Incidence and Mortality Trends after Introduction of Lung Cancer Screening in a Community-Based Healthcare System

Abstract

Background:

Lung cancer screening (LCS) enables early cancer detection but raises concerns about overdiagnosis, particularly in community-based LCS settings where evidence remains limited. This study examined trends in lung cancer incidence and mortality following the introduction of LCS in a community-based setting and used these trends to assess the potential of LCS-associated overdiagnosis.

Methods:

This retrospective study leveraged data and outcomes from the Kaiser Permanente Colorado (KPCO) LCS program between 2014 and 2023. Trends in lung cancer incidence and mortality were assessed by annual percent change (APC). Excess incidence was estimated by comparing post-LCS early-stage lung cancer incidence with pre-LCS program implementation and predicted incidence and expressed as counts and percentage of lung cancers.

Results:

Among 1,006 incident lung cancers, 619 (62%) were early stage, 367 (37%) were late stage, and 20 (2%) had unknown stage; 198 lung cancer deaths were identified. Early-stage incidence increased (APC, 4.9%; 95% confidence interval (CI), −1.6% to 1.8%), late-stage incidence decreased (APC, −5.2%; 95% CI, −9.8% to −0.3%) and mortality decreased (APC, −6.0%, 95% CI, −12.3 to 0.8). Excess incidence suggests that 11.3%-35.9% of early-stage lung cancer cases could represent potential overdiagnosis, with only a portion of these reflecting true overdiagnosis.

Conclusion:

We observed patterns consistent with a maturing LCS screening program including a non-significant increase in early-stage detection, significant reduction in late-stage disease, and a non-significant decline in lung cancer-specific mortality. These findings provide an important starting point for understanding the impact of LCS on the distribution of stage in community-based screening programs. Continued monitoring of lung cancer incidence and mortality remains essential to refine estimates and evaluate their implications for healthcare resources, survivorship care, and the net benefit of screening.

1. Introduction

The goal of lung cancer screening (LCS) programs is to detect cancer at an early and treatable stage to reduce morbidity and mortality by targeting high-risk individuals.(1, 2) However, these benefits must be balanced against potential harms, including unnecessary radiation, false positive related procedures, and overdiagnosis.(3, 4) Overdiagnosis is an unintended harm associated with screening programs that leads to serious consequences, such as unnecessary treatment and may result in increased patient anxiety, serious physical harm, unnecessary losses in quality of life, and higher healthcare costs.(1, 2, 5, 6, 7, 8, 9, 10, 11, 12, 13)

Overdiagnosis occurs when screening detects a tumor that would not have become clinically evident in the absence of screening - either because the individual dies from other causes (competing cause of death) or the disease progresses too slowly to produce symptoms during a person’s lifetime (indolent tumors).(5, 6, 7, 14, 15, 16) Lung cancer biological behavior varies widely, with some tumors progressing rapidly and others remaining slow-growing and clinically insignificant.(17) In particular, within the LCS setting, high resolution low-dose computed tomography (LDCT) may increase detection of indolent or slow-growing pathologies, leading to identification of both biologically important and biologically indolent tumors.(6, 8, 9, 10, 18, 19, 20, 21)

Estimates of overdiagnosis varied widely in LCS randomized clinical trials (RCTs) ranging from 0% to 67%.(15, 22, 23, 24, 25, 26, 27) However, little is known about overdiagnosis in community-based LCS settings where estimates may differ because individuals participating in RCTs may not reflect community-based populations. Individuals undergoing LCS in routine practice are more likely to be older, currently smoke, and have a higher comorbidity burden - particularly smoking-related cardiorespiratory disease – which increases the competing risk of death.(28, 29, 30, 31, 32) Overdiagnosis is difficult to measure at the individual level because indolent tumors cannot be reliably distinguished from clinically significant tumors.(8, 33) Consequently, overdiagnosis is most reliably estimated in high-quality randomized trials with long follow-up periods. However, studies using real-world secondary data sources can provide valuable complementary evidence by examining secular trends in cancer incidence after the introduction of screening that may help characterize patterns consistent with overdiagnosis at the population level.(8, 10, 34) Using secular incidence trends to examine changes in mortality and stage distribution provides an important first step toward understanding the impact of LCS on the distribution of stage and informing shared decision-making, healthcare resource allocation, survivorship care, and evaluations of the overall net benefit of screening.(7, 11, 34, 35, 36, 37)

The objective of this study was to evaluate changes in overall and stage-specific lung cancer incidence and mortality following the introduction of LCS in community-based LCS and to use these trends to inform initial estimates of potential overdiagnosis associated with LCS.

