Estimating Pack-Year Eligibility for Lung Cancer Screening Using 2 Yes or No Questions
Menée aux Etats-Unis par enquête auprès de 303 fumeurs (âge : 55-80 ans ; 58 % de femmes), cette étude évalue la pertinence d'un questionnaire composé de 2 questions fermées pour estimer le nombre de paquets consommés par année et identifier les personnes susceptibles de bénéficier d'un dépistage du cancer du poumon
Introduction: Lung cancer screening (LCS) for adults at high risk reduces lung cancer mortality by 16% to 20%.1 However, uptake of LCS remains low and unequal in the US.2,3 LCS differs from other types of cancer screening because eligibility assessment requires evaluation of lifetime smoking intensity (ie, pack-years). Although tobacco use (ie, current, former, or never) is well captured, calculated pack-years are often missing or inaccurate in electronic medical records (EMRs).4 Prior approaches to assessing pack-years have used open-ended questions, a format associated with increased respondent burden and lower response rates.5 There is a need for tools that simplify pack-year assessment while maintaining accuracy.4 In this cross-sectional study, we developed and tested the accuracy of using 2 yes or no questions to estimate pack-year eligibility for LCS.
Methods: This study, approved by the University of Pennsylvania institutional review board (IRB), was embedded within a parent pilot trial aimed at increasing LCS at a multisite academic health care system (NCT04612946). The parent trial used EMR data to identify primary care patients aged 55 to 80 years with no history of LCS or lung cancer and listed as currently smoking from 2018 to 2020. Prior to trial enrollment, we invited identified patients to complete an LCS eligibility survey electronically or via telephone using a waiver of written consent approved by the University of Pennsylvania IRB because the research placed no more than minimal risk to the participants. The survey asked open-ended questions regarding total number of years and mean number of packs per day smoked to calculate pack-years. The survey also included the following 2 yes or no questions: (1) If you add up all the years when you regularly smoked cigarettes, have you smoked for 20 years or more of your life? (2) At any time in your life, did you regularly smoke 1 or more packs of cigarettes per day? We evaluated the sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of the questions to estimate eligibility (ie, 20 or more pack-years) using 2 classifications of a positive response: (1) liberal (yes to at least 1 question) and (2) restrictive (yes to both questions). We calculated 95% CIs using the Clopper-Pearson method, with a 2-sided P < .05 considered statistically significant. Using the STROBE reporting guideline, we report results overall and stratified by sex, age, and race and ethnicity as documented in the patient’s EMR (due to known racial and ethnic disparities in lung cancer). All analyses were performed using Stata, version 17 (StataCorp LP).
Results: Of 303 patients who completed the survey, 156 (51.5%) were aged 55 to 64 years, 176 (58.1%) were female, and 62 (20.5%) were Black or African American; 210 (69.3%) reported 20 or more pack-years (Table 1). Using the liberal classification (Table 2), the yes or no questions had 100% NPV and 100% sensitivity for estimating pack-year eligibility for LCS but lower specificity overall (18.3%) and within subgroups (range, 13.0%-25.6%). Overall specificity (79.6%) and PPV (90.5% [range within subgroups, 88.3%-92.8%]) increased using the restrictive classification, but sensitivity decreased (85.7%), particularly among adults aged 55 to 64 years (81.4%) or Black adults (75.6%).
Discussion: Response to 2 yes or no questions showed high sensitivity for estimating pack-year eligibility for LCS, helping to identify pragmatic solutions for overcoming the challenge of assessing pack-years in routine practice.6 We observed differences in sensitivity and specificity between liberal and restrictive classifications of positive response. However, these differences might also increase the range of utility of these questions in practice. The restrictive classification could be used to equitably identify patients for direct referral to shared decision-making for LCS given the high PPV overall (90.5%) and within each subgroup (88.3%-92.8%), whereas the liberal classification could be used to efficiently exclude patients from LCS outreach given the consistently high NPV (100.0%). Our study is limited by its sample size, single site, and exclusion of years since quitting among adults who formerly smoked. Future work, including testing in a larger national sample of participants aged 50 to 80 years, could expand on these findings.
JAMA Network Open , article en libre accès, 2022