ADVANCING THE EVALUATION OF SCREENING PROGRAMS: A REGISTRY-BASED COHORT STUDY OF LONGITUDINAL SCREENING ADHERENCE AND CERVICAL CANCER RISK

Purpose: The effectiveness and efficiency of cancer screening may be improved by increasing adherence to screening guidelines, yet there is often uncertainty around which individuals to target due to limited information about screening behavior. While repeated screening at routine intervals (e.g., every three years) is recommended, current definitions of adherence only capture a single screen within a defined period of time (e.g., 3.5 years). We aimed to advance current screening evaluation by providing a longitudinal adherence metric that is more aligned with guidelines-based recommendations for cervical cancer screening.

Methods: We conducted a registry-based cohort study using data from the organized triennial cytology-based screening program in Norway to develop a longitudinal screening adherence metric, defined as adherence to routine cervical cancer screening guidelines over multiple screening intervals (i.e., the number of observed versus expected screening intervals; interval length=3.5 years). We included all women in Norway born between 1936 and 1983 eligible for at least two screening intervals (N=1,293,379). We categorized women into one of five longitudinal adherence categories: never-screeners, severe under-screeners, moderate under-screeners, guidelines-based screeners, and over-screeners. For each longitudinal adherence category we estimated the cumulative risk of developing cervical cancer by age 75 years and cancer stage at diagnosis.

Results: Using the longitudinal adherence metric, only 20% of women consistently screened every three years according to guidelines (i.e., guidelines-based screeners), while the majority of women (45%) were classified as moderate or severe under-screeners, and the remaining women were classified as either over-screeners (29%) or never-screeners (6%). The cumulative risk of developing cervical cancer and stage of diagnosis varied by longitudinal adherence category; for example, never-screeners had nearly twice the cumulative risk of developing cervical cancer (i.e., 1.19%) compared with severe under-screeners (i.e., 0.62%), and only 21% of these cancers were diagnosed at Stage 1. In contrast, women in the remaining categories were at the lowest risk (ranging from 0.27% to 0.37%), and were more likely to be diagnosed at Stage 1 (ranging from 54% to 70%). 

Conclusions: The longitudinal adherence metric more accurately reflects guidelines-based recommendations by capturing screening behavior over repeated intervals and differentiates women’s long-term cancer outcomes. Understanding screening behavior over time may help inform studies that evaluate interventions to improve screening adherence, and aid the continued refinement of screening programs.