2. Methods

2.1. Study Design and Setting

This retrospective cohort study was conducted at Kaiser Permanente Colorado (KPCO), a non-profit integrated healthcare system serving more than 500,000 individuals. This study adhered to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE)(38) reporting guidelines for cohort studies. The study was approved by the Interregional Institutional Review Board of Kaiser Permanente. Individual consent for this retrospective analysis was waived.

2.2. KPCO LCS Program

The implementation and outcomes of the KPCO LCS program have been described previously.(39, 40) Briefly, KPCO initiated LCS in May 2014 using a centralized model that incorporated two dedicated patient navigators, EHR-based provider prompts, patient follow-up reminders, and automated orders for annual repeat screening. LCS registries were created and over time, additional EHR-based support tools were created to provide eligibility guidance, billing codes, shared decision-making documentation, and smoking-cessation tools. The program gradually transitioned to a less centralized structure where responsibilities shifted from LCS-specific navigators to care teams managing broader population health efforts.

2.3. Data Sources

Data were derived from the KPCO Virtual Data Warehouse and the Population-based Research to Optimize the Screening Process (PROSPR) Lung Common Data Model which includes EHR-derived patient level demographic, smoking history, LCS screening outcomes, diagnostic, procedure, cancer registry, and death data.(41, 42, 43) Cancer registry data were derived from manually abstracted EHR data consistent with North American Association of Central Cancer Registries standards.(44) Death and cause of death data were derived from cancer registry data, health plan membership data, and state level databases. Baseline LDCT-LCS screening data including the capture of Lung CT Screening Reporting and Data System (Lung-RADs)(45) were obtained from radiology reports and identified using procedure codes G0297, 71250, and 71271.

2.4. Study Cohort

All individuals enrolled within KPCO who met the US Preventive Services Task Force (USPSTF) eligibility criteria as of December each year from 2005 through 2024 were identified via EHR captured self-reported smoking status and history and were included in the person-years-at-risk denominator calculation. Specifically, patients were considered eligible if they currently smoked or had quit within 15 years of December 31^st of each year and fell within the required age and smoking criteria: 55-80 years and ≥ 30 pack-years (before 3/1/2021) or 50-80 years and ≥ 20 pack-years (on or after 3/1/2021). This population was then divided into two groups: an incidence group of individuals identified between 2005 and 2013 (prior to implementation of LCS at KPCO) which was used to establish pre-LCS lung cancer incidence trends, and a post-LCS group of individuals identified between January 2014 and December 2023. Lung cancers diagnosed on or after baseline LDCT-LCS were considered screen-detected.

2.5. Statistical Analysis

Overall lung cancer incidence and mortality were calculated using the person-years-at-risk for the denominator and the number of lung cancer cases for incidence and number of deaths with lung cancer as the underlying cause of death for mortality. Incidence was stratified into early-stage (defined as American Joint Committee on Cancer [AJCC] stages 0 – IIIA), late-stage (defined as AJCC stages IIIB-IV), and unknown stages. Secular trends were evaluated using Joinpoint regression to estimate annual percent change (APC) and 95% confidence intervals (CIs).(46) These analyses evaluated three key features of effective screening: (1) an increase in early-stage incidence reflecting earlier cancer detection, (2) a decrease in late-stage incidence reflecting fewer advanced presentations, and (3) reduced cancer-specific mortality.(8, 21, 47, 48)

Consistent with the conceptual framework that overdiagnosis is reflected in the excess detection of cancers that would not have become clinically apparent during a patient’s lifetime, overdiagnosis was estimated using the excess early-stage incidence and excess early-stage count approaches proposed by Gao et al.(47, 48, 49, 50) Under this framework, overdiagnosis is expected to be occur primarily among early-stage cancers, whereas late-stage cancers are unlikely to represent overdiagnosed cases.(50) Estimates were calculated relative to two separate comparators: (1) 2013 incidence (the year preceding LCS implementation) and (2) predicted incidence derived from pre-LCS incidence (2005-2013). Expected incidence was extrapolated into the post-LCS period using Poisson regression, assuming linearity on a log-log scale and stable underlying lung cancer incidence. Overdiagnosis was expressed as annual excess cases defined by differences between observed and expected incidence and further summarized as a proportion of total lung cancer diagnoses and as incidence rate ratios (IRRs) with 95% CIs. Estimates were interpreted as upper-bound estimates, as it is unknown what proportion of excess incidence cancers reflect true overdiagnosis. Negative excess counts (i.e. observed counts that were less than expected counts) were set to 0. Consistent with prior methods, analyses assumed no overdiagnosis during 2005 through 2013, no overdiagnosis among late-stage cancers, and stable underlying lung cancer risk.(48)

Given the low uptake of LCS we performed a sensitivity analysis restricting cancer counts to screen-detected cancers only. Incidence was stratified by stage and incidence trends were analyzed using APC and 95% CI.

