Abstract
Background
Early detection of lung cancer through management of pulmonary nodules (PNs) may reduce lung cancer mortality. We assessed the relationship between PNs and lung cancer.
Research Question
How common are PNs in the Medicare population? What is the rate of lung cancer after detection of PNs? What is the relative proportion of early-stage lung cancer diagnosed after reporting of PNs vs through low-dose CT (LDCT) scan screening?
Study Design and Methods
Using the Surveillance Epidemiology and End Results Program-Medicare database, we defined two cohorts: those in the 5% sample with ≥ 12 months of Medicare Parts A and B coverage from 2014 through 2019 (5% sample cohort) and those with a diagnosis of lung cancer from 2015 through 2017 with coverage for the prior 18-month period (lung cancer cohort). We defined PNs as chest CT scans with accompanying codes of 793.11 (International Classification of Diseases [ICD], Ninth Revision) or R91.1 (ICD, Tenth Revision) denoting a solitary PN. Patients in the lung cancer cohort were classified by whether they had undergone LDCT scan screening and whether they had a diagnosis of PN or neither (reference) within 18 months before diagnosis. We compared cancer stage and survival across groups.
Results
Of 627,547 patients in the 5% sample cohort, 5.0% demonstrated PNs over median of 5.0 years of follow-up. Cumulative 1- and 2-year lung cancer rates after initial PN diagnosis were 3.2% and 4.7%, respectively. Of 44,194 patients in the lung cancer cohort, 15.7%, 2.9%, and 81.4% were in the PN, LDCT scan, and reference groups, respectively. Of patients in the PN, LDCT scan, and reference groups, 58.1%, 50.3%, and 24.4% respectively, had disease of a localized stage. Among all patients with localized disease, 30.0% and 4.9% were in the PN and LDCT scan and groups, respectively. Three-year lung cancer-specific survival rates were 75.0%, 75.6%, and 49.4% for the PN, LDCT scan, and reference groups.
Interpretation
Patients with lung cancer who received a diagnosis after identification of PNs tended to have localized disease. Of all patients with localized disease, almost one-third had PNs that were diagnosed previously, compared with 5% of patients who had undergone LDCT scan screening. PNs represent a relatively common presentation of potentially curable lung cancer.
Key Words: Chest CT scan, lung cancer, pulmonary nodules, stage
Take-home Points.
Study Question: How common are pulmonary nodules (PNs) among people with Medicare coverage who receive a diagnosis of lung cancer?
Results: About 16% of all patients with lung cancer and 30% of those with a diagnosis of localized disease had a PN detected within 18 months before diagnosis; comparable percentages for those with a prior low-dose CT scan screening were 2.9% and 4.9%, respectively.
Interpretation: Reporting of PNs may complement low-dose CT scan screening in detecting early-stage lung cancer.
Early detection of lung cancer with low-dose CT (LDCT) scan screening reduces lung cancer mortality.1, 2, 3 However, in the United States, where LDCT scan screening has been a covered benefit since 2015, screening rates remain low, limiting its population-level impact.4, 5, 6 Barriers to uptake include the cost of additional infrastructure, manpower, and processes required for safe and effective implementation; complicated and imperfect eligibility criteria; a paucity of infrastructure in the places with greatest need; and a knowledge gap about the benefits of LDCT scan screening among key stakeholders, including clinicians and at-risk patients.6, 7, 8
Another pathway to early lung cancer detection may be guideline-concordant management of incidentally detected pulmonary nodules (PNs) found on CT scans performed for reasons other than screening or evaluation for suspected lung cancer. In a large Mississippi Delta community-based prospective observational cohort study comparing patients in a guideline-adherent PN program with patients who underwent LDCT scan screening, 6% of patients in the PN program received a diagnosis of lung cancer within 2 years.9 Lung cancer was diagnosed more frequently through the PN program than through LDCT scan screening and the patient risk factor profiles differed, but the characteristics of lung cancer, including stage distribution and treatment, were strikingly similar.9
Using the Surveillance Epidemiology and End Results (SEER) Program-Medicare database, we examined the generalizability of these regional findings by evaluating the frequency of PNs and their relationship to future lung cancer. Specifically, we estimated rates of PNs by demographic and other factors and examined rates of lung cancer diagnosis after identification of PNs. We also assessed characteristics of lung cancers diagnosed after PNs were identified and compared them with those diagnosed after LDCT scan screening and those diagnosed after neither.
Study Design and Methods
SEER-Medicare Database
The SEER-Medicare database includes claims data on a 5% sample of all those enrolled in Medicare in SEER areas during a given coverage period; data through 2019 are available.10 Additionally, SEER-Medicare contains data on all cases of cancer in SEER areas, regardless of whether in the 5% sample, with data currently available on cases diagnosed through 2017.
