I read this article with professional anticipation. I was originally impressed by the large number of participants and by the researchers’ success in securing adequate numbers for follow-up (at least in the first year). However, on closer scrutiny of the data I find the authors’ conclusions rather debatable.

Table 2 (displayed) offers the readers an excellent overview of the outcome. Particularly the last column shows an impressively improved sexual health over the 2 years, especially in the control group! The non-specific sexual health-promoting elements in the programme must have been extraordinarily effective. Men with dysfunctions such as erectile failure, ejaculation problems and penetration difficulties were cured almost without exception, and the reported level of satisfaction also improved immensely. Comparatively, the circumcised group improved in only two of these four problem areas. ‘Low desire’ was the fifth problem that was cured in a significant number of controls. In the experimental group, P was just 0.01 lower, but a strict application of statistics inevitably implies that the level of significance is reached in the control group and not in the experimental group. Dyspareunia was the sole complaint to break the general pattern, with maximum results for the circumcised and below-significance value for controls. I return to this point later.

The inevitable conclusion therefore must be that men who are not in perfect sexual health will benefit from the non-surgical elements in this project’s protocols, and that improvement is considerably hampered by circumcision. This irrefutable outcome is all the more remarkable because closer assessment of the groups shows that randomization has not been very successful, as the controls were definitely in poorer sexual health at the start of the project. Table 2 presents the exact numbers of those with any of the six dysfunctions:

Low desire4131
Low satisfaction3828
Erectile failure3521
Penetration  problems4634
Ejaculation  problems1211

The difference between arms (controls and circumcised) was not significant for any of these dysfunctions (although for erectile failure it was close). However, there must be statistical methods to prove that these six skewed distributions (all in favour of the experimental group) cannot purely be attributed to coincidence. After all, when a coin is tossed six times, and six times it is tails, one would definitely be alerted to check the coin for bias.

It can therefore be assumed that randomization was imperfect and that the power of this research is thus further limited, because it could not be blinded. Indeed, not even single-blinded; the authors do not conceal the unfortunate fact that during the follow-up interviews the interviewer was (in most cases) aware of the interviewee’s circumcision status. This implies a serious limitation of the study’s objectivity. In the heading, the study is definitely overrated as a ‘randomized controlled trial’ (RCT).

Consequently, I am tempted to challenge the trend-breaking outcome on the genital pain issue, especially the opposite outcomes of this question and that of penetration difficulty is hard to understand. One would expect that, in at least a certain proportion of cases, penetration difficulties are caused by male genital pain. It is conceivable that the subject of pain is understood as pivotal by the circumcised interviewees, so a tendency towards bias (pleasing the interviewer) could be most outspoken for precisely this question.

In summary, Kigozi et al. are to be warmly congratulated on their tremendous success in sexual health promotion and I am looking forward to a more extensive article on the non-surgical part of their project. However, they should acknowledge that circumcision seriously jeopardizes their preventive efforts, especially when men with sexual problems are included without further consideration.

Some of this project’s control group participants might count their blessings, that they retreated from their original decision to have a circumcision at the start of their participation. If control group participants were really keen on the surgery, we would have expected the control group to show fewer discontinuations at the end, i.e. the moment they are entitled at last to have their circumcision. Numbers contradict this; withdrawal rates were equal in the experimental and the control group. One could say that many withdrawals in the control group at the 2-year follow-up ‘voted with their feet’ against circumcision.

Finally, being reasonable, for a group so highly motivated as these participants appear to have been, there are alternative, more conservative methods available for preventing HIV and sexually transmitted disease.