How Longitudinal EHR Smoking Data Could Boost Lung Cancer Screening

The use of longitudinal EHR smoking data could improve lung cancer screening, according to a study published in JAMIA.

The US Preventive Services Task Force (USPSTF) requires the estimation of lifetime pack-years to determine lung cancer screening eligibility. Market-leading EHR vendors calculate pack-years using only the most recently recorded smoking data.

Using this approach, researchers analyzed patient records for 16 ,874 current or former smokers seen at an academic health system in 2020. All patients met the UPSTF age criteria for screening (50 to 80 years old) and had no prior lung cancer diagnosis.

Only about 60 percent of the patient population had the requisite smoking data recorded in the EHR to determine USPSTF screening eligibility.

More than 80 percent of patient records had at least one issue, including missing, outdated, and inaccurate smoking data.

The study authors developed a longitudinal approach to determine lung cancer screening eligibility to address these data quality issues.

"This Longitudinal Approach can help identify patients eligible for lung cancer screening who would be missed by the Baseline Approach, which uses only the most recent EHR data and is the predominant algorithm used by market-leading EHR systems," the researchers explained.

The study found that leveraging the most recent EHR data with longitudinal data identified 49 percent more patients potentially eligible for lung cancer screening than the most recent records alone.

Notably, this combined approach identified 40 percent more high-risk, high-benefit patients.

"Screening is particularly important for this high-benefit population, and misclassifying such patients as ineligible is particularly disconcerting," the study authors wrote. "If implemented in clinical practice, the Combined Approach could substantially increase the number of individuals who are evaluated for lung cancer screening."

The approach described in this research could alert clinicians about patients with discrepancies between their most recent EHR smoking data and their longitudinal data to prompt clinical documentation improvement and inform screening eligibility.

"For example, when the Baseline Approach estimated higher pack-years than the Longitudinal Approach in our dataset, this often appeared to be a result of cigarettes-per-day being mistakenly entered as pack-per-day by EHR users, which were taken at face value in the Baseline Approach but assumed to be cigarettes-per-day in the Longitudinal Approach," the authors explained.

The researchers noted that using longitudinal EHR data could significantly increase the number of eligible patients identified for lung cancer screening, which is a critical need given that providers currently screen less than five percent of eligible patients.

Additionally, the study authors said that leveraging longitudinal data for lung cancer screening eligibility could help providers target individuals with high lifetime smoking exposure for smoking cessation outreach.

The researchers also suggested that historical smoking data should be accessible to clinical decision support systems in addition to the most recent smoking data.

Healthcare organizations could make longitudinal smoking data available from EHRs through Health Level 7 (HL7) Fast Healthcare Interoperability Resources (FHIR) application programming interfaces (APIs). This would allow external apps connected to the EHR to leverage the data for screening identification, the researchers said.

The study authors recognized several study limitations.

"The data analysis was conducted using data from one healthcare system; as such, replication is needed," they said. "However, we have no particular reason to believe that our findings are significantly divergent from what would be found in other clinical settings."

Additionally, the researchers conducted the study in Utah, where the population is 78 percent White. However, the authors noted that they conducted a fairness analysis which showed that the combined approach of using longitudinal EHR data and the most recent EHR data would boost potential lung cancer screening identification in White, Black, Hispanic, and other populations.