A few years ago, Kim Lori Sandler, MD, realized many patients newly diagnosed with lung cancer had never been screened for the disease — they received CT scans only because they were symptomatic.
But Dr. Sandler, a radiologist at Vanderbilt University Medical Center in Nashville, Tennessee, could see in medical charts that most of these patients had been eligible for a screening before becoming symptomatic. And for women, most had received decades worth of mammograms. She saw an opportunity and launched a study to find out if an intervention would work.
Low-dose CT and mammography services often are available in the same imaging facility, so women who qualified for a lung cancer screening were offered the scan during their mammography visit. Over a 3-year period, monthly rates of lung scans in women rose by 50% at one facility and 36% at the other.
“What we found is that women are really receptive, if you talk to them about it,” Dr. Sandler said. “I don’t think that lung cancer is thought of as a disease in women.”
Although lung cancer is the leading cause of cancer deaths in the United States, a recent study in JAMA Internal Medicine found only 18% of eligible patients were screened in 2022, a far cry from the rates of 72% for colon cancer — which itself falls short of goals from US medical groups like the American Cancer Society (ACS). Among those eligible, rates of lung screenings were lowest among younger people without comorbid conditions, who did not have health insurance or a usual source of care, and those living in southern states and states that did not expand Medicaid as part of the Affordable Care Act.
Getting patients screened is lifesaving: 27% of people with lung cancer survive 5 years after diagnosis. But the survival rate rises to 63% when cases are diagnosed at an early stage.
Increasing Uptake
The formal recommendation to use low-dose chest CT to screen for lung cancer is only a decade old. The approach was first endorsed by the United States Preventive Services Task Force (USPSTF) on the basis of an influential trial that found such testing was linked to a 20% reduction in mortality from the disease. Updated 2021 USPSTF guidelines call for annual screening of people aged 50-80 years who have a 20 pack-year history of smoking and currently smoke or have quit within the past 15 years.
But implementing the recommendation is not always simple. Unlike a colorectal or breast cancer screening, which is recommended primarily on patient age, eligibility for a lung cancer screening requires calculating pack-years of smoking, and, for past smokers, knowledge of when they quit.
The structured fields in most electronic medical records (EMRs) inquire about current or past use of cigarettes and the number of daily packs smoked. But few EMRs can calculate when a patient starts smoking two cigarettes a day but then increases to a pack a day and cuts down again. EMRs also do not track when a patient has stopped smoking permanently. Individual clinicians or health systems must identify patients who are eligible for screening, but the lack of automated calculations makes that job more difficult.
Dr. Sandler and colleagues turned to the informatics team at Vanderbilt to develop a natural language processing approach that extracts smoking data directly from clinician notes instead of using standard variables in their EMR.
The number of patients identified as needing a screening using the algorithm nearly doubled from baseline, from 5887 to 10,231 over a 3-year period, according to results from another study that Dr. Sandler published.
Although the algorithm may occasionally flag someone who does not need screening as eligible, “you can always have a conversation with the patient to determine if they actually meet eligibility criteria,” Dr. Sandler said.