It is unclear whether patients and providers have started to knowingly request lung cancer screening outside US guidelines and insurance coverage for risk factors besides a history of heavy smoking. The authors analyzed their institution's best practices advisory clinical decision support system to determine whether providers knowingly order guideline-discordant LCS and the indications given. Results showed referring providers knowingly ordered screening CT examinations outside Medicare coverage and US guidelines, including for never smokers, for indications including a family history of lung cancer. LCS programs may need tailored strategies to guide these patients and providers, such as help with cancer risk assessment.
It is unclear whether patients and providers have started to knowingly request lung cancer screening (LCS) outside US guidelines and insurance coverage for risk factors besides a history of heavy smoking. The authors analyzed their institution's best practices advisory (BPA) clinical decision support system to determine whether providers knowingly order guideline-discordant LCS and the indications given.
CT examinations ordered for LCS at an academic medical center that triggered BPA alerts from November 2018 to December 2019 were reviewed. Alerts were triggered by attempts to order examinations outside Medicare coverage, which resembles most US guidelines. Providers can override alerts to order the examinations. Primary outcomes were the number of examinations performed using orders with overridden BPA alerts and indications given. Qualitative exploratory and directed content analyses identified motivators and decision-making processes that drove guideline-discordant screening use.
Forty-two patients underwent guideline-discordant LCS, constituting 1.9% of all patients screened (42 of 2248): 42.9% (18 of 42) were <54 or >77 years old, 14.3% (6 of 42) had never smoked, 40.5% (17 of 42) had quit >15 years earlier, and 31% (13 of 42) had smoked <30 pack-years; 45.2% (19 of 42) fell outside all US guidelines. The most common indication was a family history of lung cancer (21.4% [9 of 42]). Perceptions of elevated cancer risk from both patients and referring providers drove guideline-discordant screening use.
Referring providers knowingly ordered screening CT examinations outside Medicare coverage and US guidelines, including for never smokers, for indications including a family history of lung cancer. LCS programs may need tailored strategies to guide these patients and providers, such as help with cancer risk assessment.
Cigarette smoking is the leading cause of lung cancer.1 For high-risk current and former smokers, the National Lung Screening Trial (NLST) showed that annual screening with low-dose CT (LDCT) can reduce lung cancer mortality by 20%2, an outcome strengthened by recently reported results of the Dutch-Belgian NELSON trial.3 Since publication of the NLST results in 2010, lung cancer screening (LCS) has received broad support from professional societies, a positive recommendation from the US Preventive Services Task Force4, and coverage under Medicare5 and by other insurers6. In general, US screening guidelines and insurance coverage criteria reflect NLST enrollment criteria, with minor variations in the age to stop screening (Table 1).
Note: Traditional Medicaid is not included because coverage may vary among plans. AATS = American Association for Thoracic Surgery; ACCP = American College of Chest Physicians; ACS = American Cancer Society; ASCO = American Society of Clinical Oncology; BPA = best practices advisory; NA = not applicable; NCCN = National Comprehensive Cancer Network; NLST = National Lung Screening Trial; USPSTF = US Preventive Services Task Force.
NCCN guidelines also recommend screening for individuals ≥50 years of age with ≥20-pack-year smoking histories and additional risk factors other than secondhand smoke exposure that increase the risk of lung cancer to ≥1.3% over a 6-year time frame  (as determined using the Tammemägi lung cancer risk prediction model ).
AATS guidelines recommend screening for all former smokers 55 to 79 years of age with ≥30-pack-year histories regardless of number of years since quitting . These guidelines also recommend screening for individuals starting 5 years after resection of stages 1A to IIIA non-small-cell lung cancer and for those ≥50 years of age with ≥20-pack-year smoking histories and additional factors that increase cumulative lung cancer risk to ≥5% within 5 years .
At our institution, BPA alerts are triggered when lung cancer screening CT is ordered outside of these criteria.
The potential benefits and harms of LCS for individuals with lighter smoking histories than the NLST cohort and for those with other risk factors, including chronic obstructive pulmonary disease, radon exposure, occupational exposure such as asbestos, and family history of lung cancer, are uncertain.1 Nevertheless, with growing public awareness and continued nationwide expansion of screening programs, LCS has started to occur outside US guidelines.7 In some cases, this may reflect insufficient patient and provider knowledge of screening guidelines. However, it is possible that providers have started to knowingly request guideline-discordant screening for patients without histories of heavy smoking because of the presence of other lung cancer risk factors. This possibility has not been well studied.
Clinical decision support (CDS) systems deliver knowledge and patient-specific information to health care providers to aid in clinical decision making, including appropriate imaging utilization.8,9 In November 2018, a best practices advisory (BPA) CDS system was integrated into the electronic ordering of LCS CT examinations at our academic medical center to promote the appropriate use of LCS and help ensure that examinations will be covered by insurers. Alerts are triggered in the electronic medical record at the point of order entry when referring providers attempt to order screening CT studies for patients outside of Medicare criteria (Table 1). Providers may acknowledge alerts, override them to order the examinations, and provide indications for screening (eg, family history of lung cancer, occupational exposure to carcinogens). The aim of this study was to investigate whether referring providers at our institution knowingly order screening examinations outside of insurance coverage criteria and US guidelines despite receiving BPA alerts and the provided indications for screening.
Study Design and Setting
This was a HIPAA-compliant, institutional review board–approved retrospective review of BPAs triggered by our tertiary care academic medical center's electronic medical record–based LCS CT order entry system (Epic Radiant; Epic, Verona, Wisconsin) from November 2018 through December 2019. Starting in November 2018, a BPA alert system was integrated into the order entry system for LDCT studies for LCS (examinations with Current Procedural Terminology code G0297). Conditions that trigger BPA alerts include ordering screening CT examinations for never smokers, former smokers who quit >15 years earlier, all smokers with <30 pack-year smoking histories, and individuals <55 or >77 years old; these are the age and smoking history criteria used by Medicare. The BPA system uses smoking history details that must be entered by the provider into corresponding screening CT order requisition data fields. BPA alert messages appear when providers attempt to sign orders that meet trigger conditions. These messages alert the provider that the patient does not meet criteria for insurance reimbursement and list the required age and smoking history criteria. Providers can complete orders following triggered alerts by either overriding the alerts or revising or placing new orders that do not trigger alerts (eg, the alert was triggered by a typographic error in the original order). When providers choose to override alerts, they are then asked to give indications for ordering screening CT. Providing an indication is not mandatory for overriding a BPA alert.
The sample population was composed of all patients who received orders for LCS CT after BPA alerts, triggered because of violations of age and/or smoking history criteria, were overridden by the referring providers. We excluded patients whose screening CT orders were canceled and replaced with nonscreening chest CT examinations and whose screening CT examinations were ordered at 54 years of age but performed after age 55. We also excluded patients who met age and smoking history criteria but had triggered BPA alerts because referring providers indicated that they were not asymptomatic for lung cancer, because this may have been caused by a misreading of the pertinent question in the order entry system and because the focus of this study was on examinations ordered outside of age and smoking history criteria.
Primary outcomes were the number of individuals who underwent screening CT by 3-month follow-up using orders with overridden BPA alerts and indications provided for overriding the alerts. We also analyzed the Lung CT Screening and Reporting Data System (Lung-RADS™)  categories of the performed examinations.
The American Association for Thoracic Surgery (AATS) and the National Comprehensive Cancer Network (NCCN) use broader screening criteria compared with CMS and other organizations, detailed in Table 1. The number of screening CT examinations ordered and performed after BPA alerts were overridden for patients who met additional AATS and/or NCCN criteria were analyzed. Our analysis considered individuals to fall within AATS and NCCN criteria if they were ≥50 years old with ≥20 pack-year smoking histories and additional lung cancer risk factors documented in the order requisitions. The individual lung cancer risk calculations incorporated into those guidelines were not performed for this study. Additionally, we considered former smokers ages 55 to 79 with ≥30 pack-year smoking histories (regardless of number of years since quitting) and all individuals with personal histories of lung cancer as documented in the order requisition to meet AATS criteria. We also analyzed the number of individuals who fell outside all US guidelines and CMS criteria.
