling frames are lacking. It is similar to snowball sampling [52] except the recruitment process is implemented in a systematic way that allows for the calculation of selection probabilities. Participants select and provide referral coupons to individuals in their peer network [50]. Verbal informed consent was obtained from all participants.

Eligibility criteria were:

RDS was initiated at each site with participants (“seeds”) selected during ethnography. Two seeds were selected from a list of 10 per site to represent varying demographic, geographic (different parts of the city), HIV status, and for MSM sites, sexual identity (insertive vs. penetrative vs. versatile) and for PWID sites, drug-related diversity (e.g., heroin vs. other opioid injection; daily vs. less frequent injection) in the local population. While the initial plan was to select 4–7 seeds, recruitment in nearly all sites proceeded at a rapid pace – therefore, in 25 of 27 sites no additional “seeds” were added. In one MSM and one PWID site, a third seed was added as recruitment was slower than the other sites. In one PWID site (Moreh), recruitment was terminated prematurely for safety considerations due to civil unrest.

“Seeds” and subsequent RDS recruits were asked to provide a scan of their fingerprint. The fingerprint image is immediately converted to a unique hexadecimal code and stored; the image itself is not stored. This code is linked to a study ID, which is used on participant forms and laboratory samples and is also used to rule out duplicate enrollments within a site and to link participation across multiple phases of the study.

All participants underwent a survey followed by HIV pre- and post-counseling and a blood draw. Survey modules and laboratory tests are provided in Table 2. English language versions of the surveys used at MSM and PWID sites are available as Additional files 1 and 2. The survey was conducted by trained interviewers who were hired expressly for the pre- or post-intervention RDS surveys, did not work with or have previous interactions with the target population in question, and had no stake in the outcome of the ICC evaluation. Interviewers recorded answers to a web-based, secure database via laptops and a local area network. RDS interviewers and support staff were trained extensively on visit flow, documentation, questionnaires, and laboratory procedures using a single training protocol across sites. Quality control for the survey was maintained by programmed logic checks and real-time data monitoring algorithms and by site supervisors who monitored 5 % of randomly selected interviews for proper interviewing technique. Constraints placed on the database required interviewers to answer every question on the survey and ensured that missing data was minimal.

Participants were also asked to recruit up to two members of their sexual (MSM) or drug-using networks (PWID) who satisfied the study eligibility criteria using bar-coded coupons that had a holographic image to hinder replication attempts. If participants successfully referred eligible participants, they received an additional incentive of INR 50 (USD 0.8) per eligible person recruited in addition to compensation of INR 250 (USD 4.1) for completing the pre-intervention assessment. Participants were recruited in successive RDS waves at each site until the desired sample size was accrued.

Randomization took place after the pre-intervention assessment was completed. First, we selected 12 PWID sites (from 15) and 10 MSM sites (from 12) for randomization. In the PWID stratum, two sites were removed because of logistical challenges that deemed them unsuitable for randomization (Moreh, Gangtok). Three additional sites were dropped based on very low HIV prevalence (Bhubaneshwar [PWID-stratum], Lucknow and Mangalore [MSM-stratum]).

We used a restricted stratified randomization approach to randomly distribute the 22 sites to either the intervention or control condition [53]. In CRTs, the number of randomized units is relatively small and large imbalances between study arms may occur if randomization is unrestricted; hence, restricted randomization is often used to ensure reasonable balance between study arms in important factors. However, the desire for balance between arms must be balanced against leaving a sufficient number of randomization options (e.g., at least 100).

Sites were stratified based on risk group (MSM and PWID) and then additional restriction criteria were used to ensure balance, first within strata and then overall. Within strata, restrictions were made on the basis of geography, HIV prevalence and the primary outcome: HCT in the prior 12 months (Table 3). Additional restrictions were made on outcomes among HIV positive persons. All within-strata restrictions were based on RDS-weighted proportions of the outcomes. After strata-specific restrictions were made, the two strata were combined to derive a combined set of allocations. Final restrictions were made using the same outcomes with the exception that both RDS-weighted and unweighted proportions were considered.

