Reports are included in this review if they satisfy the following criteria:

We will exclude reports that are SR/MAs, present self-reported HCV data, or use simulated data.

The study population is PWID. We define active PWID as individuals who have injected within the past year, while former PWID are those whose most recent injection drug use occurred over 1 year prior to study admission. Reports that do not provide a detailed description of what constitutes a PWID will be analyzed as a separate group (thus producing 3 categories of PWID: active, former, and status nonspecified). If reports explicitly state that PWID are current or active (or use similar language to convey that PWID have recently injected drugs or are currently injecting drugs) but do not specify a time frame, then we will consider these PWID as active PWID.

The primary outcome is the frequency of HCV infection. We use 4 measures to capture disease distribution:

Prevalence and incidence are measured as proportions and rates, respectively, where possible.

The preferred criteria for measuring prevalent HCV infection are both positive HCV antibody and HCV RNA test results. The alternative criterion is a single positive antibody marker.

Detection of HCV antibody or RNA in a previously seronegative individual are the preferred criteria for determining incident HCV infection. The alternate criterion is RNA positivity in an antibody-negative individual [12].

Measuring incident reinfection subsequent to spontaneous clearance or sustained virologic response is similar. The preferred criterion is a change in the HCV genotype from the cleared infection (either spontaneously or through treatment) to the recently acquired infection over consecutive tests. In the absence of genotype testing, conversion from RNA negative to RNA positive over multiple tests is the alternative criterion.

Secondary outcomes examined are associations between exposures and HCV incidence or prevalence. Exposures include biological and environmental factors such as HIV coinfection, sex, and geographic location.

Literature is searched electronically and manually. Electronic searches using a string consisting of terms related to HCV, epidemiology, and PWID are undertaken in 4 databases: Cumulative Index to Nursing and Allied Health Literature, Excerpta Medica database, ProQuest, and PubMed (see Multimedia Appendices 2-5). Results are filtered by date of publication (01/01/2006), record type (peer-reviewed journal), and language (English). Manual searches are performed on reference lists of eligible reports and other relevant papers, conference materials, and research study websites.

Reports are assessed for inclusion through 3 stages. First, each unique record in the deduplicated set of literature, all of which is retrieved from the search strategy, is screened by title and abstract to determine if it meets the eligibility criteria. Second, for every report that is considered eligible, the full text is subsequently screened. Reports that did not meet the criteria at either of these 2 stages are excluded from further consideration. Third, following data extraction, a final assessment is made on the admissibility of each report.

The quality of reports is determined through the use of an adapted version of the Quality In Prognosis Studies instrument developed by Hayden et al [13-14]. The instrument, which was modified for 2 previous SR/MAs [15-16], evaluates potential sources of bias, such as selection, misclassification, and confounding, in a report; it is also used to evaluate all study designs (eg, cross-sectional, prospective and retrospective cohort, randomized controlled trial).

Two research assistants, both with graduate training in epidemiologic research, systematic review methodology, and biostatistics, perform the literature search and identify eligible reports. Screening and coding pilots are conducted to evaluate intercoder reliability and refine the protocols governing report eligibility and data extraction. The research assistants independently review, code, and appraise all literature, which is subsequently evaluated by the principal investigator and the project director. The principal investigator and project director resolve any issues that emerge throughout the study selection and data extraction processes. The data for all included reports are aggregated into an electronic database that is managed by the research assistants, project director, and principal investigator. The research team is guided by the study protocol that they developed at the start of the project. This SR/MA is registered at PROSPERO [CRD42016035687].

The meta-analysis synthesizes report-level data that are collected on the following domains: study cohort, period, and geographic location; study design and methods; HCV incidence, prevalence, and reinfection; and participant characteristics, particularly factors understood to be associated with HCV infection (eg, age, sex, and duration of drug injection). We assess subgroups and the sensitivity of results—determined by report attributes, participant factors, and report quality—and, at each stage, heterogeneity, using Cochran Q and I². Summary estimates of the primary outcomes are derived using random-effects meta-analysis techniques. Meta-regression models assess the association between characteristics of the study or the sample (eg, location, time period, study quality, safe injection behavior, age, race/ethnicity) and HCV incidence and prevalence. Measures of relative effect are calculated, where possible. Data analysis is conducted in Stata (StataCorp LLC), and data visualization is performed in R (The R Foundation).