5% Sample Cohort
We created a 5% sample cohort that included all patients (regardless of cancer status) in the 5% SEER-Medicare sample with ≥ 12 months of non-health maintenance organizaton (HMO) Medicare Parts A and B coverage from 2014 through 2019 at 65 years of age or older. We defined the presence of a reported PN as a diagnostic chest CT scan (procedure codes 71250, 71260, 71270, and 71275) accompanied on the same date by an International Classification of Diseases (ICD), Tenth Revision, code of R91.1. Through September 2015, we used the corresponding ICD, Ninth Revision, code of 793.11.11 However, because R91.1 covers only solitary PNs and multiple PNs may be coded as R91.8 (793.19 under ICD, Ninth Revision), as a sensitivity analysis, we also defined PNs as either codes R91.1 or R91.8 (793.11 or 793.19 under ICD, Ninth Revision). As an additional sensitivity analysis, we also defined PNs as occurring within ± 30 days from a chest CT scan instead of on the same date (using R91.1 and 793.11).
Chronic conditions for those in the cohort, specifically COPD and tobacco use disorder, were assessed using the Centers for Medicare and Medicaid Services-defined chronic condition algorithms (e-Table 1).10 To assess geographic effects, SEER registry regions were grouped as follows: West Coast (four California registries and Seattle), Northeast (New Jersey, New York, Connecticut, and Massachusetts), South (three Georgia registries, Kentucky, and Louisiana), and other (Alaska Native, Detroit Metro, Hawaii, Idaho, Iowa, New Mexico, Utah, and non-SEER areas). Note that to be included in the 5% sample, patients must have resided in SEER areas for some, but not necessarily all, of the coverage period.
Lung Cancer Cohort
The lung cancer cohort consisted of all patients in the SEER-Medicare database with invasive lung cancer diagnosed between 2015 and 2017 who were 66.5 to 79 years of age and had non-HMO Medicare Parts A and B coverage for the full 18-month period before diagnosis. PNs were defined as state previously for the 5% sample cohort, with a sensitivity analysis also including code R91.8 (793.19 for ICD, Ninth Revision). Patients who underwent LDCT scans within 18 months before diagnosis were classified as the LDCT scan group; those outside the LDCT scan group with PNs identified 30 days to 18 months before diagnosis constituted the PN-only group. For sensitivity analyses, we also considered times with 12 months and within 24 months of diagnosis. We used the 30-day limit to exclude PNs observed during the diagnostic workup for lung cancer. Patients in neither group constituted the reference group. We identified LDCT imaging by procedure codes G0297 or S8032; LDCT scan Lung Imaging Reporting and Data System scores were unavailable in the database. We used the upper age limit of 79 years to allow for better comparison of the LDCT scan and PN-only groups because few patients with a diagnosis at 80 years of age or older had undergone recent LDCT imaging.
Quantitative Methods
For the 5% sample cohort, we computed the proportions with reported PNs and chest CT scans, as well as the proportion with a PN among those with chest CT scans. Follow-up began the first month of non-HMO Medicare Parts A and B coverage at 65 years of age or older from 2014 through 2019 and ended the last month of such coverage or 30 days before any lung cancer diagnosis (to exclude PNs observed as part of the diagnostic workup). Proportions were computed by sex, race or ethnicity, age, COPD status, and geographic region, with the last three factors assessed at either the start of follow-up or at the first CT scan for the proportion with PNs among those with chest CT scans. To assess differences between the ICD, Ninth and Tenth Revision, periods, we examined the proportion of chest CT scans with PNs in each period.
We developed multivariate proportional hazards models to assess factors associated with undergoing a first chest CT scan and observation of a first PN. Model factors were sex, race or ethnicity, age, geographic region, COPD, and tobacco use disorder, with the last four factors assessed at the start of follow-up. Patients were censored at the end of follow-up. We also developed a multivariate logistic regression model among patients with chest CT scans to assess factors associated with having PNs. Factors in the model, assessed at the time of the initial CT scan, were sex, race or ethnicity, age, geographic region, COPD, and tobacco use disorder; the number of CT scans also was included.
We assessed lung cancer incidence after the initial PN in patients in the 5% sample cohort who had PNs through 2017 and at least 30 days before any lung cancer diagnosis and who lived in SEER regions at the time of initial PN detection. Follow-up began at first PN detection and continued until lung cancer diagnosis, death, or December 31, 2017. Cumulative incidence was estimated using Kaplan-Meier analysis with competing risks. We used proportional hazards models to assess factors associated with cumulative lung cancer risk, including sex, race or ethnicity, geographic region, age, COPD, and tobacco use disorder.
For the lung cancer cohort, we examined stage, histologic findings, and survival by prior PN (including solitary PN) and LDCT scan status. We computed cumulative aggregate and stage-stratified lung cancer-specific and overall survival rates using Kaplan-Meier analysis, with mortality follow-up until the end of 2018. We also developed proportional hazards models for lung cancer-specific survival for patients with localized disease, controlling for age, sex, race or ethnicity, COPD, tobacco use disorder, and histologic findings.
Results
5% Sample Cohort
Of 1,264,649 people in the 5% sample from 2014 through 2019, 629,953 people (49.8%) were included in the 5% sample cohort. The other 634,696 people (50.2%) were excluded because of fewer than 12 months of non-HMO Medicare Parts A and B coverage while 65 years of age or older; this included 174,366 people who were too young in 2019 to have had such coverage. Table 1 shows the demographics of the cohort; 58.5% were women and 80.2% were non-Hispanic White, compared with the overall 5% sample, which was 56.3% female and 74.9% non-Hispanic White. Median follow-up was 5.0 years (interquartile range, 2.8-6.0 years), with 2.76 million total person-years of follow-up.
Table 1.
Demographics of 5% Sample and Lung Cancer Cohorts
| Variable | 5% Sample Cohort | Lung Cancer Cohort |
|---|---|---|
| No. of patients | 629,953 | 44,194 |
| Sex | ||
| Female | 386,630 (58.5) | 22,019 (49.8) |
| Male | 261,323 (41.5) | 22,175 (50.2) |
| Age group, ya | ||
| 65-74 | 407,947 (64.8) | 28,399 (64.3) |
| 75+ | 222,006 (35.2) | 15,795 (35.7) |
| Race or ethnicity | ||
| Non-Hispanic White | 505,175 (80.2) | 38,241 (86.5) |
| Non-Hispanic Black | 48,643 (7.7) | 3,419 (7.7) |
| Hispanic | 17,033 (2.7) | 302 (0.7) |
| Asian | 25,669 (4.1) | 893 (2.0) |
| Other/unknown | 33,343 (5.3) | 1,339 (3.0) |
| Geographic regiona,b | ||
| Northeast | 214,982 (34.1) | 17,188 (38.9) |
| South | 97,821 (15.5) | 10,113 (22.9) |
| West Coast | 177,564 (28.2) | 10,265 (23.2) |
| Other | 139,586 (22.2) | 6,628 (15.0) |
| COPDa,c | 101,068 (21.1) | 26,479 (60.1) |
| Tobacco use disordera,c | 24,183 (5.1) | 17,325 (40.0) |
Data are presented as No. (%), unless otherwise indicated.
At start of follow-up for the 5% sample cohort and at diagnosis for the lung cancer cohort.
Northeast is Connecticut, Massachusetts, New Jersey, and New York. West Coast is California and Washington; and South is Georgia, Kentucky, and Louisiana.
Twenty-four percent and 26% had unknown status for COPD and tobacco use disorder, respectively, in the 5% sample cohort; 0.4% and 2% had unknown status for COPD and tobacco use disorder in the lung cancer cohort. Percentages exclude unknowns.
Table 2 shows the proportions of patients with chest CT scans and PNs. Overall, 26.3% of patients had at least one chest CT scan; the average number of chest CT scans among those with any CT scans was 2.2. PNs were reported for 31,322 patients, 5.0% of the entire cohort and 19.0% of the subset with chest CT scans. The rate of initial nodules was 11.3 per 1,000 person-years. Modest differences in chest CT scan use rates were observed by demographics, with non-Hispanic White patients, patients 75 years of age or older, and Northeast residents having the highest rates (Table 2). Those with COPD had almost twice the rate of chest CT scan use as those without (46.5% vs 25.2%); a similar pattern was seen for tobacco use disorder.
Table 2.
Rates, HRs, and ORs of Chest CT Scans and PNs in the 5% Sample Cohort
| Variable | Chest CT Scan (≥ 1) |
PN (≥ 1) |
PN (≥ 1) |
PN (≥ 1) Among Those With a CT Scan |
|
|---|---|---|---|---|---|
| No. (%) | No. (%) | Percentage Among Those With a CT Scan | Multivariate HR (95% CI) | Multivariate OR (95% CI) | |
| All | 165,643 (26.3) | 31,322 (5.0) | 19.0 | . . . | . . . |
| Sex | |||||
| Men | 71,194 (26.9) | 13,258 (5.1) | 19.0 | Reference | Reference |
| Women | 95,449 (25.9) | 18,064 (4.9) | 18.9 | 0.93 (0.91-0.96) | 1.08 (1.05-1.11) |
| Race or ethnicity | |||||
| Non-Hispanic White | 138,906 (27.5) | 26,973 (5.3) | 19.3 | Reference | Reference |
| Non-Hispanic Black | 11,682 (24.0) | 1,776 (3.7) | 15.4 | 0.70 (0.67-0.74) | 0.73 (0.69-0.77) |
| Hispanic | 3,332 (19.6) | 472 (2.8) | 14.3 | 0.63 (0.57-0.69) | 0.74 (0.66-0.82) |
| Asian | 5,176 (20.2) | 894 (3.5) | 17.3 | 0.77 (0.72-0.82) | 0.97 (0.90-1.05) |
| Age group, ya | |||||
| 65-74 | 92,174 (22.6) | 19,716 (4.8) | 21.2 | Reference | Reference |
| 75+ | 73,469 (33.1) | 11,606 (5.2) | 15.7 | 0.89 (0.87-0.91) | 0.75 (0.73-0.77) |
| Geographic regiona | |||||
| West Coast | 40,527 (22.8) | 7,032 (4.0) | 17.5 | 0.73 (0.70-0.75) | 0.88 (0.85-0.91) |
| Northeast | 60,654 (28.2) | 12,136 (5.7) | 20.2 | Reference | Reference |
| South | 26,640 (27.2) | 5,029 (5.1) | 18.8 | 0.86 (0.83-0.89) | 0.95 (0.91-0.99) |
| Other | 37,822 (27.1) | 7,125 (5.1) | 18.8 | 0.87 (0.85-0.90) | 0.95 (0.92-0.98) |
| COPDa | |||||
| Yes | 46,993 (46.5) | 9,693 (9.6) | 20.6 | 2.29 (2.23-2.36) | 1.05 (1.03-1.08) |
| No | 95,135 (25.2) | 16,813 (4.5) | 17.9 | Reference | Reference |
| Tobacco use disordera | |||||
| Yes | 12,121 (50.1) | 3,055 (12.6) | 43.6 | 2.08 (2.00-2.17) | 1.20 (1.15-1.25) |
| No | 127,306 (28.9) | 22,931 (5.2) | 10.4 | Reference | Reference |
HR = hazard ratio; PN = pulmonary nodule.