Independent variables included potential predictors of guideline-discordant LCS utilization among patients and referring providers. For patients, sociodemographic variables included age, sex, race/ethnicity, primary language, and insurance status. For providers, characteristics included number of patients referred to LCS over the study period, degree type, years since medical school graduation, and specialty.
The electronic medical record and an integrated BPA alert report generator were reviewed to identify triggered alerts during the study period. Smoking history details, indications provided for overriding BPA alerts, and additional risk factors for lung cancer were extracted from the LCS CT order in the electronic medical record. Retrospective review of the electronic medical record was performed to identify screening CT examinations performed and the outcomes of these examinations and to extract information from clinical notes authored by referring providers for qualitative analysis. Overall LCS CT volume was extracted from the electronic medical record as all examinations ordered within the study period and performed at the time of 3-month follow-up. Characteristics of referring providers were extracted from publicly available records linked to individual National Provider Identifier numbers.
A qualitative content analysis was conducted on open-text data extracted from the electronic medical records of patients who underwent guideline-discordant screening. Specifically, clinical notes authored by the referring providers who ordered the screening CT examinations were analyzed. Two coders independently coded each data element to systematically categorize the open-text data. Coding used both an exploratory and a directed content analysis approach.11 In an exploratory content analysis approach, a coding framework is established from codes that are identified and then categorized. In a directed content analysis approach, codes are determined to be important before analysis and assigned to the data when evident. For the current study, exploratory coding was used to identify primary motivators for ordering screening CTs; directed coding was used to determine whether the referring provider or patient or both drove the decision to order LDCT.
To iteratively develop a coding framework, patients were randomly assigned into three waves (wave 1, 20% of sample; wave 2, 40% of sample; wave 3, 40% of sample). Wave 1 was used to create the coding framework, wave 2 was used to refine the coding framework, and wave 3 was used to confirm the coding framework, exhaustively capturing the reasons why screening CT examinations were ordered. Where appropriate, the decision-making process for the screening CT examinations was coded. After each wave, coders met to discuss their coding and identify discrepancies before agreeing on the final code assigned to each case. This is referred to as the constant comparative method and involves validated qualitative phases of data familiarization, condensation, categorization, and interpretation before final code assignment. For final validation, preliminary results were presented to the lead author to determine consensus on how codes had been operationalized and applied.
Characteristics of patients who underwent LCS and providers who ordered these examinations were summarized with means, medians, and proportions. Given small sample sizes, bivariate logistic regression analyses instead of multiple-variable regression analyses were performed to evaluate independent predictors of guideline-discordant lung screening utilization among patients and referring providers. P < .05 was considered to indicate statistical significance. Analyses were conducted using Stata version 11 (StataCorp, College Station, Texas). Sample sizes were derived on the basis of convenience.
Screening CT examinations were ordered after acknowledging and overriding BPA alerts for 55 individuals. Of these orders, 20% (11 of 55) were not scheduled and 3.6% (2 of 55) expired. Guideline-discordant screening CT studies were performed for 76.4% of these individuals (42 of 55), constituting the study cohort used for further analysis. This accounted for 1.9% of all screening CT examinations (42 of 2248) performed during the study period.
The mean age of patients who underwent guideline-discordant screening CT was 65 ± 11.3 years; 47.6% were women (20 of 42). Among them, 28.6% (12 of 42) were younger than 55 years of age, 14.3% (6 of 42) were older than 77 years, 14.3% (6 of 42) had never smoked, 40.5% (17 of 42) had quit >15 years earlier, and 31% (13 of 42) had smoked <30 pack-years (Fig. 1). In this cohort, 28.6% (12 of 42) fell outside of more than one of the following criteria: ages 55 to 77 years, ≥30-pack-year history, and quit within 15 years. The characteristics of these patients are summarized in Table 2. There were no statistically significant differences in the sociodemographic characteristics of those who underwent screening with orders that triggered BPA alerts compared with those with orders that did not trigger alerts (Table 2).
Note: Data are expressed as mean ± SD or as percentage (number). BPA = best practices advisory; LCS = lung cancer screening.
Given small sample sizes of nonwhite groups, comparison was performed between whites and nonwhites.
