Brian D. Earp
University of Oxford

Abstract

Feldblum et al. (2015) argue that voluntary medical male circumcision (VMMC) using the ShangRing device leads to increased sexual pleasure, universally satisfying cosmetic outcome, and virtually no delayed complications in a 2-3 year follow-up study. In this commentary, I suggest that socially desirable responding (SDR) is a likely candidate explanation for at least some of these reported findings, and I argue that this should have been controlled for using available measures. I also highlight evidence from the authors’ own study for risk compensation as a result of circumcision (including decreased condom use and an increase in number of sexual partners) and ask why this adverse outcome was not emphasized as a cause for concern. I conclude by providing 6 concrete suggestions for improving future studies on circumcision.

Introduction

This is a brief comment on the research paper entitled, "Longer-Term Follow-Up of Kenyan Men Circumcised Using the ShangRing Device" by Feldblum et al. (2015). In this paper, the authors argue that voluntary medical male circumcision (VMMC) using the ShangRing device leads to increased sexual pleasure, universally satisfying cosmetic outcome, and virtually no delayed complications in a 2-3 year follow-up study. However, certain aspects of this study's methodology as well as its broader implications seem worthy of a closer look. Indeed, by raising some basic issues here, it is hoped that the interpretability of empirical research in this area can be enhanced as the field moves forward. I ultimately suggest that validated scales, objective laboratory measures, and nuanced questionnaires, in conjunction with qualitative ethnographic interviews carried out by independent researchers, should be used to assess a wider range of sexual outcome variables, and that even longer-term follow-up data are needed before strong conclusions in this area can be drawn.

Sampling and SDR

I begin with a question about sampling. The authors state on page 3 that in order to be eligible for the study, prospective participants had to be prepared to undergo a physical examination of their genitals and “be willing to have penile photographs taken.” It seems plausible, although it is by no means certain, that men who were comparatively satisfied with their penile appearance would be more likely than those who were comparatively dissatisfied with their penile appearance to have photographs taken of their genitals. In the Supporting Information, it is stated that “If a reason for declining to make the appointment [after being contacted] is offered by the man, we will make note of the reason on the telephone log, but men will not be required to give us a reason” (p. 10). However, no discussion of how many men were initially contacted, how many declined, and what reasons they gave for declining is included in the published report. In short, the possibility of selection bias does not seem to have been adequately addressed.

Socially desirable responding (SDR) bias is also a concern (see, e.g., van de Mortel, 2008). As the authors state on page 3 of their report, all of the examiners/interviewers for the present study also worked on the initial field study during which the original circumcisions were performed. Since this was not an independent or "outside" evaluation, then, one wonders whether there was any concern on the part of the authors that (a) the interviewers may have been consciously or unconsciously motivated to elicit responses that would shed a favorable light on their own prior work (see broadly, Kaptchuk, 2003), and/or (b) that the study participants would be subject to SDR, i.e., a desire to tell the experimenters what they would (presumably) like to hear—namely that the cosmetic results were pleasing, that sexual function and/or experience was enhanced rather than diminished, and so on.

As Adams and Moyer (2015) have argued on the basis of their own investigations in Swaziland, “NGOs are known for promoting HIV prevention programmes [and therefore] being associated with them might [lead] to social desirability bias” (p. 724). Consistent with this view, Adams and Moyer report that “individual men who have undergone circumcision in Swaziland routinely complain to their peers [note: not to NGO-funded researchers] about loss of sexual pleasure following the surgery” (p. 728). If this reportage turns out to be reliable, it would stand in tension with the results reported by Feldblum et al. (2015), and this would require further explanation.

A more controlled assessment of participants’ attitudes toward circumcision, then, would entail independent, impartial evaluators who do not have a “horse in the race” in terms of funding or previous findings (see Adams & Moyer, 2015; Earp & Darby, 2015; and Martin, 1992 for further discussion). Moreover, such an assessment should be “a lengthy process”—not a one-off clinical interview, as was employed by Feldblum et al.—since “respondents may need time to reflect and consult their family, friends and partners, especially when the intervention is directed at the penis, a physically and symbolically sensitive part of a man’s body” (Adams & Moyer, 2015, p. 723; see also Martínez Pérez et al., 2015).