Assessed at start of follow-up for outcomes of chest CT scan and PN detection; assessed at first CT scan for outcome of PN detection among those with a chest CT scan.
Trends in PN detection generally mirrored trends in chest CT scan use. Modestly higher rates were seen for non-Hispanic White patients and Northeast residents, whereas substantially higher rates were seen among patients with, vs without, COPD (9.6% vs 4.5%) and with, vs without, tobacco use disorder (12.6% vs 5.2%). In the proportional hazards model, all the above factors showed significant hazard ratios (HRs), with HRs of 2.29 and 2.08 for COPD and tobacco use disorder, respectively (Table 2). HRs for undergoing chest CT scan imaging generally mirrored those for PN detection (e-Table 2).
Among patients with chest CT scans, 19.0% had at least one PN detected. Differences in PN detection rates across groups for this subcohort generally were smaller than for the entire cohort (Table 2). Men and women had similar rates. Non-Hispanic White patients showed higher rates than other groups, with the largest difference being between non-Hispanic White patients and non-Hispanic Black patients (19.3% vs 15.4%). Older patients showed slightly lower rates than younger patients (15.7% vs 21.2%; OR, 0.75). The effects of COPD and tobacco use disorder among those with chest CT scans were substantially muted compared with their effect among all patients, with ORs of 1.05 and 1.20, respectively. The proportion of chest CT scans that showed was similar during the ICD, Ninth and Tenth Revision, periods: 12.5% and 12.8%, respectively.
For the sensitivity analyses using the alternative definitions of PNs, the proportion of patients with PNs increased to 12.0% using the expanded codes definition and to 5.8% using the ± 30 days from a CT scan definition; corresponding percentages among those with chest CT scans were 45.0% and 22.1%, respectively. Rates for PNs using the expanded code definition by demographic and other factors are shown in e-Table 3.
We assessed lung cancer incidence in the 18,079 patients with PNs through 2017 residing in SEER regions. At initial PN detection, the median age was 75 years (interquartile range, 69-82 years) and 47% had COPD; 58% were women. Through a median follow-up of 1.5 years, 780 lung cancers were diagnosed; cumulative incidence rates at 6, 12, 18, and 24 months were 2.1% (95% CI, 1.9%-2.3%), 3.2% (95% CI, 2.9%-3.5%), 4.0% (95% CI, 3.7%-4.3%), and 4.7% (95% CI, 4.3%-5.0%) (e-Fig 1). In the proportional hazards model, tobacco use disorder (HR, 2.34) and COPD (HR, 1.74) were associated significantly with lung cancer incidence; women showed decreased risk (HR, 0.85) (Table 3). In the sensitivity analysis for PNs using the expanded codes, incidence rates were lower: 2.3% and 3.3% at 12 and 24 months, respectively.
Table 3.
Proportional Hazards Model of Lung Cancer Incidence after Initial PNs in the 5% Sample Cohort
| Factor | HR (95% CI) |
|---|---|
| Age group, y | |
| 65-74 | Reference |
| 75+ | 1.1 (0.95-1.3) |
| Sex | |
| Men | Reference |
| Women | 0.85 (0.73-0.98) |
| Race or ethnicity | |
| Non-Hispanic White | Reference |
| Non-Hispanic Black | 1.1 (0.8-1.5) |
| Hispanic | 0.58 (0.3-1.3) |
| Asian | 0.54 (0.3-0.99) |
| COPD | 1.7 (1.5-2.0) |
| Tobacco use disorder | 2.3 (2.0-2.8) |
| Region | |
| Northeast | Reference |
| West Coast | 0.77 (0.63-0.93) |
| Deep South | 0.93 (0.8-1.1) |
| Other SEER Program regions | 0.90 (0.7-1.1) |
HR = hazard ratio; PN = pulmonary nodule; SEER = Surveillance Epidemiology and End Results.
Lung Cancer Cohort
Table 1 shows demographics of the lung cancer cohort (n = 44,194); 15.7% were classified in the PN-only group, 2.9% in the LDCT scan group, and 81.4% in the reference group. Table 4 shows demographics, histologic findings, and stage by PN and LDCT scan status. Non-Hispanic White patients constituted a slightly higher proportion of the LDCT scan lung cancer group than the PN-only group (90.1% vs 88.8%), whereas for non-Hispanic Black patients, the reverse was true (5.4% vs 6.2%). The PN-only group included proportionally fewer patients with small cell lung cancer (7.6%) than either the LDCT scan (11.8%) or reference (13.8%) group and proportionally more adenocarcinoma (52.4%) than either the LDCT scan (45.2%) or reference (43.6%) group.