Given small sample sizes of non-English primary languages, comparison was performed between English and non-English.
Because Medicaid status is a marker of socioeconomic status, comparison was performed between Medicaid and Medicare-Medicaid and other types of insurance.
Among all patients, 21.4% (9 of 42) met the broader NCCN and AATS age and pack-year criteria for those with additional lung cancer risk factors (age ≥ 50 years with ≥20-pack-year smoking history). These risk factors were family history of lung cancer (77.8% [7 of 9]), exposure to carcinogens because of occupation as a firefighter (11.1% [1 of 9]), and exposure to asbestos (11.1% [1 of 9]). LCS is also recommended by the AATS for all former smokers ages 55 to 79 years with ≥30-pack-year smoking histories regardless of number of years since quitting. In our cohort, an additional 31% (13 of 42) of patients met these criteria.
No US guidelines support screening for never smokers without personal histories of lung cancer, smokers older than age 80, or smokers younger than age 50 and/or with <20-pack-year smoking histories, even with the presence of additional lung cancer risk factors (Table 1). In our cohort, 45.2% of all screening CT examinations (19 of 42) were ordered for patients who fell outside the age and smoking history criteria for all US guidelines, and 73.8% (31 of 42) were outside of all guidelines except for those of the AATS.
Indications provided by referring providers for overriding BPA alerts and ordering guideline-discordant screening CT were as follows: family history of lung cancer (21.4% [9 of 42]), occupational exposure to carcinogens as a firefighter (9.5% [4 of 42]), thoracic radiation therapy (4.8% [2 of 42]), asbestos exposure (2.4% [1 of 42]), and elevated lung cancer serum markers per patient self-report of laboratory tests performed outside the United States (2.4% [1 of 42]). For 59.5% of patients (25 of 42), the order requisition only documented LCS as the reason for examination and did not provide information on why screening was indicated or why the BPA alert was overridden. For 9.5% (4 of 42), referring providers noted that the patient was willing to self-pay for screening.
Exploratory qualitative analysis identified three categories of primary motivators for guideline-discordant screening CT orders: recommendation from a health care provider other than the referring provider, the referring provider's perception of the patient's lung cancer risk, and the patient's perception of his or her personal lung cancer risk. These are summarized with example texts extracted from referring provider notes in Table 3. Directed analysis identified three decision-making processes that drove screening CT orders (Table 4). First, there was documentation indicating that some providers endorsed screening despite being aware that their patients were outside guidelines. Second, we found documentation suggesting that some patients may be primarily responsible for driving the decision to screen. Finally, in some instances, documentation in the electronic medical record suggested that both the patient and the referring provider jointly arrived at the decision to screen.
Note: BPA = best practices advisory; CXR = chest radiography; PY = pack-years of cigarettes smoked.
Note: BPA = best practices advisory; CXR = chest radiography; Pt = patient; PY = pack-years of cigarettes smoked.
Guideline-discordant screening CT was ordered by 31 referring providers. Characteristics of these providers compared with those whose screening CT orders did not trigger BPA alerts are summarized in Table 5. Among referring providers, increasing numbers of years since medical school graduation was associated with higher odds of ordering guideline-discordant screening (odds ratio, 1.038; 95% confidence interval, 1.007-1.070; P = .015). Increasing overall number of patients referred for screening was also associated with guideline-discordant screening (odds ratio, 1.066; 95% confidence interval, 1.027-1.106; P = .001). There were no statistically significant differences in whether providers ordered guideline-discordant screening when analyzed by medical specialty (P = .083) or degree type (P = .629).
Note: Data are expressed as percentage (number) except as indicated. BPA = best practices advisory; IQR = interquartile range; LCS = lung cancer screening; NP = nurse practitioner.
Given small sample sizes of non-MD degrees, comparison was performed between MD and non-MD.
Given small sample sizes, comparison was performed between general internal medicine and other specialties.
