LINKING CIRCUMCISION AND HIV/AIDS
Against this historical backdrop, the HIV/AIDS pandemic is merely the latest incarnation of a 130-year-old pattern of circumcision promotion by a small, but very influential, portion of the medical community in circumcising first world countries.7-12 The idea that circumcision can prevent AIDS was developed by Fink, a long-time advocate of mass circumcision. Fink introduced the hypothesis in a letter to the New England Journal of Medicine,13 which he later admitted was based purely on speculation rather than hard data.14 Seeking to capitalise on public anxiety over the spread of HIV, other advocates of mass circumcision sought to develop Fink’s hypothesis by producing geographical analyses of Africa, which studied maps rather than men, which they argued could be used to legitimise mass circumcision in the US. Using decades-old anthropological data and extrapolating HIV incidence rates, an association between the foreskin and HIV was suggested.15 Next came a number of observational studies suggesting an association between the foreskin and an increased risk of HIV infection in men, mostly in Kenya, who exhibited high-risk behaviours.16, 17 These studies compared disparate populations that were distinguishable on other relevant independent variables, such as religion, social class, tribal affiliation, sexual practices and presence of genital ulcer disease. Subsequently, the degree of association of the initial studies and the infectivity attributed to the foreskin could not be replicated in the same population by the same team of investigators.18
Partner studies in which associations were suggested between the HIV status of a woman and the circumcision status of her sexual partner have overall failed to support an association.19, 20 Likewise, general population surveys have, as a whole, failed to demonstrate a strong association. 19, 20 It is only when limiting the analysis to African studies and using values obtained following multivariate analysis that an association can be extracted from these studies.21 One of the challenges in interpreting these various observational studies is determining whether circumcision status may be a risk factor or a marker for other risk factors. The fundamental flaw in multivariate analysis is that to be accurate it is assumed that the variables controlled for are independent of one another.Many of these variables, including sexual, religious and hygienic practices, as well as economic status, appear to be linked to tribal affiliation, which in turn is strongly correlated with circumcision status.22 These multiple, highly-correlated, confounding factors influencing sexual behaviours and HIV susceptibility create a co-linearity problem that can make these regression models unstable and yield unreliable results. Consequently, without more reliable data it is irresponsible to place blame for HIV’s spread on normal penile anatomy.
Many of the studies suggesting an association between circumcision status and HIV infection tested a wide assortment of factors, fishing for significant risk factors without making the proper adjustments for multiple comparisons. As a result, many of the positive associations asserted could be due merely to oversampling. Meta-analysis has demonstrated significant between-study variability independent of the vagaries of geography, study design and circumcision’s prevalence within a community,19 and has suggested the possibility of publication bias, whereby studies failing to find a correlation between circumcision status and HIV infection are either never submitted for publication or are passed over by editors.20 Observational studies, when compared to randomised controlled trials, have been shown to consistently overestimate odds ratios by 30%.23 In light of this unexplained heterogeneity and possible publication bias, any conclusion based on these observational studies should be viewed with scepticism.24
On the basis of weak scientific evidence, many circumcision proponents have called for universal circumcision in Africa.2-4
Although the next logical step in this scientific inquiry might be a randomised controlled trial, problems exist with such a project. A trial involving permanent amputation of a body part, the benefit of which is largely unproven, is fraught with ethical pitfalls and would not be likely to be approved in a developed nation. The subject would certainly need to be fully informed, and the potential for manipulation of the information provided would need to be prevented. Studies have already demonstrated that pro-circumcision propaganda can effectively influence attitudes regarding circumcision.25-30 Despite the clear ethical contraindications, two randomised controlled trials to determine if a relationship exists between HIV status and circumcision to be undertaken in Africa have received funding from the US National Institutes of Health. Both studies are markedly overpowered so as to find a statistically significant difference where a significant clinical difference may not exist.31-33 A report in the lay literature suggests that compliance following randomisation may pose a serious threat to the study’s completion.34 Therefore, the subsequent analysis must employ an intent-to-treat approach, as otherwise serious bias would be introduced into the results.