Preterm birth time trends in Europe: the worrying case of Greece

Authors’ reply Sir, We are delighted that the Aberdeen Department of Obstetrics and Gynaecology chose to review our article during a journal club, and generated these important comments and questions. This study was set within a President’s Emergency Plan For AIDS Relief (PEPFAR) -funded HIV care and treatment programme in Kisumu, Kenya, a programme that provides support for almost 150 000 HIV-infected individuals in the Nyanza Province of western Kenya, with demographic characteristics representative of the population within the province. However, in order to isolate the impact of the loop electrosurgical excision procedure (LEEP) on HIV shedding, we set strict eligibility criteria regarding the duration of highly active antiretroviral therapy (HAART) use, adherence, and concomitant infections that may limit the generalisability. The impact of those criteria on the number of women eligible to participate is described in detail in one of the partner studies. Among women who needed to undergo LEEP and were eligible for study participation, over 95% consented. Regarding the statistical analysis, the study was originally powered to detect an effect of 1.0 log10 copies, using prior studies to estimate the standard deviation in measurements. In fact, based on the actual variance, we had 80% power to detect a minimum change of 0.66 log10 copies in the cervical viral load at each time point in the HAART group, and equal power to detect a minimum of a 1.1 log10 copies change for the ten women in the non-HAART group (both with type-I error rates of 5%, by two-sided Student’s t test). In addition to the random-effects model, we ran fixed-effects models and population-average (generalised estimating equation, GEE) models: the results were consistent across the models. Nonetheless, the results from the non-HAART women were limited by the small sample size, and would benefit from further research with a larger sample size. Regarding the lack of adjustment for baseline clinical or demographic characteristics, the readers are correct in understanding that the shedding outcomes of individuals at different time points are compared with their baseline values. As the characteristics were measured only at baseline, there would be no change in their impact on individual shedding at various time points. Although covariates such as time since HIV diagnosis and duration of HAART could confound the results across participants, in response to your query we have found that in this particular sample they were not associated with the outcome at a conservative level (P < 0.20), supporting our decision to choose this model and excluding them as covariates. The increase in concentration at 14 weeks was driven by three participants who had viral load concentrations of 5.5–6.5 log10 copies. Two of these participants had a detectable viral load throughout the study, whereas one had a measurable viral load at 6 weeks post-LEEP. Notably, all had been on first-line HAART therapy for at least 1 year, with a reported adherence >90%; however, all had a CD4+ count <200. We hypothesize that the shedding results from a lack of systemic viral suppression; however, given the small sample size and absence of follow-up serum viral loads, we are unable to confirm this. Again, we appreciate your close reading and thoughtful input.&