Estimating the added value of PrEP where it is not the only prevention tool in the box

LeVasseur & Welles (L&W) model the potential population-level impact of PrEP in combination with other prevention strategies. They seek to quantify the additional benefit of PrEP at various levels of uptake (0%-25%) in terms of prevented infections specifically attributable to that intervention in a range of contexts involving its combination with one or more of the other currently available prevention strategies – i.e. consistent condom use; seroadaption; treatment as prevention (TasP). Unlike another modelling study of PrEP recently discussed in this blog (Modelling the potential effectiveness of PrEP(STI/Blogs)) this study does not evaluate PrEP against an enhanced provision of those other prevention strategies as an alternative way of deploying equivalent resources; it merely examines the additional impact of PrEP in the context of the provision of various combinations of those strategies at the current level. What this demonstrates very clearly is how the proportion of prevented cases specifically attributable to PrEP at various levels diminishes to the extent that PrEP is being practised along with the other preventative strategies. So, for example, assuming that the group, into which PrEP is introduced, currently practises no such strategy, PrEP uptake at 25% will prevent 30.7% of infections; but, if that group is already implementing condom use, seroadaption and TasP at average current levels, these strategies in the absence of PrEP will already be preventing 72.2%, and implementation of PrEP at 25% uptake will prevent only an additional 5% of infections.

The great limitation of this model, of course, is that it assumes the various strategies (PrEP, condoms, etc.) are independent of each other – or that ‘their interactions are stochastic’. However, the authors argue that, if we ‘assume both 100% adherence to, and 100% effectiveness of PrEP’, ‘these behavioural changes would not bias our results’; moreover, that their results helpfully represent ‘the most benefit we can likely see in the reduction of HIV’. It is evident from these comments that the independence (or otherwise) of the various strategies is seen chiefly from the angle of its importance in respect to ‘risk compensation’ (i.e. that individuals on PrEP are at risk of abandoning other risk-reduction strategies).

But the independence of strategies assumed by the model also presupposes that PrEP-users do not differ from non-users in respect to their practice of other strategies (e.g. seroadaption or consistent condom use). If, on the other hand, potential PrEP-users turn out to constitute an untypical group in this respect (e.g. if they are much less likely than non-users to practise other strategies), then we would expect the overall impact of introducing PrEP to be more beneficial than the model predicts on the scenario of combining PrEP with other prevention strategies at current levels. The authors argue that their model population is not heterogenous ‘since we only consider high-risk MSM’. But it is not clear to me that this deals adequately with the issue of the potential heterogeneity of a real-life MSM population in respect to practice of existing prevention strategies. In practice, many studies suggest that potential PrEP users are more likely than other MSM to be high-risk (Lee & Chang (STI/Blogs); Girometti & Whitlock (STI); Aghaizu & Nardone (STI); Holt & de Wit (STI)). Indeed, Griensven & Lo (STI) and Reyniers & Laga (STI) argue that PrEP could offer a solution in regions, like large cities in the Pacific rim (Griensven & Lo (STI)), where conventional strategies have little effectiveness because infection is so rapid that numerous transmissions will normally have taken place before effective engagement with TasP.