Comments
Abstract
Rationale: Lung cancer screening with low-dose chest computed tomography decreases mortality for high-risk current or former smokers. Lifetime smoking intensity (cigarette pack-years), an essential eligibility criterion, is poorly recorded in electronic health records, which may contribute to the overall low appropriate use of screening.
Objectives: We sought to assess whether elements commonly extractable from electronic health records may be useful as prescreening tools to identify individuals for formal assessment of eligibility.
Methods: This was a cross-sectional cohort study of the National Health and Nutrition Examination Survey (NHANES) continuous survey, years 2011–2016. We included all adult participants with complete smoking interview data, weighted to construct a nationally representative cohort. We determined test characteristics for five criteria, including eligibility age, smoking status (current, former, or never), and current smoking intensity, to predict lung cancer screening eligibility as defined by the U.S. Preventive Services Task Force and Centers for Medicare and Medicaid Services.
Results: Almost 9 million individuals (3.8% of the population) may qualify for screening. Simplified criteria, including the appropriate age range (55–77 yr) and smoking status, correctly discriminated individuals who were eligible for screening in most cases (area under the curve = 0.92). When the analysis was restricted to those of eligible age, smoking status retained fair predictive value (area under the curve = 0.85). Incorporating additional information about current smoking behavior would allow for refinement of approaches to identify specific populations for screening.
Conclusions: These simplified criteria may be useful for identifying individuals who are eligible for lung cancer screening. Applying these criteria as a prescreening tool may improve appropriate referral and implementation of screening.
Keywords: lung cancer; early detection of lung cancer; cancer prevention; tobacco abuse
The National Lung Screening Trial found a 20% lung cancer-specific mortality benefit for annual lung cancer screening with low-dose chest computed tomography in a subset of high-risk current and former smokers (1). The U.S. Preventive Service Task Force and the Centers for Medicare and Medicaid Services (CMS) now recommend lung cancer screening, overlapping in the following eligibility criteria: age 55–77 years, current smokers or former smokers who quit within 15 years, and a total lifetime smoking intensity of at least 30 pack-years (2, 3). The mortality benefits of lung cancer screening were recently confirmed for a similar high-risk population in the Dutch-Belgian Randomized Lung Cancer Screening Trial (NELSON). (4).
Despite evidence, recommendations, and insurance coverage, implementation of screening remains poor: multiple studies suggest that <5% of eligible Americans are enrolled in screening programs (5–7). There is also a concerning overuse of screening among noneligible individuals (7, 8). These findings suggest that a major barrier to implementation is the difficulty of identifying eligible patients on a systematic basis. One essential eligibility criterion, lifetime smoking intensity (pack-years), is inaccurately and incompletely recorded in most electronic health records (EHRs), with a recent study demonstrating a 96% discordance between EHRs and patient reports (9). Unfortunately, even resource-intensive interventions, such as training and incentives to improve documentation of pack-years, may have modest benefits (10).
As health systems consider approaches to improve health maintenance, there is a need to use commonly available information to improve the efficiency and effectiveness of screening services. In addition, it is important to identify eligible individuals across a health system to understand gaps in care and develop quality-improvement initiatives. The primary objective of this study was to understand the predictive value of criteria based on commonly extractable elements from the EHR to establish prescreening tools to identify individuals who are eligible for lung cancer screening.
We performed a cross-sectional analysis using publicly released data from the National Health and Nutrition Examination Survey (NHANES). Details regarding the methods used by NHANES have been documented elsewhere (11). Briefly, NHANES samples approximately 5,000 Americans each year and applies sampling weights to represent the noninstitutionalized civilian population of the United States. We used data from NHANES in-home interviews, including demographics (sex, age, and race/ethnicity), socioeconomic data (household income, highest education level, and insurance status), health information (self-reported comorbid conditions), and tobacco use. Tobacco exposure information included smoking status, delineated as current, former, or never (ever-smoker defined as >100 cigarettes in lifetime, and current smoker defined as “now” smoking cigarettes), initiation, and quit age. We estimated smoking duration based on initiation and quit age. We assessed intensity via questions that assessed the number of cigarettes usually smoked per day for former smokers, and the average over the past month for current smokers. Pack-years were estimated using a combination of smoking duration and intensity.
We included subjects from three cycles of NHANES, 2011–2016 (n = 29,902). We restricted the cohort to adults over 18 years of age (n = 17,969) and those with complete responses in the smoking interview (n = 17,830). We then applied appropriate subject weights per NHANES (n = 235,517,739) to create a nationally representative cohort. We defined people eligible for lung cancer screening as persons who met both CMS and U.S. Preventive Service Task Force eligibility criteria: 55–77 years of age, 30 pack-year smoking history, and current smokers or those who quit within 15 years of the interview. We summarized the total cohort as well as the eligible subjects. Based on the hypothesis that eligibility would vary significantly by socioeconomic status, we also compared eligibility by sociodemographic features, including sex, racial/ethnic group, education and income levels, and insurance status, using chi-squared tests for heterogeneity and Cochran-Armitage trend tests when appropriate, taking into account the complex survey design. We determined the test characteristics of five prescreening criteria, using combinations of age, smoking status, current smoking intensity, and quit date (for former smokers) (Table 1). We report sensitivity, specificity, positive and negative predictive values, positive and negative likelihood ratios (LRs), and accuracy. Accuracy, or diagnostic effectiveness, is the percentage of individuals correctly classified by a test (true positives and true negatives divided by all tests). We calculated estimates of the 95% confidence intervals (CI) for each of these values; however, because of the large number of sampled persons (n = 17,830), these estimates were narrow and thus are not presented. We built a receiver operator curve (ROC) for prediction of eligibility using age and smoking status (never, former, or current) based on logistic regression and calculated the area under the curve (AUC) with 95% CIs. We performed sensitivity analyses to assess test performance across different demographic groups. A two-sided P value of 0.05 was considered significant for comparative testing. All statistical analyses were performed using Stata version 15.0 (StataCorp). This work was considered exempt from institutional review board review at the University of Washington because NHANES is an exempted publicly available and deidentified dataset.
| Criteria 1: Age 55–77 yr and ever cigarette smoker |
| Criteria 2: Age 55–77 yr and current cigarette smoker or former smoker who quit less than 15 yr ago |
| Criteria 3: Age 55–77 yr and current cigarette smoker |
| Criteria 4: Age 55–77 yr and current cigarette smoker of more than half a pack per day |
| Criteria 5: Age 55–77 yr and current cigarette smoker of more than one pack per day |
In a total cohort of 235,517,739, an estimated 8,983,406 subjects (3.8%) were eligible for lung cancer screening (Table 2). There were key differences in eligibility by demographic groups. Men were more than twice as likely as women to be eligible (Table 3). Non-Hispanic whites were more likely to be eligible than individuals of non-Hispanic black or Asian race, or Hispanic ethnicity. Subjects in lower education and income quartiles were more likely to be eligible.
| Characteristic | Total cohort, N = 235,517,739 | Screening eligible, N = 8,983,406 |
|---|---|---|
| Sex | ||
| Men, % | 48 | 67 |
| Women, % | 52 | 33 |
| Age | ||
| 18–30 yr, % | 23 | 0 |
| 30–40 yr, % | 17 | 0 |
| 41–50 yr, % | 18 | 0 |
| 51–60 yr, % | 18 | 41 |
| 61–70 yr, % | 13 | 45 |
| 71–79 yr, % | 6.5 | 14 |
| 80+ yr, % | 4.4 | 0 |
| Race/ethnicity | ||
| Hispanic, % | 15 | 4.2 |
| White, non-Hispanic, % | 65 | 84 |
| Black, non-Hispanic, % | 12 | 5.8 |
| Asian, non-Hispanic, % | 5.5 | 1.6 |
| Other, non-Hispanic, % | 3.1 | 4.5 |
| Highest educational level | ||
| Some high school or less, % | 15 | 20 |
| High school or GED, % | 21 | 27 |
| Some college or associates degree, % | 32 | 35 |
| College degree or higher, % | 31 | 17 |
| Refused or don’t know, % | 0.06 | 0 |
| Annual household income | ||
| 0–$9,999, % | 4.3 | 5.1 |
| $10,000–$25,000, % | 15 | 21 |
| $25,000–$54,999, % | 26 | 34 |
| $55,000–$99,999, % | 22 | 19 |
| >$100,000, % | 25 | 16 |
| Preferred to answer <20,000, % | 0.9 | 0.9 |
| Preferred to answer >20,000, % | 3.4 | 2.3 |
| Refused or don’t know, % | 2.7 | 1.3 |
| Smoking behavior | ||
| Former smoker, % | 23 | 44 |
| Current smoker, % | 19 | 56 |
| Age at smoking initiation, median (IQR) | 17 (15–19) | 16 (14–18) |
| Years smoking, median (IQR) | 21 (9–34) | 42 (38–47) |
| Smoking pack-years, median (IQR) | 9 (2.4–24) | 45 (39–60) |
| Comorbid medical conditions | ||
| History of COPD, % | 3.2 | 15 |
| History of chronic bronchitis, % | 5.7 | 13 |
| History of emphysema, % | 1.8 | 11 |
| History of any cancer, % | 11 | 21 |
| History of lung cancer, % | 0.3 | 1.5 |
| History of heart failure, % | 2.6 | 8.7 |
| History of stroke, % | 2.8 | 6.5 |
| History of coronary artery disease, % | 3.3 | 13 |
| Characteristic | Percentage of Subjects Who Are Eligible by Characteristic (N = 235,517,739) | Percentage of Subjects Who Are Eligible among Subjects 55–77 Yr Old (N = 69,213,318) |
|---|---|---|
| All subjects | 3.8 | 13 |
| Sex*† | ||
| Men | 5.3 | 18 |
| Women | 2.4 | 8.1 |
| Race/ethnicity*† | ||
| Hispanic | 1.1 | 6.1 |
| White, non-Hispanic | 4.9 | 15 |
| Black, non-Hispanic | 1.9 | 7.7 |
| Asian, non-Hispanic | 1.1 | 4.5 |
| Other, non-Hispanic | 5.6 | 22 |
| Highest educationठ| ||
| Some high school or less | 5.2 | 18 |
| High school or General Educational Development | 5.2 | 16 |
| Some college or associates degree | 4.2 | 14 |
| College degree or higher | 2.2 | 7.1 |
| Income quartilesठ| ||
| <34,999 | 4.8 | 16 |
| 35,000–64,999 | 5.2 | 17 |
| 65,000–99,999 | 2.7 | 9.1 |
| 100,000+ | 2.5 | 8.5 |
| Health insurance coverage* | ||
| Covered by health insurance | 4.2 | 13 |
| Not covered by health insurance | 2.2 | 15 |
| Private health insurance*† | ||
| Covered by private insurance | 3.3 | 11 |
| Not covered by private insurance | 4.5 | 16 |
We examined the performance of five prescreening criteria, using an eligible age range and smoking data, for predicting full eligibility criteria. We found that even the simplest criteria—age of 55–77 years and ever-smoker—had favorable test characteristics, including a sensitivity of 100%, specificity of 88%, and accuracy of 89% (Table 4). If this prescreen was negative, the subjects were not eligible (negative LR, 0), and if positive, this test moderately increased the likelihood of eligibility (positive LR, 8.5). Including only current smokers decreased the sensitivity (57%) but increased the specificity (97%) and overall accuracy (96%). Including more information about current smokers’ smoking intensity further increased specificity and accuracy. The AUC of smoking status and age for predicting full eligibility was 0.92 (95% CI, 0.91–0.92) (Figure 1).
| Full Cohort | Sensitivity | Specificity | PPV | NPV | LR+ | LR− | Accuracy (%) |
|---|---|---|---|---|---|---|---|
| Age 55–77 yr and ever smoker | 1.0 | 0.88 | 0.25 | 1.0 | 8.5 | 0 | 89% |
| Age 55–77 yr and current smoker or former smoker quit <15 yr ago | 1.0 | 0.96 | 0.48 | 1.0 | 23 | 0 | 96% |
| Age 55–77 yr and current smoker | 0.56 | 0.97 | 0.44 | 0.98 | 20 | 0.45 | 96% |
| Age 55–77 yr and current smoker >0.5 ppd | 0.56 | 0.99 | 0.66 | 0.98 | 49 | 0.44 | 97% |
| Age 55–77 yr and current smoker >1 ppd | 0.44 | >0.99 | 0.99 | 0.98 | 2295 | 0.56 | 98% |
| Age-restricted Cohort (55–77 yr old) | Sensitivity | Specificity | PPV | NPV | LR+ | LR− | Accuracy (%) |
| Ever smoker | 1.0 | 0.56 | 0.25 | 1.0 | 2.3 | 0 | 62% |
| Current smoker or former smoker quit <15 yr ago | 1.0 | 0.84 | 0.48 | 1.0 | 6.2 | 0 | 86% |
| Current smoker | 0.56 | 0.89 | 0.44 | 0.93 | 5.3 | 0.49 | 85% |
| Current smoker >0.5 ppd | 0.56 | 0.96 | 0.66 | 0.94 | 13 | 0.46 | 91% |
| Current smoker >1 ppd | 0.44 | 0.99 | 0.99 | 0.92 | 610 | 0.56 | 93% |
Figure 1. Receiver operator characteristic (ROC) curves for prediction of lung cancer screening eligibility using age and smoking status. ROC curve A: ROC curve for prediction of U.S. Preventive Services Task Force/Centers for Medicare and Medicaid Services lung cancer screening eligibility in the National Health and Nutrition Examination Survey, using smoking status (current, former, or never) and age as predictors in the total cohort. Presented with the area under the curve (AUC). ROC curve B: ROC curve for prediction of U.S. Preventive Services Task Force/Centers for Medicare and Medicaid Services lung cancer screening eligibility in the National Health and Nutrition Examination Survey, using smoking status (current, former, or never) in cohort restricted by age eligibility (55–77 years). Presented with the AUC.
[More] [Minimize]When the analysis was restricted to age-eligible subjects (age 55–77 yr), the specificity and accuracy of smoking status alone were lower. For example, using the criterion of ever smoking in this group had a sensitivity of 100%, specificity of 56%, and accuracy of 62%. However, smoking status alone, when used in the age-eligible group, retained a fair overall predictive value (AUC 0.85; 95% CI, 0.84–0.86) (Figure 1). Despite a high degree of variability in eligibility by sociodemographic group, we found only modest differences in performance of these tests by demographic group (see Table E1 in the online supplement).
In this study, we used interview data from NHANES to determine the predictive value of simplified prescreening criteria to estimate full lung cancer screening eligibility criteria in a nationally representative cohort. In prior studies of lung cancer screening implementation, referring providers have endorsed the need for EHR-based systems to identify potentially eligible individuals, and EHR provider reminders have strong evidence in other screening modalities (12, 13). These simplified criteria may be useful as prescreening tools to identify potentially eligible persons for formal assessments of eligibility.
We found these criteria to have high accuracy. Even the simplest criteria—smoking status and age—had an AUC of 0.92. Because this predictive value is partially driven by age (which is much easier to screen by in clinical settings), we also examined the criteria in the restricted age-eligible cohort. Although the predictive value was lower, smoking status alone maintained a fair predictive value, with an AUC of 0.85. Each of the five simplified prescreening tools has advantages and disadvantages, and each could be selected by health systems based on different priorities. For example, in a well-resourced setting that attempts to screen all eligible individuals, it may be useful to use the entirely sensitive prescreening criteria of age 55–77 years and ever-smoker to further evaluate individuals for full eligibility. However, if a health system has more limited resources and wishes to initially target those who may benefit most from screening, approaches to prescreening may start with current smokers (14), acknowledging that a sizable number of potentially eligible individuals will be missed. This approach would improve efficiency and overall accuracy, and would also target a group of current smokers who would benefit from smoking-cessation resources. Using additional criteria, such as the quit date for former smokers with or without current smoking intensity, could further increase accuracy, but this information may be less readily available in the EHR.
The predictive value of a test depends largely on the prevalence of the condition of interest. Given the overall low prevalence of eligibility (3.8%), the criteria all have high negative predictive values (i.e., if the criteria are negative, the individual is unlikely to be eligible). However, they have a much less favorable positive predictive value (i.e., just because the criteria are positive does not mean an individual is necessarily eligible). Including information on pack-years greatly improves the positive predictive values (criteria 4 and 5). Therefore, we envision that these criteria (particularly criteria 1–3, which do not require any measure of intensity) would be most useful to a medical system to systematically generate EHR alerts from more accurately recorded EHR data, and serve as a prompt to inform providers and staff of the need to assess screening eligibility in a particular patient. This may save both time and resources, and limit the number of patients who are required to undergo a detailed survey of smoking history with a formal assessment of smoking pack-years. A specific measurement of pack-years is unnecessary for most patients, as it does not generally serve other important functions in risk assessment or treatment decision making, and may not be an effective use of time or resources.
This analysis demonstrates that although these criteria are quite simple and crude, they do have value in this setting, where full eligibility cannot be systematically determined. It is an important caveat that these criteria depend on an accurate recording of smoking status (current, former, or never), and national data suggest that this is only the case for 78% of outpatient encounters (15). However, this may improve with CMS reporting requirements targeting a screen for tobacco use as an electronic clinical quality measure (16). It will be useful in future studies to examine these criteria in real-world community practice settings to determine their performance using real EHR-based data.
In our analysis, we also found that the overall prevalence of U.S. adults who met screening criteria using NHANES (8.9 million) was consistent with other estimates obtained using national data (17), which suggests that our criteria for identifying eligibility were accurate. Although we note that eligibility differed prominently among sociodemographic groups, these prescreening criteria performed similarly across populations, so they may be useful across a range of clinical environments to identify eligible persons. White men are currently more likely to be eligible than other groups; however, the demographics of persons eligible for screening will likely continue to shift, as current smoking is becoming increasingly concentrated in certain minority groups and those of lower socioeconomic status (18, 19). Even at present, our results suggest that individuals in the lowest education and income strata are twice as likely as those in the highest strata to be eligible for lung cancer screening. It is important to note that eligibility for screening may not equitably reflect actual lung cancer risk across groups, which has been demonstrated for black Americans (20).
Although improving the ease and accuracy of identifying eligible individuals has the potential to improve screening uptake, there are several additional and important barriers that providers face in performing or referring for lung cancer screening. On the provider side, there are a number of time and resource barriers unique to lung cancer screening that may be additive with difficulty in identifying eligible persons. These include the time-intensive process of performing and documenting shared decision-making, limited resources for screening management and tracking, competing priorities, and limited knowledge about lung cancer screening (13, 21). Providers also face barriers related to downstream effects of screening, including a high rate of false-positive results leading to resource-intensive nodule follow-up (22–24). Screening programs with more resources and staff to provide individual patient navigation have been more successful at overcoming these barriers (25). Despite all the other provider barriers, there is some evidence that improving documentation of eligibility (pack-years) can lead to improved screening uptake. For example, a published evaluation of a large regional health system demonstrated improvement in screening rates from 2.8% to 7.3% over 2 years with attention to improved smoking history documentation (26).
Barriers to lung cancer screening are not limited to providers. As our results suggest, unlike other screening modalities defined by age and sex, lung cancer screening eligibility is concentrated in persons of lower socioeconomic status. These individuals face barriers to healthcare access and have demonstrated lower adherence to recommended screening for myriad reasons. Moreover, current smokers may have different attitudes toward screening despite an acknowledgment of lung cancer risk (27). Fatalism, perceived blame, and stigma may all play a role in current smokers’ avoidance of screening (28). Early reports suggest that current smokers may be less likely to be screened or adherent to screening follow-up (9). Although the current study may address a provider barrier, it is important to acknowledge there are several patient barriers that contribute to poor uptake.
There are limitations to this study. Although NHANES is reflective of the U.S. population, the population does not necessarily reflect individuals who actively seek care in medical systems; however, we found that these criteria performed similarly in different subpopulations. Our definition of pack-years was also limited to questions posed in the NHANES study. Therefore, smoking intensity was based on the prior month’s behavior among current smokers, rather than an estimated longitudinal measure. Also, individual preferences, comorbidities, and life expectancy should all be considered in lung cancer screening eligibility, and these elements cannot be accurately reflected in this study.
In this study, we demonstrated the accuracy of simplified prescreening criteria for predicting full lung cancer screening eligibility using a representative national cohort. These criteria have the potential to improve screening referrals, as an essential eligibility criterion, pack-years, is inaccurately and incompletely recorded in most EHRs. Further prospective studies are needed to examine these criteria in a clinical setting and their impact on referrals.
| 1 . | Aberle DR, Adams AM, Berg CD, Black WC, Clapp JD, Fagerstrom RM, et al.; National Lung Screening Trial Research Team. Reduced lung-cancer mortality with low-dose computed tomographic screening. N Engl J Med 2011;365:395–409. |
| 2 . | Moyer VA; U.S. Preventive Services Task Force. Screening for lung cancer: U.S. Preventive Services Task Force recommendation statement. Ann Intern Med 2014;160:330–338. |
| 3 . | Centers for Medicare and Medicaid Services. Decision memo for screening for lung cancer with low dose computer tomography; 2015 [accessed 2018 May 25]. Available from: https://www.cms.gov/medicare-coverage-database/details/nca-decision-memo.aspx?NCAId=274&bc=AAAA. |
| 4 . | De Koning H, Van Der Aalst C, Ten Haaf K, Oudkerk M. Effects of volume CT lung cancer screening: mortality results of the NELSON randomized-controlled population based trial. J Thoracic Oncol 2018;13:S185.journal |
| 5 . | Jemal A, Fedewa SA. Lung cancer screening with low-dose computed tomography in the United States—2010 to 2015. JAMA Oncol 2017;3:1278–1281. |
| 6 . | Pham D, Bhandari S, Malgorzata O, Pinkston CM, Kloecker GH. Lung cancer screening rates: data from the Lung Cancer Screening Registry. J Clin Oncol 2018;36:6504. |
| 7 . | Huo J, Shen C, Volk RJ, Shih YT. Use of CT and chest radiography for lung cancer screening before and after publication of screening guidelines: intended and unintended uptake. JAMA Intern Med 2017;177:439–441. |
| 8 . | Gould MK, Sakoda LC, Ritzwoller DP, Simoff MJ, Neslund-Dudas CM, Kushi LH, et al. Monitoring lung cancer screening use and outcomes at four Cancer Research Network sites. Ann Am Thorac Soc 2017;14:1827–1835. |
| 9 . | Modin HE, Fathi JT, Gilbert CR, Wilshire CL, Wilson AK, Aye RW, et al. Pack-year cigarette smoking history for determination of lung cancer screening eligibility. Comparison of the electronic medical record versus a shared decision-making conversation. Ann Am Thorac Soc 2017;14:1320–1325. |
| 10 . | Brenner AT, Malo TL, Margolis M, Elston Lafata J, James S, Vu MB, et al. Evaluating shared decision making for lung cancer screening. JAMA Intern Med 2018;178:1311–1316. |
| 11 . | Centers for Disease Control and Prevention. National Health and Nutrition Examination Survey data [accessed 2018 May 25]. Available from: https://www.cdc.gov/nchs/nhanes/. |
| 12 . | Baron RC, Melillo S, Rimer BK, Coates RJ, Kerner J, Habarta N, et al.; Task Force on Community Preventive Services. Intervention to increase recommendation and delivery of screening for breast, cervical, and colorectal cancers by healthcare providers a systematic review of provider reminders. Am J Prev Med 2010;38:110–117. |
| 13 . | Triplette M, Kross EK, Mann BA, Elmore JG, Slatore CG, Shahrir S, et al. An assessment of primary care and pulmonary provider perspectives on lung cancer screening. Ann Am Thorac Soc 2018;15:69–75. |
