Data analysis

Ten per cent of data were checked independently for errors by a member of the research team who was not involved in the initial data entry. Error rates were substantially below the prespecified cut‐off for no further action to be taken (<1% error for all outcomes). Demographic characteristics were compared between the six groups using one‐way ANOVA and Chi‐squared tests as appropriate.

We compared the demographic characteristics of responders and nonresponders (including age, test results, number of previous screens, NHS site and IMD quintile) which revealed small variations. See Supporting Information Table S1 for an overview of nonresponder demographic characteristics. To adjust for the fact that our approached sample may not have been fully representative of the screening population in the pilot sites, we generated and applied population weights based on age group (24–34; 35–44; 45–54; 55–65) and IMD quintile within each test result group. With permission from the Office for Data Release, we used data from 955,387 women who attended HPV primary screening (and primary cytology for the control group) within the NHSCSP in the five sites included in our study in 2017–2018 to calculate the weights.

For each of the primary outcomes (anxiety and distress), we compared the mean scores between the six groups using univariate regression analysis. Further to this, multiple regression analysis was performed to adjust for confounding factors: age, IMD score, ethnicity, marital status, education, number of previous cervical screens and NHS site. Results are presented as mean difference (MD) compared to the control group, along with 95% confidence intervals. We also present descriptive mean values and standard deviations for each of the six groups. For the secondary outcomes (very high anxiety, case‐level distress, worry about cancer, concern and reassurance about results), we fitted both univariate and multiple logistic regression models adjusting for the same confounding factors. Results are presented as odds ratios, indicating the odds of the outcome for each of the groups relative to the control group, with 95% confidence intervals. Due to skewed responses to the concern and reassurance items in the control group, we used the HPV positive with normal cytology group as the reference category for analyses of these outcomes.

Data completeness was >96% for the majority of outcomes and factors, with the exception of anxiety (89%) and IMD (93%). We used multiple imputation assuming data were missing at random to account for missing data. The imputation model included primary outcomes and all sociodemographic factors, which we assumed included all predictors of missingness. The final models were derived by fitting a regression model including all confounders, and estimates were combined using Rubin's rules.20

Demographic characteristics have been presented using nonweighted data. All primary and secondary results have been adjusted using the weights described above and using the imputed data. Supporting Information Table S2 presents the results of the primary and secondary analyses using unweighted data. Analyses were carried out using Stata v1521 and SPSS v25,22 and p‐value <0.05 was considered statistically significant.