Electronic health records significantly underestimate smoking history in women, causing approximately 37% of those eligible for lung cancer screening to be missed, while patient-reported surveys and general chart reviews perform much more accurately in identifying screening candidates.
- EHR social history showed only 8.4% sensitivity compared to 99.3% sensitivity from general chart history in identifying lung cancer screening-eligible women
- Approximately 37% of eligible women did not undergo lung cancer screening despite meeting criteria, primarily due to incomplete EHR documentation
- Patient-reported smoking surveys outperformed EHR data and showed accuracy comparable to general charting, achieving 96% accuracy versus 79.2% for social history alone
- EHR social history limitations include allowing only a single start and quit date, which cannot capture variable smoking patterns across a lifetime
- Researchers recommend automated extraction of all smoking periods from general charts into social history and patient surveys to improve lung cancer screening identification
Electronic health records (EHRs) may underestimate smoking history in women undergoing mammography who may be eligible for lung cancer screening.
Patient-reported smoking surveys outperformed EHR social history and showed accuracy on par with general charting, which could help identify women at mammography who are eligible for lung cancer screening, wrote a team led by Jeffrey Velotta MD, from Kaiser Permanente in Pasadena, CA.
“We propose automated extraction of all smoking periods from the general chart into the social history to calculate lifetime pack-years, with a patient-reported survey updating current status,” Velotta and colleagues wrote in the Annals of Thoracic Surgery Short Reports.
Breast cancer survivors have a higher risk of subsequent lung cancer. And while routine breast cancer screening uptake is high among these women, lung cancer screening uptake is suboptimal, the researchers wrote.
Accurate documentation of smoking history is needed to determine which patients are eligible for lung cancer screening. The researchers noted that smoking history is often incomplete in the EHR. This is due to inconsistent documentation and recall limitations. This means some eligible patients are still unscreened.
The Velotta team studied smoking history accuracy across three data sources to assess lung cancer screening eligibility among women undergoing screening mammography. It collected survey data in 2025 for self-reported smoking history from women undergoing screening mammography at a tertiary breast center.
The team gathered smoking history from patient surveys, the EHR, and EHR general chart history. It determined lung cancer screening eligibility using 2021 criteria from the U.S. Preventive Services Task Force (USPSTF).
The study included 1,299 women, of whom 19 were eligible for lung cancer screening through survey data. Of the eligible women, 12 had prior lung cancer screening documentation in the EHR. This means that about 37% of eligible women did not undergo lung cancer screening.
Using the survey as reference, general chart history showed higher sensitivity and accuracy than social history. General chart history showed more alignment with survey data than social history.
Performance of social history, general chart history for identifying lung cancer screening-eligible women (using survey as reference) | ||
Measure | Social history | General chart history |
Sensitivity | 8.4% | 99.3% |
Specificity | 99.2% | 95.1% |
Positive predictive value | 73.9 | 84.8 |
Negative predictive value | 79.3 | 99.8 |
Accuracy | 79.2% | 96.0% |
F1 score | 15.0 | 91.5 |
The results suggest that EHR eligibility alone does not guarantee screening. Women who are eligible for lung cancer screening may be missed without clinical prompting, the study authors wrote.
The team also noted that social history “only allows a single start and quit date” with an averaged pack-year exposure. This limits accurate assessment for the variable smoking patterns common across a lifetime, it added.
“Standardized workflows that longitudinally capture all smoking episodes may improve [lung cancer screening] determination,” the authors wrote.
Read the full study here.
