(B-149) Constructing a frailty index (FI) using the Behavioral Risk Factor Surveillance System: Application in lung cancer screening

Background: Frailty measures are widely used in research and clinical practice to assess increased vulnerability to adverse health outcomes. Using the deficit accumulation model, a frailty index (FI) can be constructed retrospectively by synthesizing a variety of health-related variables. The Behavioral Risk Factor Surveillance System (BRFSS), the largest ongoing phone-based survey in the US, provides extensive health-related data, enabling FI development.

Objectives: To construct an FI among individuals aged 50-79 using the 2022 BRFSS data and assess frailty status in populations eligible and screened for lung cancer.

Methods: Following a standard procedure for FI development, all variables from the 2022 BRFSS dataset were screened for inclusion. Each selected variable was recoded on a scale from 0 (no deficit) to 1 (deficit), with higher scores indicating increasing frailty. The continuous FI (mean of all selected variables) was categorized as robust ( < 0.15), prefrail (0.15-0.24), mildly frail (0.25-0.34), and moderate-to-severely frail (≥0.35). We applied the BRFSS design weights to reflect US population characteristics and assessed FI score distribution in the US population aged 50-79. We then applied the FI to individuals eligible for lung cancer screening (LCS) and those who underwent LCS. LCS eligibility was determined based on the 2021 US Preventive Services Task Force criteria: age 50-80 years, currently smoking or having quit within the last 15 years, and having a smoking history of ≥20 pack-years. LCS in the last year was ascertained by self-report.

Results: We identified 55 health-related variables from the BRFSS data, of which 30 were selected to construct the FI. We identified 245,342 respondents aged 50-79 (weighted n=113,059,723), of whom 235,326 (weighted n=107,066,840) had a valid FI score. In the weighted overall population, the mean FI score was 0.156 (standard error [SE]=0.0005), with 59.3% classified as robust, 22.5% as prefrail, 11.1% as mildly frail, and 8.2% as moderate-to-severely frail. In the applied example, the mean FI score was 0.218 (SE=0.0017) in the screening-eligible population and 0.244 (SE=0.0033) in the screened population. Mean FI score increased with age in the screening-eligible population (50-54y: 0.198, 55-59y: 0.215, 60-64y: 0.217, 65-69y: 0.217, 70-74y: 0.232, 75-79y: 0.239). The percent screening uptake among those eligible increased with increasing frailty: 12.1% in robust, 17.8% in prefrail, 19.8% in mildly frail, and 19.7% in moderate-to-severely frail.

Conclusions: We developed an FI for use in the BRFSS data for those aged 50-79 and applied this index to two LCS populations. This measure may be used to address a range of aging-related research questions using the BRFSS data.