The risk network approach to HIV detection: something like contact tracing?

There has been considerable debate on the most effective and cost-effective means of accessing untested HIV- or STI-infected individuals. One frequently canvassed strategy is that of respondent driven sampling (RDS). This involves issuing suitable ‘seeds’ (e.g. recently diagnosed MSM) with coupons to distribute to others in their sexual or social networks.  Wei & Raymond (STI) in a recent study of black MSM in San Francisco compare RDS very favourably with time location sampling (TLS) as a method of recruitment of those most likely to be at risk. Similary, Rosenberg & Miller (STI) in a small study in Malawi observe the superior effectiveness of infected, over non-infected seeds in discovering untested individuals (31% vs. 11%). Much, of course, depends on the target group; so, for example, Weir & Chen (STI) find in favour of ‘venue-based’ sampling over RDS in the case of Chinese FSW.

Smyrnov & Friedman (S&F) report a Ukrainian intervention which used HIV-infected seeds to access networks of people who inject drugs (PWID) and other at-risk individuals over the period 2013-16, and resulted in 1,252 tests (Transmission Reduction Intervention Project, or TRIP). The intervention is a refinement of RDS which the researchers describe as a ‘risk network’ approach. Seeds – some recently-infected, others longer-term infected – recruit through networks that are relatively strictly defined (by comparison with normal RDS). Such networks are restricted to: persons with whom the seeds had performed sex or injected drugs; people present where this had taken place; those at small-sized venues frequented by these participants. The researchers then compare the outcomes of their own study (undiagnosed HIV-positive as a proportion of persons tested, and cost per person identified) with the outcome of a more traditionally conceived RDS trial amongst PWID in Odessa involving 400 persons, and a large (13,936) community-based ‘Outreach Testing’ programme, also amongst PWID.

The proportion of undiagnosed positives identified by TRIP (14.6%) was much higher than it was for the RDS (5%) or Outreach (2.4%). This gives odds rations for TRIP of 3.25 as against conventional RDS and 7.03 as against Outreach. Within TRIP the networks seeded by recently infected persons contained a greater proportion of undiagnosed positives (16.3%) than those seeded by the longer-term infected (12.2%). As for cost per HIV positive identified, this the researchers estimate at $250 for TRIP as against $387 for RDS and $653 for Outreach. They conclude that interventions prioritizing networks of the recently effected offer the most efficient way of detecting undiagnosed HIV positive persons.

On the face of it, the ‘risk-network’ approach looks rather like an enhanced version of the well-known practice in sexual health of contract tracing – a ‘contract tracing plus’. It may seem evident, but contact tracing has shown itself to be a very effective intervention (Rayment & Sullivan (STI)); perhaps it offers a basis on which to model larger-scale interventions. The study of S&F would at least suggest so.