A population-based study of primary care predictors of non-attendance for cervical screening

Publisher Website
Google Scholar
Abstract
Objective: To identify and estimate the population impact of primary care service delivery factors that independently predict non-attendance for cervical screening. Setting: Screening records of all eligible women aged 30 years and over and resident in the former Manchester Health Authority area were analysed (n=72,613). Methods: Practice-level and GP-level explanatory variables, along with area-level covariates, were obtained and merged to the study data set. Logistic regression models were used to identify factors associated with having no recorded history of attending NHS cervical screening services. A multivariate model was created to identify independent predictors of non-attendance with comprehensive adjustment for women's age, area-level socio-demographic factors, and other primary care factors. Attributable fraction estimates were used to assess the population impact of the independent predictors. Results: Large practice size (>4,000 patients), single-handed practice, South Asian male GP, part-time GP employment status, older age and birthplace overseas, and area-level measures of deprivation and transience independently predicted non-attendance. Women born overseas and registered at larger practices were especially unlikely to have ever attended. The combined population attributable fraction estimate for the independent predictors reflecting primary care service delivery was almost 40%, and that for all variables in the final model was over 70%. Conclusions: Independent predictors of non-attendance reflecting general practice structure, workload and GP characteristics were identified. Although relative risks were modest, the collective population impact of these factors was considerable, which has implications for the implementation of informed targeting and the development of new screening services by Primary Care Trusts.

This publication has 0 references indexed in Scilit: