To the Editor—A simulation study on the impact of vaginal-rectal ultrasound examinations on infectious risks in France was published recently by Leroy and colleagues.Reference Leroy, M’Zali, Kann, Weber and Smith ¹ Although statistical methods with Monte Carlo simulations could be contributive, we would like to raise some points which might limit the interpretation of their results.

The uncertainty of several parameters was possibly very wide, and simulation did not take such variability into account. The probability that a pathogen lingered on the probe after cleaning and disinfection was derived from data on bacterial agents in 2 single-center studies,Reference Kac, Podglajen, Si-Mohamed, Rodi, Grataloup and Meyer ^{2 ,} Reference Kac, Gueneret and Rodi ³ whereas most infections simulated by Leroy et al were viral. The probability of probe contamination from an infected patient was extracted from observational data on sexual intercourse. However, the probability of transmission differed according to type of sexual intercourse, inoculum or viral load.Reference Hughes, Baeten and Lingappa ⁴ Sexual exposure was most probably very dissimilar from endocavitary ultrasound exposure. With hepatitis C virus, the rate of transmission differed strongly between infection observed among drug usersReference Thorpe, Ouellet and Hershow ⁵ and patients after nosocomial exposure, such as hemodialysis.Reference Petrosillo, Gilli and Serraino ⁶ Similarly, with human immunodeficiency virus, the probability of infection after accidental bloodReference Henderson, Fahey and Willy ⁷ and male-to-female sexual exposuresReference Hughes, Baeten and Lingappa ⁴ is distinct with 0.003 and 0.0019 probability densities, respectively. Sensitivity analyses should have been conducted to properly interpret the results.

In a hypothetical cohort of 4 million exposed patients in France,Reference Leroy, M’Zali, Kann, Weber and Smith ¹ the authors ascertained that a mean (SD) of 40 (20) would be infected by human immunodeficiency virus and 151 (63) by hepatitis C virus annually. Recently, our group studied a French prospective, observational, hospital-based cohort of 16,474 individualsReference Bénet, Ritter and Vanhems ⁸ and found that the incidence of human immunodeficiency virus seroconversion was 0 (n=0) per 10,000 patient-years in patients with endocavitary probe exposure within 12 months before testing and 6.7 (n=13) in nonexposed patients (log-rank test: P=.64). The incidence of hepatitis C virus seroconversion was 16.1 (n=1) per 10,000 patient-years in patients exposed to endocavitary probes and 23.4 in nonexposed patients (log-rank test: P=.69).

In a letter published elsewhere,Reference Bénet and Vanhems ⁹ our group underlined that statistical analysis of a previous meta-analysis by Leroy,Reference Leroy ¹⁰ based on 2 published studies, would be questionable owing to lack of weighting according to study size. However, similar data were analyzed, again with a dearth of details regarding the calculation of pooled prevalence.Reference Henderson, Fahey and Willy ⁷ We agree with Leroy et alReference Leroy, M’Zali, Kann, Weber and Smith ¹ that the issue of probe contamination is important and could be a public health concern, particularly with human papillomavirus infection related to endocavitary ultrasound exposure. Additional sensitivity analysis would have improved the accuracy of estimations in the present study.Reference Leroy, M’Zali, Kann, Weber and Smith ¹ Appropriate prospective investigations are needed with a view to proposing the best preventive measures for patient safety regarding these exposures.