Given these concerns, it seems reasonable to argue that there ought to have been at least some effort to control for such confounds as experimenter demand effects (e.g., Zizzo, 2010), and/or socially desirable responding. This could have been accomplished, for example, by administering any number of well-validated measures for detecting SDR, and/or by recruiting an independent research team with expertise in ethnographic assessment.

Leading questions?

To dig into this matter in a little more detail, consider that on page 4, it is stated that participants were asked, “What is the one best thing about your circumcision?” This particular way of phrasing things calls to mind what might be referred to (in a legal context) as “leading the witness.” That is, it rhetorically presumes and therefore forces a positive answer, such that participants must think of a “best thing” about their circumcision that they can duly report to the interviewer—an interviewer who, as noted earlier, was directly involved in the original field study by which the circumcision in question was performed. But were participants also asked, “What is the one worst thing about your circumcision?”

A more appropriate question, it seems, would be something along these lines: “Have you noticed any positive or negative effects as a result of your circumcision, and if so, what did you notice?” Neutral phrasing such as this gives the participant a chance to formulate and reflect upon his own views, and thence report on them in a less biased manner.

In summary, one can see how the framing of questions (as well as which questions were asked vs. not asked) might influence participant responses. Unfortunately, the full script and questionnaires that were used in this study do not appear to be publicly available, making it difficult for other scientists to evaluate their suitedness to the relevant research question(s). On a related note, on page 7, it is stated that the researchers lacked a validated quantitative scale for assessing sexual function. But since such scales are widely available, why were none employed?

SDR and risk compensation

The issue of socially desirable responding on the part of the participants is brought into particular relief by the discussion of the authors on pages 6 and 7. As the authors write:

"Comparing participants’ responses regarding sexual behavior at this longer-term follow-up to their responses at the time of the ShangRing circumcision yielded hints of increased sexual risks among these Kenyan men. In the earlier study, 26.3% of the Kenyan participants reported two or more sex partners in the past six months; 60.6% reported using condoms about half the time or more, including 40.6% who reported always using condoms. [By contrast] 42.3% of men in the follow-up study reported two or more partners; 44.3% of the men reported using condoms half of the time or more post-MC, including 24.2% who reported using condoms all the time. These figures are difficult to reconcile with the 77% of responders in the follow-up study who stated that their condom use had increased after circumcision. The veracity of self-reported condom use in reproductive health studies is problematic" (emphasis added).

Several points are relevant here. First, 77% of participants were (apparently) willing to report increased condom use after their circumcisions, clearly a socially desirable response, and one that was perhaps influenced by the “HIV prevention and risk-reduction counseling” they were exposed to, as noted on page 2. Yet this is so despite the fact that -- when they were asked more specific questions about condom use -- the percentages actually decreased from 60.6% to 44.3% (for condom use “about half the time or more”) and from 40.6% to only 24.2% (for condom use “all the time”). Given this discrepancy, why would participants’ responses concerning other sensitive issues – such as their and their partners’ sexual pleasure post circumcision – not also be regarded by the authors as being similarly “problematic” (i.e., as likely to be influenced by SDR)?

Second, given the apparent evidence here for risk-compensation (i.e., increased risky behavior in the form of reduced condom-use and, in more than a quarter of participants as reported on page 5, increased number of sexual partners following circumcision), why were these results not highlighted in the abstract of the paper as being a cause for potentially serious concern (see Cassell et al., 2006)? Since this research is part of a Bill and Melinda Gates Foundation-funded campaign to “scale up” VMMC (see Supporting Information), presumably any evidence of risk-compensation—which would be expected to lead to greater exposure to the HIV virus, as well as to increased likelihood of transmitting it to others—would be of paramount interest.

Nevertheless, almost immediately after reporting that “more than a quarter of the men (28.1%) reported an increased number of partners post-MC [and] less than half of the men (44.3%) reported using condoms half of the time or more,” the authors blithely conclude that, “This study supports the safety and acceptability of ShangRing male circumcision during 2–3 years of follow-up” (pp. 1-2).

