ASPO Abstracts
More than one: Evaluating intra-system variation in the implementation and impact of lung cancer screening
Category: Early Detection & Risk Prediction
Conference Year: 2020
Abstract Body:
Purpose. To compare multilevel factors associated with lung cancer screening (LCS) in two medical centers
affiliated with a single healthcare organization. Methods. Using electronic medical record data and contextual
inquiry from stakeholders, we evaluated differences by medical center in baseline and subsequent annual LCS
using low-dose computed tomography in patients undergoing baseline screening in 2014-2018. The two centers
are located about 90 miles apart in urban and semi-rural locations in Southeastern Pennsylvania. Bivariate and
multivariate analyses were conducted to compare differences by center in patient characteristics, (age, sex,
race/ethnicity, and smoking status), clinician specialty, baseline LungRADS, and subsequent annual screening.
Key informant interviews with clinical leadership were used to describe LCS implementation at each center.
Results. To implement LCS, the semi-rural center used a centralized approach, coordinated through a nurse-led
program for shared decision-making and coordination; whereas, the urban center used a de-centralized approach,
without direct outreach or centralized coordination. In comparison to patients screened at the urban center
(n=2,611), patients screened at the semi-rural center (n=3,186) were significantly more likely to be Non-Hispanic
White (90.5% vs 64.5%), younger than age 65 (52.7% vs 48.6%), and male (54.7% vs 50.9%), and less likely to
be current smokers (48.3% vs 54.4%). Nearly-half (47.3%) of semi-rural center patients received a timely annual
screen (<15 months) following baseline in contrast to 17.5% of urban center patients. Semi-rural center patients
were 3-times more likely to receive any subsequent LCS in comparison to urban center patients, controlling for
patient characteristics (AOR=3.34; 95%CI: 2.62-4.00). Conclusions: Intra-system variation in LCS implementation
was associated with differences in optimal adherence, not attributable to observed patient-level factors. As U.S.
healthcare systems continue to consolidate, it is important to understand how contextual factors between and
within systems impact outcomes. Further research is needed to identify effective strategies to improving screening
to ensure that existing disparities are not exacerbated by LCS implementation differences.
Keywords: Lung cancer screening, implementation, healthcare systems