Table 4.
Demographics, Histologic Findings, and Stage Distribution by PN and LDCT Scan Status in the Lung Cancer Cohort
| Variable | PN-Only Group | LDCT Scan Group | Reference Group (No PN Detection or LDCT Scan) |
|---|---|---|---|
| All | 6,942 (100) | 1,271 (100) | 35,981 (100) |
| Women | 3,941 (56.8) | 650 (51.1) | 17,428 (48.4) |
| Men | 3,001 (43.2) | 621 (48.9) | 18,553 (51.6) |
| Race or ethnicity | |||
| Non-Hispanic White | 6,167 (88.8) | 1,145 (90.1) | 30,929 (86.0) |
| Non-Hispanic Black | 429 (6.2) | 69 (5.4) | 2,921 (8.1) |
| Hispanic | 42 (0.6) | a | 257 (0.7) |
| Asian | 120 (1.7) | a | 760 (2.1) |
| Histologic findings | |||
| Adenocarcinoma | 3,636 (52.4) | 575 (45.2) | 15,684 (43.6) |
| Squamous cell carcinoma | 1,463 (21.1) | 341 (26.8) | 8,472 (23.6) |
| Small cell lung cancer | 529 (7.6) | 150 (11.8) | 4,969 (13.8) |
| Other/not specified | 1,314 (18.9) | 205 (16.1) | 5,833 (19.0) |
| SEER Program stage | |||
| Localized | 3,864 (58.1) | 627 (50.3) | 8,369 (24.4) |
| Regional | 1,519 (22.8) | 382 (30.6) | 8,248 (24.0) |
| Distant | 1,271 (19.1) | 238 (19.1) | 17,734 (51.6) |
| Unknown | 358 | 24 | 1630 |
| TNM stage | |||
| I | 2,661 (58.5) | 429 (51.0) | 5,792 (23.0) |
| IA | 2,140 (47.0) | 342 (40.7) | 3,964 (15.8) |
| IB | 521 (11.5) | 87 (10.3) | 1,828 (7.3) |
| II | 399 (8.8) | 87 (10.0) | 2,023 (8.0) |
| III | 673 (14.8) | 179 (21.3) | 5,067 (20.1) |
| IV | 817 (18.0) | 146 (17.4) | 12,282 (48.8) |
| Unknown | 2,392 | 430 | 9,552 |
Data are presented as No. (%) or No. Percentages for stage exclude unknowns. LDCT = low-dose CT; PN = pulmonary nodule; SEER = Surveillance Epidemiology and End Results.
Too few in cell to specify.
Fifty-eight percent of the PN-only group and 50.3% of the LDCT group showed localized SEER stage disease. In contrast, only 24.4% of reference group showed localized disease. Among patients with localized disease, 30.0%, 4.9%, and 65.0% were in the PN-only group, LDCT scan group, and reference group, respectively, compared with 9.5%, 2.1%, and 88.4%, respectively, among patients with nonlocalized disease. Results generally were similar for TNM as for SEER stage (Table 4).
The results of the sensitivity analysis using the expanded codes definition of PNs are shown in e-Tables 4 and 5. The proportion in the PN-only group with localized SEER stage decreased slightly, from 58% to 52%. However, the proportion of all patients with localized disease who were in the PN-only group increased from 30% to 45%. Also shown are sensitivity analyses using 12- and 24-month windows before diagnosis instead of the 18-month primary analysis window; results were similar to those of the primary analysis.
Median follow-up for survival for patients with SEER stage localized disease and all patients was 1.9 and 1.2 years, respectively. Figure 1 shows lung cancer-specific and overall survival curves for the three groups. Three-year lung cancer-specific survival rates for all stages combined were similar for the LDCT scan and PN-only groups (75.6% vs 75.0%); both were substantially higher than the reference group (49.4%) (Table 5). For patients with localized disease, 3-year lung cancer-specific survival rates were 89.7%, 91.9%, and 82.4% for the PN-only group, LDCT scan group, and reference group, respectively; corresponding overall survival rates were 75.2%, 81.0%, and 63.2% (Table 5). The proportional hazards model for lung cancer-specific survival for patients with localized disease showed HRs of 2.40 (95% CI, 1.7-3.4) and 1.26 (95% CI, 0.9-1.8) for the reference and PN-only groups, respectively, compared with the LDCT scan group. Results of sensitivity analyses for survival are shown in e-Tables 4 and 5. Results generally were similar as for the primary analysis.
Figure 1.
A, B, Graphs showing lung cancer-specific (A) and overall (B) survival in patients with lung cancer diagnosed after LDCT scan screening (gray), PN only detection (blue), neither LDCT scan screening nor PN only detection (reference group; red). Vertical bars represent 95% CIs. LDCT = low-dose CT; PN = pulmonary nodule.
Table 5.