All examinations in our cohort were reported using Lung-RADS, with 9.5% (4 of 42) assessed as Lung-RADS category 1, 81% (34 of 42) as category 2 or 2S, 2.4% (1 of 42) as category 3 or 3S, 4.8% (2 of 42) as category 4A or 4AS, and 2.4% (1 of 42) as category 4B. One of the patients categorized as Lung-RADS 4A was a 79-year-old current smoker with a 61-pack-year history. This patient was recommended for 3-month follow-up chest CT, which showed the initial category-defining nodule to be atelectasis and did not recommend any additional follow-up or management. The other Lung-RADS category 4A patient was a 53-year-old never smoker who worked as a firefighter. This patient underwent pulmonology evaluation including transbronchial biopsy, which revealed sarcoidosis. The Lung-RADS category 4B patient was a 79-year-old current smoker with a 35-pack-year history. This patient was recommended for 1-month follow-up chest CT to address potentially infectious or inflammatory conditions, with resolution of the previously seen findings documented on follow-up CT.
At our institution, referring providers have begun to order LCS CT knowingly outside of Medicare coverage and US screening guidelines for indications that include lung cancer risk factors other than a history of heavy smoking. Although variations exist among US guidelines, nearly half of our cohort fell outside the age and smoking history criteria of all guidelines, and 3 in 4 were outside all except those of the AATS. Notably, 1 in 7 patients were never smokers. Guideline-discordant screening CT occurred but was uncommon, accounting for 2 of every 100 screening examinations performed over the study period.
Both quantitative and qualitative analyses suggest an individual risk–based approach to screening by these patients and referring providers. Personal and family histories of lung cancer, occupational exposure, and asbestos exposure are all lung cancer risk factors incorporated by the NCCN12 and/or AATS13 into their guidelines, which support screening starting at age 50 with a 20-pack-year history if these or other risk factors increase cumulative lung cancer risk. These guidelines incorporate risk prediction models and reason that patients outside of NLST enrollment criteria should still be recommended for screening if their lung cancer risk is similar to the trial cohort.12,13 However, the American College of Chest Physicians cautioned that some patients assessed as high risk by modeling may have less favorable benefit-versus-harm profiles than the cohort identified by its eligibility criteria, which reflect the NLST criteria.12 Furthermore, per the ACR Appropriateness Criteria, there is insufficient evidence to guide screening use among this population of younger, lighter smokers, though it does conclude that screening for these patients may be appropriate.14 Finally, no US guidelines recommend screening before age 50, or with less than a 20-pack-year smoking history, or for never smokers without personal histories of lung cancer even in the presence of other risk factors. Thus, screening is unlikely to bring a net benefit to half of the patients in our study, even when assessed through an individual risk–based approach.
Although steadily increasing, utilization of LCS among eligible individuals nationwide remains low.15 Thus, LCS stakeholders must continue to focus on the expansion of screening programs and overcome existing barriers.16 At the same time, it is important to avoid screening in cases in which potential harms may outweigh the benefits. CDS systems can encourage appropriate imaging utilization across health care systems.8 As highlighted by this study, CDS systems including the BPA system described here can also be analyzed to identify specific groups of referring providers for targeted outreach to further promote appropriate use. For example, our analysis identified a group of providers who may benefit from tailored guidance and education on the issue of LCS for patients with lung cancer risk factors who fall outside of current guidelines. Support for providers on this topic could come from program navigators, clinical decision aids, educational resources such as provided by the ACR17 and other organizations6, and/or online case-based educational tools, exemplified by Radiology-TEACHES.18 This can facilitate shared decision making when providers encounter patients inquiring or requesting screening because of lung cancer risk factors other than a heavy smoking history and ultimately encourage appropriate, guideline-concordant screening.
We did not find a statistically significant association between patient characteristics and guideline-discordant screening use. On the provider side, increased years since medical school graduation and greater number of patients referred for screening increased the odds of ordering guideline-discordant examinations. It is possible that experienced providers may be more inclined to refer patients for screening on the basis of their own and their patients' perceptions of cancer risk in situations in which guidelines do not offer clear and consistent guidance, such as for light smokers with strong family histories or prolonged occupational exposures to carcinogens (eg, firefighters).