Restrictions related to geography and HIV prevalence were imposed because both were related to the stage of the HIV epidemic in the target population and HIV service availability. Further, geographic restrictions were important for political reasons, such that all intervention sites were not restricted to one region or state. Additional covariates used in the restriction process represented the primary and secondary outcomes. Of a total 232,848 possible allocations, 596 allocations remained after all restrictions. Across all combinations, there were only two sites that had >75 % chance of being randomized to the same arm [53]. Three individuals independent from the study who were blinded to the allocation sequence associated with each of the possible numbers chosen (e.g., 001 to 596) were asked to draw numbered balls from an opaque container to arrive at the final 3 digit number, corresponding to a numbered randomization combination. A recording of the randomization is available at: https://​www.​youtube.​com/​watch?​v=​vmHYHgv_​uS0. The final allocation is shown in Fig. 2.

For tuberculosis care, samples for screening and diagnosis are collected at the ICC, with confirmatory testing taking place at government centers. If a participant is diagnosed with TB, he/she is linked to the government DOT center most convenient to him/her for treatment initiation and follow-up. There are over 400,000 DOT centers in India ensuring easy access to all populations with excellent retention rates. Peer-health workers from the ICCs follow up with clients diagnosed with TB to ensure completion of the DOT program.

A post-intervention assessment will be conducted in all 22 intervention and control sites approximately 24 months after the establishment of the ICCs (beginning in August 2016). RDS will be used to accrue samples of 1000 persons in each site using identical eligibility criteria as the baseline pre-intervention assessment. An additional module will be added to the survey to collect data on utilization and perceptions of the ICCs. Participants will undergo a blood draw and laboratory testing as in the pre-intervention assessment.

An important consideration is the selection of seeds for the post-intervention RDS. We will select either the same seeds, if possible (i.e., if the seeds are alive, residing in the same community and still regarded as peer leaders in the community) or seeds that are as similar as possible to the seeds used in the pre-intervention RDS (e.g., age, area of town they reside, marital status, sexual preference, drug of choice, etc.).

For the pre- and post-intervention assessments, we developed software in-house that references features of RDS coupon manager software and tracks recruitment and coupon numbers, links coupons of recruiters and recruits, tracks non-eligible referrals and determines reimbursements. The software also incorporates biometric data capture (fingerprint images) allowing storage of system-generated unique non-identifiable reference keys that are linked to study identification numbers. This biometric information is used to identify duplicate recruitments within and across sites in the pre- and post-intervention assessments. This information will also be used to identify participants who participated in both the pre- and post-intervention assessments and utilized an ICC.

Electronic surveys for the pre- and post-intervention interviews were developed using the Lime Survey Open Source Tool embedded with JAVA Scripts that includes logic checks, skip patterns and data constraints. The interview data is linked to the information captured in the in-house coupon manager software using the coupon ID. We utilize PHPMYADMIN with a secure password-protected encrypted database to store all data in a cloud. All data is exported from the individual sites via a Virtual Private Network tunnel to the central database maintained at YRGCARE in Chennai. When the local sites are connected to the internet, fingerprint–generated codes from all sites are pushed from the central server to the fingerprint database at each site. Storing this data centrally allows for the identification of duplicates in real-time within and across sites.

Laboratory specimens for the pre- and post-intervention surveys are collected on-site and processed and transported daily via courier to the central YRGCARE lab in Chennai for further testing and long-term storage. Only rapid HIV testing is performed on site where the data collection takes place. Additional CD4, HIV RNA, syphilis and HSV-2 testing are performed at YRGCARE. Indeterminate HIV results were resolved using Western Blot testing at YRGCARE. Remaining plasma and serum specimens are being stored at -70 °C at YRGCARE.

The trial is being conducted under regulatory review by institutional review boards at YRG CARE, Johns Hopkins School of Medicine, and Johns Hopkins Bloomberg School of Public Health. Additional trial oversight is provided by a data safety monitoring board (for the PWID stratum) and an advisory board (for the MSM stratum), both comprised of expert members external to the investigators’ organizations.

