In their analysis of the relationship of concurrency to the prevalence of HIV and sexually transmitted infections (STI), Lagarde et al.  drew on observations from over 10 000 participants in five major African urban areas. These communities vary widely in HIV prevalence (1–28.4%) as well as in values for three measures of concurrency – the kappa (κ) statistic of Morris and Kretzschmar [2,3], the number of days of overlapping relationships per individual in a 12 month period; and the ‘iic ', a new statistic, introduced by the authors, which measures the individual propensity to retain or dissolve an ongoing relationship before acquiring another one. The authors found little apparent relationship between HIV prevalence and such measures.
The data they present are strong, and challenge an emerging belief that concurrency may be an important measure of sexual transmission dynamics. To pursue such a notion further, it is worthwhile to re-arrange and re-present their data. Drawing on their Tables 1 and 2, we reconstructed the information presented for the four initial cities that participated in the multicentre study in rank order by κ (Table 3). Dakar is excluded because biological specimens were not collected.
There is almost perfect rank order correlation among the three measures of concurrency and the two bacterial STI. The exceptions are a reversal of the first and second rank for iic, the rank order for gonorrhea in men in Kisumu, whose zero prevalence was the lowest among small values, and a reversal of ranks 3 and 4 for women with chlamydia. The measure of concurrency introduced in this article, iic, does not distinguish between individuals and communities that have more versus fewer overlapping relationships, but rather measures the difference between actual overlaps and expected (random) overlaps. Its lack of perfect agreement with the other measures of concurrency, and with the frequencies of chlamydia and gonorrhea, may indicate that it contains different information, possibly related to the difference between collecting egocentric versus sociometric data.
An association between HIV prevalence and concurrency may be masked by the measurement of current (as opposed to historical) concurrency, a possibility reinforced by the fact that measures of concurrency do correlate well with acute bacterial sexually transmitted diseases (STD). In addition, the relationship between concurrency and bacterial STD was likely to have been underestimated in this type of study, as well, because of the sampling design. If it is true that individuals at genuine (not just theoretical) risk of STD compose a small proportion of the general population, and that sustained transmission takes place within even smaller (`core') groups, then the real influence of concurrency must be assessed in these smaller networks. Classic sampling methods are unlikely to reach the important populations, who are more likely not to have a fixed address, to be incarcerated, or to be unavailable for other reasons. As the authors point out, difficulties recruiting ‘… eligible men who … were never found at home despite repeated visits by the study teams’ (p. 879) suggests the undersampling of important groups. In addition, limiting, by design, the number of partners elicited may in fact truncate the upper portion of the distribution of the number of partners.
Therefore, either current concurrency is not a good surrogate for historical concurrency, in which case the question is still moot, or it is a good surrogate, in which case there is little or no association. If the latter is true, we are left with an impressive disparity in the sexual transmission dynamics of acute STI and HIV. Coupled with (i) the remarkable heterogeneity of HIV prevalence in Africa (noted in the study and in many others); (ii) the lack of association between measures of sexual activity and HIV prevalence in the multicentre study (cited by the authors in a manuscript in press); and (iii) the lack of correlation, in these four cities, between bacterial STI that presumably enhance HIV transmission, and HIV prevalence – it is difficult to avoid the hypothesis that the sexual transmission of HIV may not play the role that has been vigorously ascribed to it in Africa .
The discrepancy between the correlation of concurrency with STD and HIV leads to the consideration of an alternative hypothesis of the potential importance of the medical transmission of HIV. Considerable recent interest in this area [5,6] has focused attention on differences in medical practice, and in the exposure to non-sterile sharp equipment as mechanisms for HIV transmission. The study of Lagarde and colleagues  does not discuss the use of medical injections in these areas, nor the exposure of HIV-positive and HIV-negative individuals to potentially unsafe injections, and it would seem that such information could illuminate the dissonance between HIV prevalence and measures of sexual transmission.
1. Lagarde L, Auvert B, Carael M. et al
. Concurrent sexual partnerships and HIV prevalence in five urban communities of sub-Saharan Africa. AIDS 2001, 15: 877–884.
2. Kretzschmar M, Morris M. Measures of concurrency in networks and the spread of infectious disease. Math Biosci 1996, 133: 165–195.
3. Morris M, Kretzschmar M. Concurrent partnerships and transmission dynamics in networks. Social Networks 1995, 17: 299–318.
4. Auvert B, Ballard R, Campbell C. et al
. HIV infection among youth in a South African mining town is associated with herpes simplex virus-2 seropositivity and sexual behaviour. AIDS 2001, 15: 885–898.
5. Kane A, Lloyd J, Zaffran M. et al
. Transmission of hepatitis B, hepatitis C, and human immunodeficiency viruses through unsafe injections in the developing world: model-based regional estimates. WHO Bull 1999, 77: 801–807.
6. Simonsen L, Kane A, Lloyd J. et al
. Unsafe injections in the developing world and transmission of bloodborne pathogens: a review. WHO Bull 1999, 77: 789–800.