JMIR Publications

ABSTRACT

Background:

Annual low-dose computed tomography (LDCT) screening reduces lung cancer mortality among high-risk adults, yet uptake has remained persistently low. In 2021, the US Preventive Services Task Force (USPSTF) expanded eligibility by lowering the starting age from 55 to 50 years and the smoking-history threshold from 30 to 20 pack-years, and in 2022 the Centers for Medicare & Medicaid Services (CMS) issued a matching National Coverage Determination (NCD). Whether these policies changed public awareness and real-world screening behavior, and through what mechanism, has not been established using real-time data.

Objective:

We aimed to determine whether the 2021 USPSTF recommendation and 2022 CMS NCD increased public information-seeking for LDCT screening and actual LDCT utilization, and whether utilization changes differed between the newly eligible (aged 50-54) and previously eligible (aged 55-74) populations.

Methods:

We used a three-arm quasi-experimental design spanning January 2016 to June 2023. Arm A applied a two-way fixed-effects difference-in-differences model to a state-level panel of Google Trends search volume, using mammography search as a concurrent control. Arm B applied interrupted time series (ITS) analysis to the within-disease ratio of screening-specific to general lung cancer search volume. Arm C applied ITS analysis to monthly LDCT encounter counts from the Epic Cosmos electronic health record network (>200 million patients), with pre-specified age- and payer-stratified analyses. Event-study models assessed pre-policy trends.

Results:

National LDCT encounter volume rose 53-fold over the study period. The utilization response was concentrated in newly eligible adults aged 50-54, who showed a 517% immediate increase above trend following the USPSTF Final Recommendation (β = +1.82, 95% CI 1.40 to 2.25, P < .001) and a 46-fold increase over baseline by June 2023; no previously eligible age group showed a significant discrete level shift. Responses were near-identical across commercial, Medicare, and Medicaid payers, arguing against insurance access as the binding mechanism. Search interest and encounter volume were highly correlated (r = 0.86, P < .001). The Arm A parallel-trends assumption was formally rejected (χ²₁₁ = 54.0, P < .001); causal interpretation therefore rests on convergence across the three arms rather than on the difference-in-differences estimate alone.

Conclusions:

The 2021-2022 policy expansion successfully activated the newly eligible population but did not measurably increase uptake among the larger previously eligible majority who remain under-screened. Because the response did not vary by payer, closing the residual gap will require implementation infrastructure rather than further eligibility or coverage changes. Paired infodemiology and EHR data provide a validated, near-real-time framework for monitoring screening-policy impact.