Indeed, our analyses found that perceptions of lung cancer risk from both patients and referring providers contributed to guideline-discordant screening use. This suggests that it may be useful to provide resources that can guide both groups on the assessment of lung cancer risk, especially within the context of weighing potential benefits and harms of screening. Such complex, individualized considerations of risk and whether to screen may benefit from consultation with specially trained personnel associated with LCS programs, such as from pulmonary nodule clinics.
Radiologists interpreting screening examinations should also be aware that some may be ordered for patients outside screening guidelines to help avoid the use of Lung-RADS for reporting. This system was developed by the ACR for LCS to standardize the reporting and management of screen-detected findings. For patients who do not meet criteria for screening, it may be more appropriate to manage CT findings according to Fleischner Society guidelines19 or usual institutional practice for incidental pulmonary findings on routine chest CT examinations. All LCS CT examinations in this study were reported using Lung-RADS, indicating the need to evaluate this use at our own institution.
The performance of our study at a single academic medical center limits the generalizability of our results. However, ours is a high-volume screening program with multiple strategies, including educational outreach, CDS, a multidisciplinary nodule clinic, and a navigator to expand screening while promoting appropriate use. In these respects, our program likely resembles other programs nationwide. Furthermore, LCS CT orders originated from a wide spectrum of providers at our institution. These support the generalizability of our findings.
In analyzing patients who may have met NCCN and/or AATS guidelines, we did not incorporate the recommended individual risk calculations, which we believed to be outside the scope of this study. For example, the risk calculator cited by the NCCN includes data not collected as part of this study, including body mass index, history of chronic obstructive pulmonary disease, and educational level.20 However, this affects only a small fraction of our cohort, with 1 in 5 patients meeting even the age and smoking history criteria of NCCN and AATS guidelines.
Finally, we are limited in our understanding of the reasons LCS CTs were ordered. Our conclusions were based on inference from the order requisitions and qualitative analyses of clinical notes. However, no indication was provided on the order requisition for nearly 60% of our cohort. Furthermore, qualitative analysis was limited to only the information documented by referring providers, which may not fully reflect the decision-making process leading to the order. Also, we assumed that providers understood that Medicare and other insurance coverage criteria reflect US guidelines and, therefore, receiving an alert that a patient is outside of coverage approximately equated to an alert that the patient is outside of screening guidelines. Although we believe this to be common knowledge among health care providers, this may not be the case for all. Ultimately, additional investigation is warranted to further understand the reasons providers ordered guideline-discordant examinations and to determine whether this occurs at other institutions. This will help advance efforts to promote appropriate screening utilization.
Referring providers have started to order LCS examinations for patients despite being alerted that they fall outside of insurance coverage and, by extension, outside of current US guidelines.
Indications provided for guideline-discordant screening examinations include lung cancer risk factors other than heavy smoking, such as family history of lung cancer and occupational exposure to carcinogens.
Guideline-discordant screening use was driven by perceptions of lung cancer risk from both patients and referring providers.
Radiologists and screening programs should be aware of demand for guideline-discordant screening to help guide patients and providers toward appropriate imaging utilization.
Sign in to take quiz and track your certificates
To help improve the quality of its educational content and meet applicable education accreditation requirements, the content provider will receive record of your participation and responses to this activity.
As the official journal of the American College of Radiology, JACR will inform, educate, and engage radiologists, giving them the information and tools they need to provide evidence-based, patient-centric imaging care while moving the science of radiology forward.
This content and the individual contributions contained in it are protected under copyright by the American College of Radiology.
To read more about imaging and patient-centered care, visit the Journal of the American College of Radiology.
Disclosure: Dr Fintelmann has received grants from the, the Society of Thoracic Radiology, and the Society of Interventional Oncology; personal fees from Jounce Therapeutics; and nonfinancial support from BTG plc, outside the submitted work. Dr Little is an academic textbook author and associate editor for Elsevier and receives royalties for his work. Dr Flores has received grant funding from the ACR Innovation Fund and the National Cancer Institute Research Diversity Supplement for work not related to the preparation of this report. All other authors state that they have no conflict of interest related to the material discussed in this article. Drs Wang, Neil, Fintelmann, Little, Narayan, and Flores are employees.
You currently have no searches saved.
You currently have no courses saved.