Development of a multivariable prediction model for anal high-grade squamous intraepithelial lesions in persons living with HIV in Puerto Rico: a cross-sectional study

Risk assessment and the prevention of anal cancer among people living with HIV

“…It is a strength that Soto-Salgado and colleagues used a sex-stratified approach for their risk prediction model development. 4 For women living with HIV, only past abnormal cytology was statistically significantly associated with HSIL in multiple logistic regression models, whereas for men, a self-reported history of gonorrhoea was independently associated with HSIL (in addition to abnormal cytology). It was somewhat surprising that detection of high-risk HPV infection was not statistically significantly associated with HSIL for either sex, although the direction of association was positive for both sexes.…”

“…In this issue, Soto-Salgado and colleagues report on a possible approach to prioritize which patients should go forward for thorough assessment. 4 The authors developed a risk prediction model for anal HSIL in a clinical population of 522 people living with HIV in Puerto Rico, a territory that experiences higher rates of HIV and anal cancer relative to the continental United States. Most referred patients to the Anal Neoplasia Clinic of the University of Puerto Rico Comprehensive Cancer Center were exhibiting symptoms or had a history of abnormal anal cytology such that, strictly speaking, this was not a screening study.…”

“…Altogether, the reported multivariable risk prediction models correctly discerned risk for HSIL with a 73% probability among women and a 69% probability among men. 4 There was considerable missingness in the available chart-abstracted data, which the authors attempted to mitigate with multiple imputation analysis, although they found it provided no improvement in model prediction. For future explorations, standardized measures would be preferable to medical chart data not collected for this purpose, as would measures not dependent on self-report that may be prone to recall error and social desirability bias.…”