The recent discovery of the preventative potential of anti-retroviral therapy (ART) (STIs/blog/modelling ART impact)  throws into sharp relief the challenge represented for the US by the very inadequate proportion of its 1.2 million HIV+ citizens (<30%) who are virally suppressed.  Nunn & Mayer  use new geographical mapping tools to bring home forcibly the epidemiological dimension of the problem by visualizing the association which HIV+ incidence/mortality show with social status and ethnicity as reflected in residence.  The picture that emerges is of an enormous concentration of the problem in certain very circumscribed neighbourhoods.  To give just one example of what is best conveyed in the diagrams (figures 1 & 2), age-adjusted death rates rise from <11.2 per 1000 people living with AIDS (PLWHA) to 19.4-32.5 per 1000 PLWHA as one passes from a predominantly white neighbourhood with large gay population and high rates of HIV/AIDS (≥2142 per 100,000 population) to the predominantly Afro-Caribbean neighbourhood of Harlem.

For Nunn & Mayer (N&M), these visualizations raise the question whether either (1.) the allocation of resources to metropolitan areas, or (2.) the nature of the strategies employed by public health interventions, reflects the very geographically focussed nature of the problem of HIV/AIDS incidence and mortality.  Their response to the epidemiological dimension of the problem revealed by their mapping tools is to urge the importance of implementation research as a vital component of HIV initiatives.

N&M’s emphasis on viraemia suppression, rather than just HIV incidence, accords well with their insistence of the epidemiological importance of the local dimension.  Retention in care is a factor that is presumably amenable to initiatives at local level – whereas HIV incidence may owe much to transmission through sexual contacts external to the community (STI/blog/Grabowski & Gray).

Their message is in line with increasing public health interest over recent years in “program impact evaluation methods that take account of the complex interactions among interventions and between intervention packages and the context into which they are introduced” (STIs/Aral & Blanchard).  There is surely a strong argument in favour of designing interventions to take place within an evaluative framework allowing a reflection on the kind of program mix likely to be most effective in a given context.  On the other hand, N&M may be in danger of undervaluing the potential of interventions of a non-localized character that act on the socio-economic determinants of the HIV problem, and especially non-retention in care – for example, the wider provision of medical insurance (STI/blogs/ObamaCare).  It would be interesting to see how far a geographical mapping of the incidence of other health problems in New York or Philadelpia coincided with N&M’s mapping of HIV/AIDS mortality.  How far is the effect of “micro-epidemics”, conjured up by epidemiological language, just a reflection of socio-economic determinants that produce identical results wherever they happen to be present?