| 14 . | Pinsky PF, Church TR, Izmirlian G, Kramer BS. The National Lung Screening Trial: results stratified by demographics, smoking history, and lung cancer histology. Cancer 2013;119:3976–3983. |
| 15 . | Bae J, Ford EW, Kharrazi HHK, Huerta TR. Electronic medical record reminders and smoking cessation activities in primary care. Addict Behav 2018;77:203–209. |
| 16 . | Centers for Medicare and Medicaid Services. Clinical quality measures basics 2019. [accessed 2019 May 30]. Available from: https://www.cms.gov/Regulations-and-Guidance/Legislation/EHRIncentivePrograms/ClinicalQualityMeasures.html. |
| 17 . | Ma J, Ward EM, Smith R, Jemal A. Annual number of lung cancer deaths potentially avertable by screening in the United States. Cancer 2013;119:1381–1385. |
| 18 . | Jamal A, Phillips E, Gentzke AS, Homa DM, Babb SD, King BA, et al. Current cigarette smoking among adults—United States, 2016. MMWR Morb Mortal Wkly Rep 2018;67:53–59. |
| 19 . | Zhuang YL, Gamst AC, Cummins SE, Wolfson T, Zhu SH. Comparison of smoking cessation between education groups: findings from 2 US National Surveys over 2 decades. Am J Public Health 2015;105:373–379. |
| 20 . | Li CC, Matthews AK, Rywant MM, Hallgren E, Shah RC. Racial disparities in eligibility for low-dose computed tomography lung cancer screening among older adults with a history of smoking. Cancer Causes Control 2019;30:235–240. |
| 21 . | Kanodra NM, Pope C, Halbert CH, Silvestri GA, Rice LJ, Tanner NT. Primary care provider and patient perspectives on lung cancer screening. A qualitative study. Ann Am Thorac Soc 2016;13:1977–1982. |
| 22 . | Wiener RS, Gould MK, Slatore CG, Fincke BG, Schwartz LM, Woloshin S. Resource use and guideline concordance in evaluation of pulmonary nodules for cancer: too much and too little care. JAMA Intern Med 2014;174:871–880. |
| 23 . | Golden SE, Wiener RS, Sullivan D, Ganzini L, Slatore CG. Primary care providers and a system problem: a qualitative study of clinicians caring for patients with incidental pulmonary nodules. Chest 2015;148:1422–1429. |
| 24 . | Kinsinger LS, Anderson C, Kim J, Larson M, Chan SH, King HA, et al. Implementation of lung cancer screening in the Veterans Health Administration. JAMA Intern Med 2017;177:399–406. |
| 25 . | Gesthalter YB, Koppelman E, Bolton R, Slatore CG, Yoon SH, Cain HC, et al. Evaluations of implementation at early-adopting lung cancer screening programs: lessons learned. Chest 2017;152:70–80. |
| 26 . | Li J, Chung S, Wei EK, Luft HS. New recommendation and coverage of low-dose computed tomography for lung cancer screening: uptake has increased but is still low. BMC Health Serv Res 2018;18:525. |
| 27 . | Silvestri GA, Nietert PJ, Zoller J, Carter C, Bradford D. Attitudes towards screening for lung cancer among smokers and their non-smoking counterparts. Thorax 2007;62:126–130. |
| 28 . | Quaife SL, Marlow LAV, McEwen A, Janes SM, Wardle J. Attitudes towards lung cancer screening in socioeconomically deprived and heavy smoking communities: informing screening communication. Health Expect 2017;20:563–573. |
*K.C. is an Associate Editor and D.H.A. is a Deputy Editor of AnnalsATS. Their participation complies with American Thoracic Society requirements for recusal from review and decisions for authored works.
Supported by the Clinical Research Division of Fred Hutchinson Cancer Research Center.
Author Contributions: M.T. contributed to the conception of the work and conducted the analysis. D.H.A. contributed to the conception of the work. All of the authors contributed to interpreting the data and drafting and revising the manuscript, and approved the submitted version of the manuscript. All of the authors agree to be accountable for all aspects of the work.
This article has an online supplement, which is accessible from this issue’s table of contents at www.atsjournals.org.
Author disclosures are available with the text of this article at www.atsjournals.org.
