Article activity feed

  1. Evaluation Summary:

    This study finds that the levels of many immune markers are higher in vaginal samples in women taken after initiation of vaginal sex than before initiation of vaginal sex. This result may indicate that initiation of vaginal sex potentially influences vaginal immune responses, but it is possible that unmeasured confounding and selection bias might contribute to some of the difference across samples. This study will be of highest interest to those interested in how immune markers can change within individuals over time.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. The reviewers remained anonymous to the authors.)

    Was this evaluation helpful?
  2. Reviewer #1 (Public Review):

    This study compares concentrations of immune mediators in vaginal samples of young women who report having had or report not having had vaginal sex. The study finds that the concentration of many immune markers is higher in samples of women who report having had sex than in samples of women who report not yet having had sex. While the results are interesting and suggestive, I do not believe this result necessarily indicates that vaginal sex increases levels of these immune mediators (a causal relationship) and that the evidence presented here is strong enough to draw this conclusion.

    This study presents many methodological strengths. The sample size is amply sufficient to achieve high statistical power for this research question. A particular strength of this analysis is the relatively large number of participants who provided paired before and after sex samples. These samples are particularly valuable because stronger conclusions can be drawn from them, as their comparison is less likely to be confounded by unmeasured confounders. The statistical methods are largely appropriate for the research question, with the use of random effects to account for the correlation in multiple measures per participant.

    The reason I would not draw causal conclusions from this analysis is that there is a high potential for unmeasured confounding of the association between sex and the concentration of immune mediators. The variables that were included in the multivariable analysis were for the most part not confounders, so the authors cannot claim that their results are free from potential confounding. Confounders are in general variables which are common causes of both the exposure of interest (vaginal sex) and the outcome (level of immune markers), and which are not on the causal pathway and are not a downstream effect of the outcome (inverse causality). The only variable included that is potential confounders is age. Most other variables (pregnancy, contraception, Nugent score, Chlamydia infection, and HSV-2 seropositivity) are either potential mediators of the effect of sex or downstream effects of the level of immune markers. It does not follow that adjustment for these variables would necessarily lead to an underestimation of the causal effect, as it is possible some of these variables have complex relationships with immune mediators, so it is difficult to predict how adjusting for these variables would influence results. Some of these variables are also potentially colliders, so adjustment for them may lead to bias (see an introduction to this topic in Holmberg MJ, Andersen LW. Collider Bias. JAMA. 2022;327(13):1282-1283. doi:10.1001/jama.2022.1820). There is no consideration of general social determinants of health that are more likely to be confounders because they potentially influence both sexual behavior and the immune system: socioeconomic status, ethnicity, education, employment, housing, food security, access to health care, etc. There is overwhelming evidence that young people who are sexually active tend to have very different socioeconomic characteristics than young people who are not sexually active. It is therefore difficult to assess whether the higher level of immune markers in women who are sexually active truly represents a causal effect of sex or simply reflect differences in the type of women who have sex.

    The paired analysis also suggests that the main analysis is likely to be confounded. The evidence from the paired analysis is much stronger than the evidence from the unpaired main analysis because the paired analysis inherently adjusts for many unmeasured confounders that lead to women having sex by a certain age; the differences in paired samples are likely much closer to the causal effect of sex than the differences from the unpaired samples. We see that, in the paired analysis, the differences in levels of immune mediators before and after sex is systematically much smaller and non-significant for most immune markers. This suggests to me that the main analysis is confounded and overestimates the effect of sex on immune markers. If there is a causal effect, it is likely to be much smaller than the one estimated in the main unpaired analysis.

    The authors argue that the smaller effects seen in the paired analysis might be due to an effect of time, where samples closer to the start of sex show smaller differences. However, I would need more evidence to be convinced of this. Notably, they use a spline analysis in Figure 4 to show the effect of time since vaginal sex. However, I would have liked to see the p-values for the time-dependent spline effect, in order to see whether the data supports that a difference in slopes before and after sex significantly improves the model. I suspect many of the splines are not significant and may not lend strong support to the hypothesis that time since sex has an effect. It is however difficult to assess this visually without a formal test.

    While the results from the systematic review and meta-analysis are interesting and show that at least two other studies have shown similar results, I wonder whether these other studies do not have similar issues of confounding. The other previous studies have even fewer paired samples, so are likely to have weaker evidence than the current study.

    In summary, I think this study has some important methodological strengths in terms of sampling and study design. However, I believe the interpretation of the results should be more tempered and cautious; while there are differences in levels of immune markers in women who have had and not had sex, there is not to my mind sufficient evidence that this difference is the result of a causal effect of initiation of vaginal sex, as there is likely to be some collider bias and unmeasured residual confounding in the analysis.

    Was this evaluation helpful?
  3. Reviewer #2 (Public Review):

    This is a strong paper that evaluates an important gap in the literature, one that has been difficult to study (how mucosal immunological variables change after first sex). The authors do a thorough job addressing this with a combination of primary data and meta-analyses of other studies, providing novel insights into how sexual activity impacts mucosal immunology.

    There are several important strengths to the paper including that the field lack of critical information in this sub-group in the literature, despite their higher-than-average HIV risk. Use of multiple cohorts in the final analysis is important, as is careful consideration for possible confounders and/or mediators. Apart from minor issues, I feel as though the main claims are well-supported by the data.

    Was this evaluation helpful?