Using stated preference methods to inform public health decision making

Croatia has recognised the importance of prevention programmes in the field of public health, although their effectiveness is not satisfactory due to the low population response, which has a negative impact on the rationalisation of public spending. One of the possible solutions is to consider the stated preferences of the target population. Therefore, the main objective of this study is to evaluate the use of stated preference methods in improving public health prevention programmes. For the purpose of the study, a questionnaire was designed using three different methods - Discrete Choice Experiment (DCE), Contingent Rating method (CR) and Best-Worst Scaling method (BWS). This study shows that the attributes of the Croatian cervical cancer screening programme are significantly associated with the respondents' utility level, which in turn is related to women's response. Since BWS, DCE, and CR measure the same construct - utility - we can say that convergent validity partially confirms the external validity of the methods. The author concludes that it is necessary to implement market principles, i.e. the demand-side analysis using stated preference methods, in the planning, implementation and re-evaluation of public health programmes.

Real-world effectiveness of primary screening with high-risk human papillomavirus testing in the cervical cancer screening programme in China: a nationwide, population-based study

Background Randomized controlled trials have shown a higher sensitivity and longer negative predictive value of high-risk human papillomavirus (HPV) testing than cytology for cervical cancer screening; however, little is known about the effectiveness of HPV testing in middle-income countries. Understanding the characteristics of HPV testing may increase the priority of HPV testing in health policies. The study aims to evaluate the effectiveness of HPV testing in the national cervical cancer screening programme in China. Methods We performed a nationwide, population-based study using individual data from the national cervical cancer screening programme in rural China between 2015 and 2017. The analyses included 1,160,981 women aged 35–64 years who underwent cytology alone or high-risk HPV testing with cytology or genotyping triage. The main outcome was cervical intraepithelial neoplasia 2 or worse (CIN2+). We used multivariate logistic regressions and performed sensitivity analyses with propensity score matching to compare the screening positive, colposcopy referral, detection rate, and positive predictive value (PPV). Results The screening positive rates for HPV testing and cytology were 10.1% and 4.0%, respectively. The per protocol colposcopy referral rate of HPV testing was significantly lower than that of cytology (3.5% vs 4.0%), and this difference was mostly due to the low referral threshold of cytology (≥ASC-US). Overall, HPV testing detected more CIN2+ (5.5 vs. 4.4 per 1000, adjusted odds ratio [aOR]=1.18, 95% confidence interval 1.11–1.25) and had a higher PPV (13.8% vs 10.9%, aOR 1.29, 95% CI 1.21–1.37) than cytology. The colposcopy referrals of HPV testing in comparison to cytology differed by income status; it significantly increased in lower-middle-income areas (3.7% vs 3.1%, aOR 1.21, 95% CI 1.17–1.25) and significantly decreased in upper-middle-income areas (3.4% vs 4.9%, aOR 0.69, 95% CI 0.67–0.71). Sensitivity analyses demonstrated the reliability and robustness of the results. Conclusions The introduction of HPV testing could improve both the CIN2+ detection rate and efficiency of cervical cancer screening programme, supporting the introduction of primary screening with high-risk HPV testing in China. Further study is needed to investigate the long-term effect of this change.

A nationally quasi-experimental study to assess the impact of partial organized breast and cervical cancer screening programme on participation and inequalities

Background Organized breast and cervical cancer screening programme could only provide services at no cost for a fraction of socioeconomic-deprived women in China and other low-resource countries, however, little evidence exists for whether such a programme effectively affect the participation and inequalities at the population level. Methods We used individual-level data from a nationally cross-sectional survey in 2014–2015 for breast and cervical cancer screening participation. Data for intervention and comparison grouping were from 2009 to 2014 national breast and cervical cancer screening programme, and counties covered by the programme were divided into intervention group. We assessed the impact of the intervention on the participation rates and the magnitude of inequalities with two approaches: multivariable multilevel logistic regressions adjusting for individual- and region-level covariates; and a difference analysis combined with propensity score matching that estimated the average intervention effect. Results Of 69,875 included women aged 35–64 years, 21,620 were classified into the intervention group and 43,669 into the comparison group for breast cancer screening; and 31,794 into the intervention group, and 33,682 into the comparison group for cervical cancer screening. Participation rate was higher in intervention group than comparison group for breast cancer screening (25.3, 95% confidential interval [CI], 22.8–27.7%, vs 19.1, 17.4–20.7%), and cervical cancer screening (25.7, 23.8–27.7%, vs 21.5, 19.6–23.5%), respectively. Intervention significantly increased the likelihood of participation for both breast cancer and cervical cancer screening in overall women, rural women and urban women, whereas the effect was significantly higher in rural women than urban women. The average intervention effect on the participation rate was an increase of 7.5% (6.7–8.2%) for breast cancer screening, and 6.8% (6.1–7.5%) for cervical cancer screening, respectively. The inequalities were significantly decreased by 37–41% (P < 0.001) between rural and urban, however, were slightly decreased or even increased in terms of age, education status, and household income. Conclusions Organized breast and cervical cancer screening programme targeting for a fraction of women could increase the participation rates at population level, however, it could not significantly affect socioeconomic-introduced inequalities. Further studies are need to conduct time-series analyses and strengthen the causal correlation.