Sexual outcomes

Turning to sexual outcome variables, on page 4, it is reported that “Among those men who reported more pleasure, the most common reason given was more prolonged intercourse (57.4%).” Assuming that this self-reporting is reasonably accurate—that is, setting aside the concerns about SDR raised above—one is left wondering what the mechanism for this apparent effect could be. One potential lead for an explanation comes from the prior literature on this issue, where the possibility has been raised, with supporting evidence, that circumcision may increase the risk of diminished tactile sensation in the concomitantly exposed penile glans (see, e.g., Sorrells et al., 2007; but see the criticism by Waskett & Morris, 2007, as well as the reply by Young, 2007; see also Bossio et al., 2014, for a more general discussion; and see Frisch, 2012; Svoboda and Van Howe, 2013; and Earp & Darby, 2015, for further discussion of critical letters-to-the-editor by Morris and colleagues).

This would be in addition to the 100% risk that circumcision will eliminate sexual sensation in the penile prepuce itself (see Darby, 2015; Earp, 2015a), a richly-innervated genital structure with an adult surface area of approximately 30-50 square centimeters (Cold and Taylor, 1999; Taylor et al., 1996; Bronselaer et al., 2013; Bronselaer, 2013; Kigozi et al., 2009; Werker et al., 1998).

Seemingly, such an outcome (i.e., reduced sensitivity) could be regarded either as a benefit or as a harm, depending upon the particulars of each man’s baseline sensitivity, as well as how this factors into his unique sexual experience (for a related discussion, see Johnsdotter, 2013). Since the majority of the men in the present study were in approximately their mid-20s, i.e., relatively young, diminished sensitivity could plausibly be expected to correspond to “prolonged intercourse” in a way that might be regarded as being beneficial (again, setting SDR aside).

However, as these participants grow older, a reasonable hypothesis is that they may find that decreased sensitivity (assuming here for the sake of argument that this is in fact the mechanism responsible for the reported findings) becomes associated with increased sexual difficulties, rather than sexual pleasure. This suggests that future studies should (1) directly assess penile (including glans) sensitivity both before and after VMMC, and (2) that the follow-up investigation should ideally track outcomes over a much longer period, i.e., into older age, when sexual difficulties typically become more of a problem (e.g., Johannes, 2000).

Limited engagement with prior literature

On page 5, the authors state that “Male circumcision has not been shown to result in adverse changes in sexual function or satisfaction,” and cite a review article by Morris and Krieger (2013). However, this article has been criticized on methodological grounds (e.g., Bossio et al., 2014; Boyle, 2015; but see Morris and Krieger, 2015), as have the key RCT studies upon which its primary conclusions most heavily rest (see Frisch, 2012; Earp, 2015a).

As Bossio et al. (2014) state, the review by Morris and Krieger is “not a meta-analysis, thus, no statistical analyses of the data have been performed; instead, the article presents the authors’ interpretation of trends.” Problematically, “Morris and Krieger do not report the results of [their] review collapsed across study quality. The conclusion they draw – that circumcision has no impact on sexual functioning, sensitivity, or sexual satisfaction – does not necessarily line up with the information presented in their review, which is mixed” (p. 2854).

In a later publication, Bossio and colleagues go further: “Based on the Statement of Authorship in Morris and Krieger, it appears that the two authors alone composed the group who rated the articles in their review. According to the SIGN criteria that Morris and Krieger utilize, would their entire review in question not warrant a rating of ‘low quality’ based on the ‘high risk of bias’ introduced by the authors’ well documented, unconditional support of the practice of circumcision?” (Bossio et al., 2015; for further discussion, see Earp & Darby, 2015).

With respect to the RCTs alluded to above, one of which is cited by Feldblum et al. in their discussion (namely, Krieger et al., 2008), Frisch (2012, p. 313) has stated:

"Rather than blindly accepting … findings [from RCTS] as any more trustworthy than other findings in the literature, it should be recalled that a strong study design, such as a randomized controlled trial, does not offset the need for high-quality questionnaires. Having obtained the questionnaires from the authors … I am not surprised that these studies provided little evidence of a link between circumcision and various sexual difficulties. Several questions were too vague to capture possible differences between circumcised and not-yet circumcised participants (e.g. lack of a clear distinction between intercourse and masturbation-related sexual problems and no distinction between premature ejaculation and trouble or inability to reach orgasm). Thus, non-differential misclassification of sexual outcomes in these African trials probably favoured the null hypothesis of no difference, whether an association was truly present or not."

And as Krieger et al. (2008, p. 8) themselves state:

"We did not have direct observation of sexual function, partner reports, or physiologic or laboratory indicators of sexual dysfunction. [We also] did not use validated instruments, such as the International Index of Erectile Function or a recently validated sexual quality of life questionnaire for use in men with premature ejaculation or erectile dysfunction … The prevalence of sexual dysfunction is subject to the definition and period of recall used. Lack of validated instruments may prove especially difficult in assessing items such as premature ejaculation, increased sensitivity, and the enhanced ease of reaching orgasm reported in our study [note that these findings are in tension with those reported by Feldblum et al.]; the latter might be another way of describing undesired premature ejaculation."

It seems reasonable to expect that limitations such as these would be carefully emphasized in any scientific discussion of what has (or has not) “been shown” with respect to “adverse changes in sexual function or satisfaction” as a consequence of VMMC (Feldblum et al., 2015, p. 5). However, Feldblum et al. appear instead to adopt a stance of credulity toward the work of Morris and Krieger, the first of whom has argued that male circumcision “should be made compulsory” and that “any parents not wanting their child circumcised really need good talking to” (see Frisch, 2012).

What these limitations highlight, instead, is the need for validated instruments, “objective” indicators of sexual function (such as might be derived from physiological or laboratory studies), and precise definitions of sexual outcome variables that are meaningfully interpretable by research participants, as well as sufficiently well-theorized (Johnsdotter, 2013). By contrast, broad, vague, and potentially “leading” questions such as the one(s) used by Feldblum et al.—and which are typical of much of the research in this area—raise serious concerns about the haste with which some authors seem prepared to conclude that there is nothing to worry about in terms of adverse effects.

Conclusion

Taking all of these points into consideration, it seems that the authors’ headline conclusion may be premature. As they state in their abstract on page 1, their results “should allay worries that the ShangRing procedure could lead to delayed complications later than the observation period of most clinical studies.” But given the various limitations of their approach to collecting data on sexual complications in particular (i.e., no use of validated measures or quantitative scales, use of broad and arguably “leading” questions, no attempt to correct for SDR or experimenter demand effects, etc.), as well as the preliminary evidence that they themselves provide of behavioral risk compensation (decreased condom use, increased number of sexual partners), it would seem that a much more heavily qualified summary statement would have been more appropriate.

As Yavchitz et al. (2012) have shown, it is quite common for medical researchers to “spin” their findings in a favorable direction, especially in the conclusion of their articles’ abstracts, as I have argued is the case with the Feldblum et al. (2015). But it is no less problematic for being common. Indeed, this “spin” is often recycled by the media, and may in turn influence the general public as well as influential policymakers and even other scientists. Given, therefore, the ambitious scale and yet highly controversial nature of the ongoing campaign to circumcise millions of African men, it is important for scientists to stress the limitations of what they have shown in their article abstracts, not only the parts that would seem to favor “scaling up.”

In future research, I suggest that investigators working in this area should attempt to:

(1) engage more seriously with previous criticisms in the literature of studies purporting to show little or no adverse effects of circumcision on sexuality (see, e.g., Bossio et al., 2014, 2015; Earp, 2015a; Earp & Darby, 2015; Frisch, 2012; see more generally, Johnsdotter, 2013), as well as with more recent evidence that does suggest certain adverse effects (e.g., Dias et al., 2014),

(2) employ validated questionnaires, objective measures, and quantitative scales to index sexual function, sensation, and satisfaction,

(3) employ outside, impartial researchers with expertise in both qualitative and quantitative sexual ethnography,

(4) institute controls for experimenter demand effects and socially desirable responding,

(5) engage in even longer-term follow-up, tracking a wider range of sexual outcome variables (see Earp, 2015b), and

(6) dig further into the issue of risk-compensation, making sure that “absence of evidence” (based upon limited self-report surveys, for example) is not conflated with “evidence of absence.”

