We completely agree with the conclusion of Nosyk et al.  that ‘optimizing implementation of the HIV treatment as prevention strategy’ is essential. However, we believe that true optimization will require further research into the generalizability of the individual-level effectiveness of treatment as prevention (TasP) beyond HIV-infected mothers and stable, serodiscordant, heterosexual couples, as well as the scalability of individual-level effects to population-level HIV incidence. We have two main concerns about the application of ‘the Bradford-Hill criteria’ to apparently obviate the need for this further research.
First, a ‘checklist’ approach to causal inference is generally questionable, and specifically, the list that Sir Austin Bradford Hill proposed is notoriously flawed for establishing causality [2–6]. Even Hill himself cautioned that ‘none of my nine viewpoints can bring indisputable evidence for or against the cause-and-effect hypothesis.’ Only one of the items – temporality – is a necessary (though not sufficient) feature of a causal effect . Each of the other ‘viewpoints’ (Hill did not call them ‘criteria’) has fundamental limitations and is neither a necessary nor sufficient causal criterion. For example, a biological gradient between a presumed cause and effect can be partially or wholly due to a confounding factor that has a dose–response relationship with the disease . In the case of the population-level impact of increased antiretroviral treatment (ART) use, a concomitant behavioral intervention could be such a confounder.
Although the Bradford Hill criteria cannot prove causality, they can provide a valuable framework for evaluating the available evidence for a causal effect and identifying gaps in current knowledge. However, our second main concern about the argument put forth by Nosyk et al.  is that, rather than evaluating a single causal question, their approach attempts to simultaneously evaluate three: the individual-level effect of TasP within heterosexual couples and mother–child pairs, the individual-level effects in other populations, and the population-level effect of expanded ART use on HIV incidence. An approach that examines all of these related – but fundamentally different – questions with a single pass of a causal lens conflates the three distinct questions and obscures gaps in the evidence for each individual question.
There is clear evidence for a strong individual-level effect of TasP for stable, serodiscordant, heterosexual couples and for prevention of mother-to-child transmission; however, the magnitude of an individual-level effect in MSM and IDUs is much less certain. ART may be less effective at preventing transmission among MSM and IDUs, given higher transmission probabilities associated with unprotected anal sex  and parenteral exposure , a potentially greater possibility for drug resistance , and differences in the pharmacological performance of antiretroviral agents in the rectum . Most of the existing evidence for TasP among MSM and IDUs is based on ecological studies, which are inappropriate for drawing causal conclusions about individual-level effects . Additionally, as noted (but essentially dismissed) in the authors’ discussion of consistency , the results of these ecological studies have been inconsistent. Upcoming observational studies among serodiscordant MSM couples  and IDU networks  will provide valuable information on individual-level TasP effects in these groups.
Conclusions about the population-level impact of ART based on current evidence are also tenuous. The only empirical examinations of population-level effects are based on ecological studies with inconsistent results and critical limitations, including the potential for confounding and the use of proxy measures – such as community viral load and new HIV diagnoses – for ART exposure and HIV incidence [11,14]. Additional empirical investigations are needed to evaluate the potential effects of behavioral disinhibition , transmission during acute HIV infection , and imperfect retention throughout the HIV treatment cascade , all of which may undermine the population-level impact of TasP to varying degrees across settings. The results of ongoing cluster randomized trials  will provide important information on the population-level effects of TasP and will guide future implementation of treatment-based HIV prevention strategies.
We wholeheartedly agree that ART has great promise for both preventive and clinical benefits, and that universal ART access is absolutely imperative. However, we believe that there is still much to be learned about the individual-level effect of TasP across fundamentally different risk groups, as well as the population-level impact of expanded ART. Numerous ongoing studies will shed light on some of these critical questions, as well as some of the potential pitfalls and unintended consequences of TasP, allowing true optimization of TasP strategies.
Conflicts of interest
There are no conflicts of interest.
1. Nosyk B, Audoin B, Beyrer C, Cahn P, Granich R, Havlir D, et al. Examining the evidence on the causal effect of highly active antiretroviral therapy on transmission of HIV using the Bradford Hill criteria
2. Lippincott Williams & Wilkins, Rothman K, Greenland S, Lash TL. Modern epidemiology
3. Phillips C, Goodman KJ. The missed lessons of Sir Austin Bradford Hill
. Epidemiol Perspect Innov
4. Ward AC. The role of causal criteria in causal inferences: Bradford Hill's
. Epidemiol Perspect Innov
5. Höfler M. The Bradford Hill considerations on causality: a counterfactual perspective
. Emerg Themes Epidemiol
6. Phillips C, Goodman KJ. Causal criteria and counterfactuals; nothing more (or less) than scientific common sense
. Emerg Themes Epidemiol
7. Powers K, Poole C, Pettifor AE, Cohen MS. Rethinking the heterosexual infectivity of HIV-1: a systematic review and meta-analysis
. Lancet Infect Dis
8. Baggaley R, Boily MC, White RG, Alary M. Risk of HIV-1 transmission for parenteral exposure and blood transfusion: a systematic review and meta-analysis
9. Truong H-H, Kellog TA, McFarland W, Louie B, Klausner JD, Philip SS, Grant RM. Sentinel surveillance of HIV-1 transmitted drug resistance, acute infection and recent infection
. PLoS One
10. Zuckerman RA, Whittington WL, Celum CL, Collis TK, Lucchetti AJ, Sanchez JL, et al. Higher concentration of HIV RNA in rectal mucosa secretions than in blood and seminal plasma, among men who have sex with men, independent of antiretroviral therapy
. J Infect Dis
11. Smith M, Powers KA, Muessig KE, Miller WC, Cohen MS. HIV treatment as prevention: the utility and limitations of ecological observation
. PLoS Med
12. World Health Organization. Geneva, Switzerland: 2011. WHO and U.S. NIH Working Group Meeting on Treatment for HIV Prevention among MSM: What Additional Evidence is Required? http://www.who.int/hiv/pub/msm_meeting_report.pdf
. [Accessed 30 April 2013].
14. Miller W, Powers KA, Smith MK, Cohen MS. Community viral load as a measure for assessment of HIV treatment as prevention
. Lancet Infect Dis
15. Bezemer D, de Wolf F, Boerlijst MC, van Sighem A, Hollingsworth TD, Prins M, et al. A resurgent HIV-1 epidemic among men who have sex with men in the era of potent antiretroviral therapy
16. Powers K, Ghani AC, Miller WC, Hoffman IF, Pettifor AE, Kamanga G, et al. The role of acute and early HIV infection in the spread of HIV and implications for transmission prevention strategies in Lilongwe, Malawi: a modelling study
17. Gardner E, McLees MP, Steiner JF, Del Rio C, Burman WJ. The spectrum of engagement in HIV care and its relevance to test-and-treat strategies for prevention of HIV infection
. Clin Infect Dis
18. Boily MC, Mâsse B, Alsallaq R, Padian NS, Eaton JW, et al. HIV treatment as prevention: considerations in the design, conduct, and analysis of cluster randomized controlled trials of combination HIV prevention
. PLoS Med