Three-Year Lung Cancer-Specific and Overall Survival Rates by PN and LDCT Scan Status in the Lung Cancer Cohort
| Survival Statistic and Stage | PN-Only Group | LDCT Scan Group | Reference Group |
|---|---|---|---|
| Lung cancer-specific survival | |||
| All | 75.0 (73.6-76.3) | 75.6 (70.5-80.0) | 49.4 (48.7-50.1) |
| SEER Program stage | |||
| Localized | 89.7 (88.2-90.9) | 91.9 (88.5-94.3) | 82.4 (81.3-83.5) |
| Regional | 70.1 (66.9-73.1) | 75.1 (64.7-82.8) | 58.7 (57.2-60.1) |
| Distant | 36.8 (33.0-40.7) | 33.9 (21.6-46.7) | 26.6 (25.6-27.7) |
| TNM stage | |||
| I | 84.9 (82.9-86.7) | 88.9 (84.1-92.4) | 79.2 (77.7-80.5) |
| II | 68.5 (61.7-74.3) | 72.2 (56.0-83.3) | 56.8 (53.9-59.6) |
| III | 44.6 (39.0-50.0) | 51.8 (38.3-63.8) | 36.7 (34.9-38.5) |
| IV | 19.5 (15.6-23.7) | 19.4 (8.3-34.1) | 12.6 (11.7-13.6) |
| Overall survival | |||
| All | 58.2 (56.7-59.7) | 59.7 (54.4-64.6) | 31.7 (31.1-32.3) |
| SEER stage | |||
| Localized | 75.2 (73.3-76.9) | 81.0 (76.0-85.1) | 64.6 (63.2-65.9) |
| Regional | 54.3 (51.1-57.3) | 54.8 (43.8-64.5) | 42.0 (40.7-43.3) |
| Distant | 20.5 (17.8-23.3) | 19.3 (11.6-28.4) | 13.4 (12.8-14.1) |
| TNM stage | |||
| I | 71.2 (69.5-74.0) | 80.0 (74.2-84.8) | 63.8 (62.1-65.3) |
| II | 56.2 (49.6-62.2) | 60.3 (39.1-76.3) | 45.2 (42.4-47.9) |
| III | 34.8 (29.9-39.7) | 39.1 (26.3-51.7) | 28.6 (27.0-30.2) |
| IV | 11.4 (8.6-14.7) | 16.0 (6.9-28.6) | 8.6 (7.9-9.3) |
Data are presented as rate (95% CI). LDCT = low-dose CT; PN = pulmonary nodule; SEER = Surveillance Epidemiology and End Results.
Discussion
In this study of Medicare enrollees, 26% underwent chest CT scans and 5% demonstrated PNs over a 5-year median coverage period. PN rates were similar across sexes, but were substantially lower for other racial and ethnic groups compared with non-Hispanic White patients. These lower rates in non-Hispanic White patients were the result of both lower chest CT scan rates and lower rates of PN detection among those with chest CT scans. The 1- and 2-year cumulative lung cancer incidence rates after initial PN detection were 3.2% and 4.7%, respectively, with COPD and tobacco use disorder associated with increased rates of subsequent lung cancer. A study of LDCT scan screening in this same Medicare population showed a slightly lower 1-year rate after initial LDCT scan screenings of 2.4%.12 Gould et al13 reported a rate of new nodules in those 65 to 79 years of age of about 15 per 1,000 person-years, slightly higher than our rate of initial nodules of 11.3 per 1,000 person-years. A study in two health-care systems of a slightly younger population showed a modestly lower 2-year cumulative incidence of lung cancer after PN detection of 3.7%.14
Among the lung cancer cohort, 16% had PNs detected within 18 months before diagnosis, of whom 58% had localized disease, slightly higher than the 50% localized disease rate for patients with prior LDCT scan screenings. Of all patients with localized disease, 30% had prior PNs, compared with only 4.9% with prior LDCT scan screenings, a 6:1 ratio. This was similar to the 5:1 PN to LDCT scan ratio for patients with stage I-II disease observed in the Detecting Early Lung Cancer study in the Mississippi Delta cohort.9 Lung cancer-specific survival, for all stages combined and for localized disease, was similar for those with prior PNs (PN group) as for those with prior LDCT scan screenings (LDCT scan group). The effectiveness of LDCT scan screening in reducing mortality resulting from lung cancer has been demonstrated through randomized trials.1,2 The similar stage distribution and survival observed here of the PN and LDCT scan groups suggest that the identification and reporting of PNs also may reduce lung cancer mortality; however, because this study was observational, it does not establish a causal association. Specifically, we do not know the natural history of the localized cancers diagnosed after PN detection in this study and what their stage and survival would have been absent PN identification.
The ICD, Tenth Revision (and Ninth Revision), diagnosis codes for PNs have not been validated and have unknown sensitivity and specificity. No specific code covers all PNs; R91.1 covers only solitary PNs. Nonsolitary PNs may be coded nonspecifically under R91.8. A recent study reported that 57% of patients with any PNs had multiple nodules.14 Therefore, our 5.0% rate likely underestimates the proportion of the population with PNs. In our sensitivity analysis, which included R91.8 abnormalities as PNs, the proportion with PNs increased to 12.0%. Additionally, the proportion of all patients with localized disease who had prior PNs increased from 30% to 45%, with the PN to LDCT scan ratio increasing to 9:1. Thus, our primary analysis results provide a lower bound on the prevalence of PNs among the general Medicare population and among the subset with lung cancer, and especially localized lung cancer.
