Data Analysis
Summary statistics were generated for demographic variables. Summary statistics were also generated for clinical characteristics, such as whether the participants had been screened for cervical cancer, the vaginal pH, the severity of the outcome among those who were screened and tested positive on visual inspection with acetic acid (VIA), and the cytology result of those who were VIA positive. Additionally, summary statistics were tabulated for preferences for cervical cancer screening, including gender preference of the provider, collection type preference (physician or self), whether the respondents would test with self-collection again, and whether they would recommend self-collection to a friend. A 2-by-2 table was generated to compare self-collected versus physician-collected specimens. From this table, performance measures and their associated 95% confidence intervals were generated, including sensitivity and specificity, predictive values positive and negative, and the proportion of concordant pairs and the proportion of concordant pairs among those pairs for which the physician-collected sample was HPV DNA testing.
Acceptability measures included: how well cared for respondents felt; how well the respondents’ privacy was handled during the tests; the extent to which respondents felt embarrassed during the test; whether the test caused the respondents genital discomfort; and whether the test caused the respondents genital pain.
Data were compared using a paired t-test, or, as appropriate, the Wilcoxon signed-rank test. Cohen’s kappa, a measure of agreement between two raters, was also computed. For Cohen’s kappa, a value near zero represented agreement being due to random chance, 0.01-0.20 represented slight agreement; 0.21-0.40, fair agreement; 0.41-0.60, moderate agreement; 0.61-0.80, substantial agreement; and 0.81-1.00 almost perfect agreement. All tests were carried out using an α=0.05 significance level. Stata version 11.2 (StataCorp, College Station, TX) was used for all analyses except for Cohen’s Kappa and its associated confidence interval, for which the ckap command, part of the rel package in R version 3.6.2 (R Core Team 2019), was used.