The primary analysis will be to compare the prevalence of community–level outcomes across intervention and control clusters, adjusting for pre-intervention prevalence levels. For a given outcome, we will first log-transform the 22 cluster-level proportions obtained from the post-intervention RDS. Then, via weighted least squares linear regression, these will be regressed on a dummy term for the control arm (vs. intervention), another for the MSM stratum (vs. PWID), and a term for the log-transformed cluster-level pre-intervention proportions. The exponentiated coefficient for the control arm term is thus the prevalence risk ratio (PRR), and (1-PRR) × 100 % is the percentage increase in service utilization associated with the intervention. The primary analyses will be conducted using the RDS-II weighted cluster-level proportions from both pre- and post-intervention RDS samples. These RDS-II weights, which account for personal network size (number of PWID or MSM seen in the past 30 days), will be calculated using the RDS Analyst Software Version 0.5 (http://​hpmrg.​org). For secondary outcomes that are continuous (e.g., community viral load) we will use a similar regression approach.

We will conduct several sensitivity analyses for the primary and all secondary outcomes First, we will repeat analyses using unweighted cluster-level proportions from both the pre-and post intervention RDS surveys. Second, we will consider adjustment for demographic covariates (age, sex, marital status and educational attainment) measured at the post-intervention RDS that are associated with the outcome and are differentially distributed across intervention and control clusters. We will consider adjustment for these factors if the p values for associations with the outcome and the intervention vs. control clusters are <0.05 and the OR is >2 or < 0.5. A two-stage approach will be used when adjusting for individual-level covariates: at the first stage, for a prevalence outcome, individual responses are modeled with a logistic regression model adjusting for all relevant covariates except the dummy term for control vs. intervention. In the second stage, observed and expected prevalence counts for each cluster are calculated, followed by t-test-like analyses of log-transformed ratios of observed to expected [53]. We will also consider an approach that models the difference in outcome prevalence between the pre- and post- intervention surveys using the same cluster-level comparison approach. We will also consider individual-level analyses using multi-level random effects regression approaches (Stata GLLAMMs program) to account for dependence of responses within clusters [54‐56]. These models allow inclusion of fixed effects (e.g., intervention), random effects (e.g., clusters), adjustment for pre-intervention covariates at the individual and cluster level, and incorporation of scaled RDS weights as sampling weights. Finally, we will conduct descriptive analyses of the HIV care continuum, before and after the intervention phase of the study, in which completion of earlier steps are assumed to be necessary to complete later steps. Additionally, we will consider sensitivity analyses of outcomes in the HIV care continuum, where biologic markers such as HIV RNA and serum antiretroviral drug testing, are used to supplement self-reported data on access to care.

Several subgroup analyses are also planned. First, we will analyze all outcomes separately within each stratum (PWID and MSM). Using the combined sample of MSM and PWID sites, we will further compare all outcomes within subgroups defined by age, marital status, educational attainment, substance use (drug and alcohol use), and personal network size (number of persons in risk group [PWID or MSM] known and seen in the prior 30 days). Using only the PWID sites, we will also analyze subgroup differences by age, sex, marital status, educational attainment, substance use (including alcohol use), personal network size and region. In the MSM sites, we will also analyze subgroup differences by age, sexual identity, marital status, educational attainment, substance use (including alcohol use), personal network size and region. We also plan subgroup analyses by HIV serostatus and awareness of status for risk behaviors, HIV testing of spouses, substance use, stigma, and depression.

Network effects will be ascertained by comparing utilization of ICCs and services within ICCs across networks as defined by RDS. For example, we will examine utilization patterns across recruiters and recruits in the evaluation RDS. We will also ascertain whether utilization of ICCs varied by wave of RDS. Individual-level comparisons will draw on data from post-intervention RDS participants in the intervention clusters, in which extra questions will address participants’ use of ICC services. In addition, biometric data at the post-intervntion RDS will be linked with the biometric data from the ICCs to determine utilization. Individual analyses will use descriptive statistics and log-binomial regression to compare the level of each outcome (e.g., proportion accessing HCT in prior 12 months) by the main exposure of interest (visiting an ICC). We will adjust for individual-level confounders including demographic characteristics. Analyses will use multi-level random effects regression approaches to account for dependence of responses within personal networks [56]. In addition, using the biometric data to link persons between the pre-and post intervention RDS samples, we will conduct exploratory within-individual comparisons of the primary outcome and secondary outcomes. For example, restricting the sample to persons who participated in both the pre- and post intervention RDS samples and are eligible for the outcome, we will compare across control and intervention clusters the proportions of person who transition from not having the outcome to having it and vice versa.