H2R Newsletter

New Approaches in H2R Research Methods

Mapping Virtual Platforms

If most H2R population members do not attend physical locations but use virtual platforms such as social networking platforms (Facebook, WhatsApp) and apps (Badoo, Grindr, Tagged, Planet Romeo, and Hornet, etc.) to seek partners or clients, how can you find and count them?

Emmanuel et al conducted virtual mapping in three counties in Kenya to map, estimate, and measure the typology of men who have sex with men (MSM). Using a multi-stage approach, they identified 24 apps, 18 Facebook accounts/pages, and 18 WhatsApp groups across the three counties, with more than 6600 MSM users who could have been missed in conventional physical mapping.

Methods & Quality Checklist

Today's review is about Capture-Recapture (1,2)

Overview

In a two-round capture-recapture, key population (KP) members are surveyed initially and “tagged” (first round), perhaps by receipt of a small token, gift, or unique object. KP population members are then resurveyed at a later date (second round), with participants asked if they participated or were tagged by the prior survey.

In three- or more- rounds of capture-recapture, KP members are surveyed three times or more. The number of participants in each survey, as well as the size of the overlap, provides a basis to estimate the total population size.

Capture-recapture requires at least two rounds, but with three (or more) rounds, you can adjust for round dependencies, if they exist. Instead of survey rounds, you can use existing sources (i.e., data or lists) to run capture-recapture if you can identify the size of overlaps between the sources.

Examples of existing sources are case-based reporting systems, laboratory-based surveillance data, registries, insurance, or other existing program or study data sources that have a unique identifier to be used to determine the overall.

Advantages

• It has well-developed methods for analysis.
• If conducted properly it is scientifically rigorous.

Weaknesses

It has five key assumptions:

• Each round (source) needs to be representative of the target KP population.
• Require the accurate measurement of the overlap size between rounds (or sources)
• Closed population (no significant in or out migration or mortality)
• Each individual has the same chance of being included.
• Independent rounds

Possible Biases

If the two rounds (or data sources) are correlated (i.e., the rounds are dependent), meaning that participating in one round would change the probability of participating in the second round, the size will be underestimated.

Capture-recapture Quality Checklist

• The key population definition/eligibility is carefully defined (e.g., PWID: people who inject non-therapeutic drugs at least once in the last month) and being evaluated accurately for each participant in each round (data source)
• The number of eligible participants in each round (source) as well as the size of the overlap between them is accurately recorded. Outreach staff to be trained to ensure good quality data and recording of the numbers.
• There should be enough lag between the two survey rounds (i.e., one month or more) to ensure the second round is independent from the first round by leaving some time for the key population to mix up.
• Include three or more rounds if feasible and so you can adjust for round dependencies (if exist).
• For the analyses of three or more rounds, capture-recapture includes the interaction terms into regression models (e.g., log-linear model) to account for the round decency (if exists)
• Compare and contrast the population size estimates resulting from your capture-recapture analysis to prior knowledge and estimates from other methods, triangulate, and systemize the results. Keep in mind that there is no one perfect method for population size estimation.
• Data sources are in depend on each other (being in one data source/list does not automatically lead for appearing in another list).

[1] Rasheed A, Sharifi H, Wesson P, Pashtoon SJ, Tavakoli F, Ghalekhani N, Haghdoost AA, Atarud A, Banehsi MR, Hamdard N, Sadaat SI, McFarland W, Mirzazadeh A. Mapping and population size estimates of people who inject drugs in Afghanistan in 2019: Synthesis of multiple methods. PLoS One. 2022 Jan 28;17(1):e0262405.



[2] McIntyre AF, Mitchell A, Stafford KA, Nwafor SU, Lo J, Sebastian V, Schwitters A, Swaminathan M, Dalhatu I, Charurat M. Key Population Size Estimation to Guide HIV Epidemic Responses in Nigeria: Bayesian Analysis of 3-Source Capture-Recapture Data. JMIR Public Health Surveill. 2022 Oct 26;8(10):e34555.

Data Corner & Consultation

Data Request

If you are a student, post-doc, researcher, faculty, physician, or health officer, it is never too late to start your H2R research or conduct secondary data analysis and write a paper for a health topic on H2R populations.

Submit your data request online to be involved in one of our past or current projects on an H2R population.

Office Hours

We have office hours for those who want to meet virtually or in person and consult with one of our team members about an H2R population study or method.



• Tuesdays (17:00 to 18:00 PT)
• Wednesdays (16:00 to 17:00 PT)

Team Achievements

Publications

• The association between religiosity and resilience among young trans women. Wang JC, McFarland W, Arayasirikul S, Wilson EC.PLoS One. 2023 Jul 31;18(7):e0263492.

• Stigma as a barrier and sex work as a protective factor for HIV testing among trans women in Nepal. Wilson EC, Turner CM, Dhakal M, Sharma S, Rai A, Lama R, Banik S, Arayasirikul S. PLOS Glob Public Health. 2023 Mar 17;3(3):e0001098.

H2R Studies

• Mapping, Population Size Estimation, and Bio-Behavioral Surveys of PWID, MSM, TG, and FSW in 31 Districts of Pakistan in 2023-2024 (Ali Mirzazadeh: Technical Consultant)