![Sample our Geography journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/5937/TOC-Subject-Sample-2021-Geography.jpg"})
Abstract
Billions of dollars to circumcise millions of African males as an HIV infection prevention have been sought, yet the effectiveness of circumcision has not been demonstrated. Data from 109 populations comparing HIV prevalence and incidence in men based on circumcision status were evaluated using meta-regression. The impact on the association between circumcision and HIV incidence/prevalence of the HIV risk profile of the population, the circumcision rates within the population and whether the population was in Africa were assessed. No significant difference in the risk of HIV infection based on the circumcision status was seen in general populations. Studies of high-risk populations and populations with a higher prevalence of male circumcision reported significantly greater odds ratios (odds of intact man having HIV) (p < .0001). When adjusted for the impact of a high-risk population and the circumcision rate of the population, the baseline odds ratio was 0.78 (95% CI = 0.56–1.09). No consistent association between presence of HIV infection and circumcision status of adult males in general populations was found. When adjusted for other factors, having a foreskin was not a significant risk factor. This undermines the justification for using circumcision as a primary preventive for HIV infection.
Disclosure statement
No potential conflict of interest was reported by the author.
Supplemental data
Supplemental data for this article can be accessed here.
Notes
1. The research question of the three randomised clinical trials was to determine the impact of circumcision on the incidence of heterosexually transmission of HIV from female to males. Without determining the source of the infections during these trials, it was impossible for these trials to answer their research question.
2. Men at low risk for HIV infection would be less likely to pursue circumcision as a means of lowering their risk.
3. Meta-regression allows one to create a linear regression model using the weighted natural logarithm of the odds ratio of each included study as the dependent variable in an effort to determine the impact of identified independent variables. For example, one could have a number of studies of the impact of a drug on a categorical outcome (such as survival), but the studies included in the analysis may have studied different doses of the medication. Using simple meta-analysis techniques the summary effect of all the studies together can be calculated, but using meta-regression, the impact of medication dose on the results of the individual studies results can be determined. For example, in a model with three variables the methods of meta-regression can be used to estimate the values for each of parameter (β) in the equation:
4. Studies with statistically significant findings are more likely to be submitted for publication and, if submitted, more likely to be published. Consequently, there will be studies that did not have statistically significant findings that go unpublished. If all studies were published, regardless of their findings, based on basic statistics, larger studies would be expected to be closer to the true value and smaller studies would be expected to show more variation. If one plots the size of a study on the y-axis and outcome of the study on the x-axis, a plot that looks like an inverted funnel would be expected. With publication bias, there is often a paucity of studies in the left lower portion of the plotting (small studies with non-significant results). The measures of publication bias attempt to quantify any asymmetry in the funnel plot.
5. For example, for a high-risk population in Africa with a circumcision prevalence of 70%, the predicted natural logarithm of the odds ratio of risk of HIV in intact men versus circumcised men would employ the equation for the trivariate model in . For this population x1 and x2 would equal 1 and x3 would equal 0.70. The estimate would be (0.2337 × 1) + (0.6223 × 1) + (0.6269 × 0.70)− 0.2442, or 0.2337 + 0.6223 + 0.4388 – 0.2442, or 1.0506. Converting from the natural logarithm gives a predicted odds ratio of 2.86.
6. In a linear regression equation for the natural logarithm of odds ratio for HIV incidence or prevalence in men who are not circumcised where x3 = 1 if the prevalence of circumcision in a population was 100%, the estimated second-order equation would be: y = 1.3794 (SE = 1.0319) x3 − 0.3891 (SE = 0.9861) (x3 × x3) − 0.03118 (SE = 0.2015). The t value for the second-order parameter is −0.39, which is not statistically significant.
7. In a linear regression equation for the natural logarithm of odds ratio for HIV incidence or prevalence in men who are not circumcised where x1 = 1 if the study took place in Africa and x2 = 1 if a population at high risk for HIV infection was studied, the equation to evaluate an interaction between x1 and x2 would be: y = −0.00225 (SE = 0.2426) x1 + 0.2870 (SE = 0.3005) x2 + 0.5841 (SE = 0.3301) (x1 × x2) + 0.2071 (SE = 0.2286). The t value for the interaction parameter is 1.77.
8. The fragility index (Walsh et al., Citation2014) for the three African randomised clinical trials (Auvert et al., Citation2005; Bailey et al., Citation2007; Gray et al., Citation2007) ranged from four to seven, which indicates that the results of these trials were not very robust.
9. If the rate of disease in the control group was 2% and the rate of disease in the treated group was 1%, the relative risk reduction, the ratio of these percentages, would be 50%. The absolute risk reduction is the difference between these percentages, or 1%.
10. Since the cost of circumcision is up front, the yearly costs of ART treatment should be discounted. At a 3% discount rate, cost of the circumcisions to prevent one case of HIV infection would pay for two people to be on ART for 52 years or three people to be on ART for 25 years. At a 5% discount rate, ART could be provided for three people for 39 years or four people for 21 years. These estimates are based on the marginal cost per patient as one would expect the delivery of both interventions to require a similar amount of infrastructure and support staff.
11. At least one attendee noted that the resolutions at Montreux were railroaded through with a minimal of discussion (Dowsett & Couch, Citation2007).
12. When an analysis is made, some have suggested that the harms of circumcision be compared to the harm of ART treatment. This is a false comparison: harms from a procedure from which only 1% will see a benefit (and provides no protection for one's sexual partners) versus the harms from a treatment from which nearly 100% of those treated will see a benefit and will provide a strong benefit to their sexual partner.
13. A primary objective of the American Academy of Pediatrics's new policy was to shore up reimbursement for the procedure from third-party payers (Task Force on Circumcision, Citation2012).