Predictors of Premature or Unnecessary Cervical Cancer Screening—Oregon, 2010–2014

BACKGROUND: Prevailing cervical cancer screening guidelines changed in 2009 and again in 2012. The extent to which providers adhere to revised guidance is not known. Excessive screening wastes resources and can lead to unnecessary emotional distress. We sought to identify risk factors for unnecessary or premature screening.

METHODS: We analyzed statewide administrative claims for service provided 2010 through May 2014 to female Oregon residents aged ≥18 years. Claims contain a unique identifier allowing distinction of unique individuals. We defined a woman as having had a screening test if a claim in her name listed ≥1 ICD-9 diagnoses or procedure codes determined by the National Committee for Quality Assurance to be indicative of routine cervical cancer screening. We calculated proportions of women screened by age group, insurance type and location of residence. We created two logistic regression models to identify predictors of screening in women aged <21 years and women aged >65 years, since guidelines advise against screening of average-risk women in these groups. We used multivariate time-to-event analysis to identify predictors of premature repeat screening among women aged 21–29 years first screened between June 2011 and April 2014. All models adjusted for evidence cervical dysplasia and related conditions to examine predictors of screening among average-risk women.

RESULTS: The proportion of women screened decreased annually in all age groups, most substantially in in women aged 18–21 years from 9.8% in 2010 to 0.8% in 2014. Both women <21 years and women >65 years were significantly less likely to be screened during each succeeding year. Compared to 2010, adjusted odds of screening in 2014 were 0.28 (95%CI:0.24–0.32) for women <21years and 0.57 (95%CI:0.53–0.61) for women >65 years. For women <21 years but not for women >65 years, rural relative to urban residence (OR:2.47; 95%CI:1.78–3.43) and public relative to commercial insurance (OR:2.35; 95%CI:2.24–2.48) were associated with screening. For women 21-29 years old, average risk of a second screening increased with age (61%/year), use of public  insurance (21% higher than commercial), and rurality (RR: 1.36; 95% CI:1.05-1.76)

CONCLUSIONS: Cervical cancer screening has declined in Oregon since guidelines were changed. Though screening was rare in this age group by 2014, women <21 years were more likely to have been screened—perhaps unnecessarily—if they were publicly insured or rural residents. Rural and publicly insured patients and their health care providers might benefit from additional education about current screening recommendations.