Hopefully, such measures will enhance the real world interpretability of future studies on circumcision and sexuality.

References

Adams, A. & Moyer, M. (2015) Sex is never the same: Men's perspectives on refusing circumcision from an in-depth qualitative study in Kwaluseni, Swaziland. Global
Public Health, 10(5-6), 721-738. Available at http://www.tandfonline.co....

Boyle, G. J. (2015). Does male circumcision adversely affect sexual sensation, function, or satisfaction? Critical comment on Morris and Krieger (2013). Advances in Sexual Medicine, 5(02), 7-12. Available at http://www.scirp.org/Jour....

Bossio, J. A., Pukall, C. F., & Steele, S. (2015). Response to: “The literature supports policies promoting neonatal male circumcision in N. America.” The Journal of Sexual Medicine, 12(5), 1306-1307. Available at http://onlinelibrary.wile....

Bossio, J. A., Pukall, C. F., & Steele, S. (2014). A review of the current state of the male circumcision literature. The Journal of Sexual Medicine, 11(12), 2847-2864. Available at http://www.ncbi.nlm.nih.g....

Bronselaer, G. A., Schober, J. M., Meyer‐Bahlburg, H. F., T'sjoen, G., Vlietinck, R., & Hoebeke, P. B. (2013). Male circumcision decreases penile sensitivity as measured in a large cohort. BJU International, 111(5), 820-827. Available at http://onlinelibrary.wile....

Bronselaer, G. A. (2013). Reply. BJU International, 11(5), E270–E271. Available at http://onlinelibrary.wile....

Cassell, M. M., Halperin, D. T., Shelton, J. D., & Stanton, D. (2006). Risk compensation: the Achilles’ heel of innovations in HIV prevention? BMJ, 332, 605-7. Available at http://www.ncbi.nlm.nih.g....

Cold, C. J., & Taylor, J. (1999). The prepuce. BJU international, 83(s 1), 34-44. Available at http://www.cirp.org/libra....

Darby, R. (2015). Risks, benefits, complications and harms: Neglected factors in the current debate on non-therapeutic circumcision. Kennedy Institute of Ethics Journal, 25(1), 1-34. Available at http://muse.jhu.edu/login....

Dias, J., Freitas, R., Amorim, R., Espiridião, P., Xambre, L., & Ferraz, L. (2014). Adult circumcision and male sexual health: a retrospective analysis. Andrologia, 46(5), 459-464. Available at http://onlinelibrary.wile....

Earp, B. D. (2015a). Sex and circumcision. The American Journal of Bioethics, 15(2), 43-45. Available at https://www.academia.edu/....

Earp, B. D. (2015b). Do the benefits of male circumcision outweigh the risks? A critique of the proposed CDC guidelines. Frontiers in Pediatrics, 3(18), 1-6. Available at http://journal.frontiersi....

Earp, B. D., & Darby, R. (2015). Does science support infant circumcision? The Skeptic, 25(3), 23-30. Available at https://www.academia.edu/....

Feldblum, P. J., Okech, J., Ochieng, R., Hart, C., Kiyuka, G., Lai, J. J., & Veena, V. (2015). Longer-term follow-up of Kenyan men circumcised using the ShangRing device. PloS One, 10(9), e0137510. Available at http://journals.plos.org/....

Frisch, M. (2012). Author's response to: Does sexual function survey in Denmark offer any support for male circumcision having an adverse effect? International Journal of Epidemiology, 41(1), 312-314. Available at http://ije.oxfordjournals....

Johannes, C. B., Araujo, A. B., Feldman, H. A., Derby, C. A., Kleinman, K. P., & McKinlay, J. B. (2000). Incidence of erectile dysfunction in men 40 to 69 years old: longitudinal results from the Massachusetts male aging study. The Journal of Urology, 163(2), 460-463. Available at http://www.ncbi.nlm.nih.g....

