Jump to comment:

• Re:Alternative forms of penile foreskin cutting and HIV infection in Papua New Guinea
  David J MacLaren
  Published on: 14 March 2016
• Alternative forms of penile foreskin cutting and HIV infection in Papua New Guinea
  Kathleen E Wirth
  Published on: 17 August 2015

• Published on: 14 March 2016

  Re:Alternative forms of penile foreskin cutting and HIV infection in Papua New Guinea

  • David J MacLaren, Senior Research Fellow
  • Other Contributors:
    • Reinhold Muller

  We thank the author for her interest in our paper and are happy that she was able to exactly reproduce our findings.

  As explicitly stated - ours was an ecological study, and thus used aggregate data only. Here we comment on the author's additional statistical analysis and interpretations:

  First, the author of the reply apparently did not test "individual results" which we assume means the single, diff...

  Show More

  We thank the author for her interest in our paper and are happy that she was able to exactly reproduce our findings.

  As explicitly stated - ours was an ecological study, and thus used aggregate data only. Here we comment on the author's additional statistical analysis and interpretations:

  First, the author of the reply apparently did not test "individual results" which we assume means the single, different types of foreskin cutting. The statistical analysis stated in this context do not test the "single types of foreskin cuttings" vs no cutting, but rather refer to "circumcision" vs "dorsal/longitudinal cut or no cut at all" and "dorsal/longitudinal cut" vs "circumcision or not cut at all". These tests do not confer any additional information: in contrast, if assume that "any cut" is associated with HIV prevalence (as our results clearly suggest), then it should logically not come as a surprise that "one specific type of cut" vs "another type of cut plus no cut at all' is not significant. This approach is muddling up the baseline of "no cut" with "another type of cut" and thus will water down any association of penile cutting with HIV.

  Second, we point out that the regression coefficient per se cannot be used to judge the strength of an association (only its direction / slope of the resulting regression line). It is consequently a misconception when the author of the reply states - based on regression coefficients - that the association between condom use and HIV prevalence is stronger than that between any cuts and HIV. In fact, the reverse can be seen to be true when correct correlation coefficients (or their squares, the coefficients of determination) are used.

  Finally, more sophisticated multivariate analyses cannot - and should not - be applied to this type of ecological data to clarify things further. In contrast to making "bold" and "declarative" statements, every effort was made to restrict the interpretation of the results to "association" as opposed to a causal effect. Please see the paragraph where we explicitly highlight caution with respect to the interpretation of results in this study (at the end of the discussion). Our paper also explicitly stated that the study is of "hypothesis generating" character rather than of any "confirmative" nature.

  Conflict of Interest:

  None declared

  Show Less

  Conflict of Interest:
  None declared.

  • Back to top
• Published on: 17 August 2015

  Alternative forms of penile foreskin cutting and HIV infection in Papua New Guinea

  • Kathleen E Wirth

  I read with concern the manuscript by MacLaren DJ et al which states: "Alternative forms of penile foreskin cutting may be associated with reduced HIV infection risk in Papua New Guinea." Using data described in the manuscript, I was able to exactly replicate the authors' primary numerical finding. However, results of additional analyses not reported in the manuscript directly conflict with the key message regarding a po...

  Show More

  I read with concern the manuscript by MacLaren DJ et al which states: "Alternative forms of penile foreskin cutting may be associated with reduced HIV infection risk in Papua New Guinea." Using data described in the manuscript, I was able to exactly replicate the authors' primary numerical finding. However, results of additional analyses not reported in the manuscript directly conflict with the key message regarding a potential association between alternative forms of penile foreskin cutting and reduced HIV risk.

  First, the manuscript does not present individual results according type of penile foreskin cutting. Neither of the individual associations between HIV prevalence and prevalence of circumcision or dorsal longitudinal cut alone reach statistical significance (P=0.1756 and P=0.1327, respectively).

  Second, no results are reported for associations between HIV prevalence and other risk factors for HIV acquisition. Yet, prevalence of condom use at last sex is significantly associated with HIV prevalence (P=0.0089) and has a high coefficient of determination (R2=0.9823). Furthermore, the strength of the association appears to be as strong, or stronger, than that found for circumcision/dorsal longitudinal cut. The regression coefficient for condom use at last sex (?=-0.07848) is approximately three -fold stronger than the coefficient for circumcision/dorsal longitudinal cut (?=-0.02322).

  The manuscript makes bold statements about alternative forms of penile foreskin cutting and reduced HIV risk. Yet, these claims are not supported by the individual associations with each type of penile of foreskin cutting. Additionally, the claims are predicated in part on an apparent ruling-out of other risk factors for HIV acquisition. Applying the same analytic approach with prevalence of condom use yields a similar, if not stronger, association with HIV prevalence. Taken together, I question the appropriateness of the manuscript's declarative title and the credibility of its key message regarding alternative forms of penile foreskin cutting and reduced HIV risk.

  1. MacLaren DJ, McBride WJH, Kelly GC, et al. HIV prevalence is strongly associated with geographical variations in male circumcision and foreskin cutting in Papua New Guinea: an ecological study. Sexually Transmitted Infections. Published Online First: 30 June 2015 doi: 10.1136/sextrans-2014-051970.

  Conflict of Interest:

  None declared

  Show Less

  Conflict of Interest:
  None declared.

  • Back to top