Implementation of both PN and LDCT scan programs may be a pragmatic way to expand access to early lung cancer detection. Uptake of LDCT imaging increased from 3.3% of eligible people in 2016 to 5% in 2018, within the period covered in the present lung cancer cohort.15 The COVID-19 pandemic further slowed the already slow adoption of lung cancer screening.16 Notwithstanding the expanded United States Preventive Services Task Force and Centers for Medicare and Medicaid Services eligibility criteria, lowering the minimum smoking history from 30 pack-years to 20 pack-years, the striking ratio of early lung cancers detected by LDCT imaging vs after detection of a PN is unlikely to change substantially any time soon. Even with a threefold rise in LDCT imaging-detected lung cancers, other things being equal, approximately twice as many early-stage cases in the Medicare-aged population would be identified after reporting of PNs as through screening.
In the Detecting Early Lung Cancer study in the Mississippi cohort, > 50% of all patients with lung cancer (20% of patients with a diagnosis of stage I or II disease) in a program designed to ensure guideline-concordant management of PNs were ineligible for LDCT scan screening (even under the new guidelines) and would have had no pathway to early detection. Furthermore, a higher proportion of racial minorities and socioeconomically disadvantaged patients received a diagnosis of lung cancer through the PN program than through LDCT scan screening.9 The differences were especially striking in the pre-Medicare population 50 to 65 years of age, when lung cancer risk begins increasing and health insurance coverage is not guaranteed.17 Similarly, in this SEER-Medicare cohort, non-Hispanic Black patients were more likely to receive a diagnosis after detection of a PN than through LDCT scan screening.
Evidence-based management of PNs therefore may provide an epidemiologically powerful pathway to early lung cancer detection. Starting from lesion detection in pre-existing scans, guideline-concordant management of PNs may bypass the steepest implementation barriers to LDCT imaging, such as the need to develop new care-delivery infrastructure, the need to recruit eligible candidates, and the imperfect eligibility criteria, which by understating risk in certain demographic segments, exacerbate lung cancer disparities.18,19 Because guidelines for managing PNs are followed infrequently, we may have underestimated the potential impact of implementing guideline-concordant PN programs.20,21 Analysis of a database representing the commercially insured US population from 2013 through 2016 revealed that only 36% of patients with a PN received a guideline-concordant workup.22 Such is likely to be the case with Medicare enrollees. The yield of early-stage lung cancer may be even greater with the implementation of infrastructure to ensure guideline compliance in the management of patients with PNs.23,24
In addition to the limitations mentioned previously concerning the use of ICD codes to define PNs, our study has several other limitations. First, inherent limitations exist in using claims databases for research because they were developed primarily for reimbursement purposes. Second, the reasons for the chest CT scans on which PNs were observed are unknown. Third, we lacked other important clinical details, such as smoking history (beyond a diagnosis of tobacco use disorder), LDCT scan Lung Imaging Reporting and Data System scores, performance status of patients with lung cancer, and details of physician-patient interactions, including shared decision-making. Fourth, we did not have information on procedures and diagnoses before the start of follow-up in 2014, so the initial PNs as defined herein may not always have been the first PNs detected in a patient. Finally, the representativeness of this SEER-Medicare cohort to younger patients and those receiving care outside of SEER areas is uncertain.
Interpretation
In conclusion, the rate of lung cancer after reported PNs was relatively high. More than half of the patients who received a diagnosis of lung cancer after PN detection demonstrated localized disease, and almost one-third of all patients with localized had a PN detected previously. We recommend implementation of processes to promote guideline-concordant management of PNs as a means of expanding access to early lung cancer diagnosis.
Funding/Support
The authors have reported to CHEST that no funding was received for this study.
Financial/Nonfinancial Disclosures
None declared
Acknowledgments
Author contributions: All authors contributed substantially to this research. P. P. takes responsibility for (is the guarantor of) the content of the manuscript, including the data and analysis. P. P. and R. O. contributed to concept, study design, data analysis, drafting of the manuscript, and revising of the manuscript. E. M. and N. F. contributed to study design, data analysis, and revising of the manuscript.
Additional information: The e-Figure and e-Tables are available online under “Supplementary Data.”