Johnsdotter, S. (2013). Discourses on sexual pleasure after genital modifications: the fallacy of genital determinism (a response to J. Steven Svoboda). Global Discourse, 3(2), 256-265. Available at http://www.tandfonline.co....

Kaptchuk, T. J. (2003). Effect of interpretive bias on research evidence. BMJ, 326, 1453-5. Available at http://www.bmj.com/conten....

Kigozi, G., Wawer, M., Ssettuba, A., Kagaayi, J., Nalugoda, F., Watya, S., ... & Gray, R. H. (2009). Foreskin surface area and HIV acquisition in Rakai, Uganda (size matters). AIDS (London, England), 23(16), 2209-2213. Available at http://europepmc.org/arti....

Krieger, J. N., Mehta, S. D., Bailey, R. C., Agot, K., Ndinya-Achola, J. O., Parker, C., & Moses, S. (2008). Adult male circumcision: effects on sexual function and sexual satisfaction in Kisumu, Kenya. The Journal of Sexual Medicine, 5(11), 2610-2622. Page numbers cited in this comment from the publically accessible version available at http://www.ncbi.nlm.nih.g....

Martin, B. (1992). Scientific fraud and the power structure of science. Prometheus, 10(1), 83-98. Available at https://www.uow.edu.au/~b....

Martínez Pérez, G., Triviño, D. L., Gasch, A., & Desmond, N. (2015). Towards a gender perspective in qualitative research on voluntary medical male circumcision in east and southern Africa. Global Public Health, 10(5-6), 626-638. Available at http://www.ncbi.nlm.nih.g....

Morris B. J., Krieger J. N (2013). Does male circumcision affect sexual function, sensitivity, or satisfaction? A systematic review. The Journal of Sexual Medicine, 10, 2644–2657. Available at http://www.ncbi.nlm.nih.g....

Morris, B. J., & Krieger, J. N. (2015). Male circumcision does not reduce sexual function, sensitivity or satisfaction. Advances in Sexual Medicine, 5(03), 53-60. Available at http://www.scirp.org/Jour....

Sorrells, M. L., Snyder, J. L., Reiss, M. D., Eden, C., Milos, M. F., Wilcox, N., & Van Howe, R. S. (2007). Fine-touch pressure thresholds in the adult penis. BJU International, 99(4), 864-869. Available at http://www.ncbi.nlm.nih.g....

Svoboda, J. S., & Van Howe, R. S. (2014). Circumcision: a bioethical challenge. Journal of Medical Ethics, E-letter. Available at http://www.arclaw.org/sit....

Taylor, J. R., Lockwood, A. P., & Taylor, A. J. (1996). The prepuce: specialized mucosa of the penis and its loss to circumcision. BJU International, 77(2), 291-295. Available at http://onlinelibrary.wile....

van de Mortel, T. F. (2008). Faking it: social desirability response bias in self-report research. Australian Journal of Advanced Nursing, 25(4), 40-48. Available at http://epubs.scu.edu.au/c....

Waskett, J. H., & Morris, B. J. (2007). Fine-touch pressure thresholds in the adult penis. BJU International, 99(6), 1551. Available at http://onlinelibrary.wile....

Werker, P. M., Terng, A. S., & Kon, M. (1998). The prepuce free flap: dissection feasibility study and clinical application of a super-thin new flap. Plastic and Reconstructive Surgery, 102(4), 1075-1082. Available at http://journals.lww.com/p....

Yavchitz, A., Boutron, I., Bafeta, A., Marroun, I., Charles, P., Mantz, J., & Ravaud, P. (2012). Misrepresentation of randomized controlled trials in press releases and news coverage: A cohort study. PLoS Medicine, 9(9), e1001308. Available at http://www.plosmedicine.o....

Young, H. (2007). Fine-touch pressure thresholds in the adult penis. BJU International, 100(3), 699. Available at http://onlinelibrary.wile....

Zizzo, D. J. (2010). Experimenter demand effects in economic experiments. Experimental Economics, 13(1), 75-98. Available at http://papers.ssrn.com/so....