Supplementary Data
References
- 1.Aberle D.R., Adams A.M., Berg C.D., et al. Reduced lung-cancer mortality with low-dose computed tomographic screening. N Engl J Med. 2011;365(5):395–409. doi: 10.1056/NEJMoa1102873. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.de Koning H.J., van der Aalst C.M., de Jong P.A., et al. Reduced lung-cancer mortality with volume CT screening in a randomized trial. N Engl J Med. 2020;382(6):503–513. doi: 10.1056/NEJMoa1911793. [DOI] [PubMed] [Google Scholar]
- 3.United States Preventive Services Task Force Screening for lung cancer. U.S. Preventive Services Task Force recommendation statement. JAMA. 2021;325:962–970. doi: 10.1001/jama.2021.1117. [DOI] [PubMed] [Google Scholar]
- 4.Richards T.B., Soman A., Thomas C., et al. Screening for lung cancer—10 States, 2017. MMWR. 2020;69:201–206. doi: 10.15585/mmwr.mm6908a1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Zgodic A., Zahnd W.E., Miller D.P., Studts J.L., Eberth J.M. Predictors of lung cancer screening utilization in a population-based survey. J Am Coll Radiol. 2020;17:1591–1601. doi: 10.1016/j.jacr.2020.06.015. [DOI] [PubMed] [Google Scholar]
- 6.Zgodic A., Zahnd W.E., Advani S., Eberth J.M. Low-dose CT lung cancer screening uptake: a rural–urban comparison. J Rural Health. 2022;38:40–53. doi: 10.1111/jrh.12568. [DOI] [PubMed] [Google Scholar]
- 7.Rankin N.M., McWilliams A., Marshall H.M. Lung cancer screening implementation: complexities and priorities. Respirology. 2020;25(suppl 2):5–23. doi: 10.1111/resp.13963. [DOI] [PubMed] [Google Scholar]
- 8.Zeliadt S.B., Hoffman R.M., Birkby G., et al. Challenges implementing lung cancer screening in federally qualified health centers. Am J Prev Med. 2018;54:568–575. doi: 10.1016/j.amepre.2018.01.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Osarogiagbon R., Liao W., Faris N.R., et al. Lung cancer diagnosed through screening, lung nodule and neither program: a prospective observational study of the Detecting Early Lung Cancer (DELUGE) in the Mississippi Delta cohort. J Clin Oncol. 2022;40(19):2094–2105. doi: 10.1200/JCO.21.02496. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.National Cancer Institute SEER-Medicare linked database. National Cancer Institute website. https://healthcaredelivery.cancer.gov/seermedicare/
- 11.Danforth K.N., Early M.I., Ngan S., Kosco A.E., Zheng C., Gould M.K. Automated identification of patients with pulmonary nodules in an integrated health system using administrative health plan data, radiology reports, and natural language processing. J Thorac Oncol. 2012;7(8):1257–1262. doi: 10.1097/JTO.0b013e31825bd9f5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Pinsky P.F., Miller E. Use and outcomes of low-dose CT lung cancer screening in the Medicare population. Chest. 2022;162(3):721–729. doi: 10.1016/j.chest.2022.03.031. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Gould M.K., Tang T., Liu I.L., et al. Recent trends in the identification of incidental pulmonary nodules. Am J Respir Crit Care Med. 2015;192(10):1208–1214. doi: 10.1164/rccm.201505-0990OC. [DOI] [PubMed] [Google Scholar]
- 14.Farjah F., Monsell S.E., Greenlee R.T., et al. Patient and nodule characteristics associated with a lung cancer diagnosis among individuals with incidentally detected lung nodules. Chest. 2023;163(3):719–730. doi: 10.1016/j.chest.2022.09.030. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Fedewa S.A., Kazerooni E., Studts J.L., et al. State variation in low-dose computed tomography scanning for lung cancer screening in the United States. J Natl Canc Inst. 2021;113:1044–1052. doi: 10.1093/jnci/djaa170. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Fedewa S., Smith R.A., Silvestre G.A., Jemal A. Lung cancer screening rates during the Covid-19 pandemic. Chest. 2022;161(2):586–589. doi: 10.1016/j.chest.2021.07.030. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Liao W., Faris N.R., Fehnel C., et al. Lung cancer risk in persons in low-dose CT screening (LDCT) versus incidental lung nodule programs (ILNP). Abstract presented at. American Society of Clinical Oncology Annual Meeting, June 3-7. 2022 [Google Scholar]
- 18.Pinsky P.F., Lau Y.K., Doubeni C. Potential disparities by sex and race or ethnicity in lung cancer screening eligibility rates. Chest. 2021;160(1):341–350. doi: 10.1016/j.chest.2021.01.070. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Haiman C.A., Stram D.O., Wilkens L.R., et al. Ethnic and racial differences in the smoking-related risk of lung cancer. N Engl J Med. 2006;354(4):333–342. doi: 10.1056/NEJMoa033250. [DOI] [PubMed] [Google Scholar]
- 20.Tanner N.T., Aggarwal J., Gould M.K., et al. Management of pulmonary nodules by community pulmonologists: a multicenter observational study. Chest. 2015;148(6):1405–1414. doi: 10.1378/chest.15-0630. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Wiener R.S., Gould M.K., Slatore C.G., Fincke B.G., Schwartz L.M., 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(6):871–880. doi: 10.1001/jamainternmed.2014.561. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Pyenson B.S., Bazell C.M., Bellanich M.J., Caplen M.A., Zulueta J.J. No apparent workup for most new indeterminate pulmonary nodules in US commercially-insured patients. J Health Econ Outcomes Res. 2019;6(3):118–129. doi: 10.36469/9674. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Dyer D.S., Zelarney P.T., Carr L.L., Kern E.O. Improvement in follow-up imaging with a patient tracking system and computerized registry for lung nodule management. J Am Coll Radiol. 2021;18(7):937–946. doi: 10.1016/j.jacr.2021.01.018. [DOI] [PubMed] [Google Scholar]
- 24.Carr L.L., Dyer D.S., Zelarney P.T., Kern E.O. Improvement in stage of lung cancer diagnosis with incidental pulmonary nodules followed with a patient tracking system and computerized registry. JTO Clin Res Rep. 2022;3(3) doi: 10.1016/j.jtocrr.2022.100297. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.