Analyses were performed using SAS^® Software version 9.4M6 (SAS Institute Inc., Cary, North Carolina), and Joinpoint Regression Program version 4.8.0.1.(46)

3. Results

A total of 1,006 patients with incident lung cancer and 198 deaths with an underlying cause of lung cancer were identified between 2014 and 2023. Overall incidence increased slightly (APC, 1.01%; 95% CI, −2.9% to 5.1%) and overall mortality decreased (APC, −6.0; 95% CI, −12.3% to 0.8%) between 2014 and 2023, but neither trend was significant (Figure 1). Among lung cancer cases, 619 (62%) were early stage, 367 (37%) were late stage, and 20 (2%) had unknown stage at diagnosis. There was more than a two-fold increase of early-stage disease from 309 cases per 100,000 in 2014 to 741 cases per 100,000 in 2023 (absolute difference (D), 432 cases per 100,000; 95% CI, 235 to 629 cases per 100,000; APC, 4.9; 95% CI, −1.6 to 11.8) (Figure 2). There was a concomitant decrease in late-stage disease from 386 cases per 100,000 in 2014 to 208 cases per 100,000 in 2023 (D, −178 cases per 100,000; 95% CI, −332 to −25 cases per 100,000; APC, −5.2; 95% CI, −9.8 to −0.3). The incidence of unknown-stage cancer did not change substantially, with an absolute difference of −30 cases per 100,000 (95% CI, −98 to 37 cases per 100,000).

Figure 1. Overall lung cancer incidence and mortality among KPCO LCS-eligible individuals, 2014-2023.

Figure 2. Lung cancer incidence by stage among KPCO LCS-eligible individuals, 2013-2023.

The estimated excess early-stage incidence and counts relative to 2013 are presented in Table 1. Using the excess incidence approach, 85.8 cases (13.9%) were potentially overdiagnosed. Using the excess count approach, 222 cases (35.9%) were potentially overdiagnosed. Using 2013 as the comparator, IRRs and 95% CIs ranged from a low of 0.57 (95% CI, 0.35 to 0.92) in 2014 to a high of 1.46 (95% CI, 1.01 to 2.13) in 2018 (Figure 3A).

Table 1.

Excess early-stage incidence and estimates of potential overdiagnosis using pre-LCS comparator

				Excess early-stage incidence approach			Excess early-stage count approach
Year	Population [a]	Cases [b]	Incidence [c]	Excess Incidence [d = c – 545.2]	Count Overdiagnosed [e = a * d / 100,000]	% Overdiagnosed [f = e / b]	Excess Count [g = b – 41]	% Overdiagnosis [h = g / b]	% of cases that were screen-detected	% of Population with LCS- LDCT b	% of Population with non-LCS Chest CT c
2013 a	8616	41	545.2	--	--	--	--	--	--	--	--
2014	9566	28	308.8	−236.4	0 d	0 d	0 d	0 d	10.7	5.8	19.1
2015	10265	57	580.2	35.0	3.6	6.3	16	28.1	31.6	13.5	27.0
2016	10900	58	554.7	9.5	1.0	1.7	17	29.3	41.4	21.2	21.5
2017	10971	66	624.8	79.6	8.7	13.2	25	37.9	53.0	28.4	22.4
2018	10998	84	798.4	253.2	27.8	33.1	43	51.2	41.7	33.5	22.8
2019	10682	43	416.8	−128.4	0d	0 d	2	4.7	76.7	37.4	22.9
2020	9959	63	659.6	114.4	11.4	18.1	22	34.9	57.1	32.6	23.5
2021	12477	69	576.0	30.8	3.8	5.5	28	40.6	47.8	28.1	20.5
2022	11949	69	601.6	56.4	6.7	9.7	28	40.6	58.0	29.4	23.6
2023	11633	82	741.2	196.0	22.8	27.8	41	50.0	65.9	40.4	25.7
TOTAL	--	619	--	--	85.8	13.9	222	35.9	50.0	--	--

^aComparator year (pre-LCS)

^bLCS-LDCT identified using codes G0297, 71250, 71271

^cNon-LCS Chest CTs identified using codes 87.41, 87.42, 71260, 71270, 71275, 75571, 75572, 75573, 75574, 76380, B24ZZZ, BB240ZZ, BB2400Z, BB241ZZ, BB2410Z, BB24YZZ, BB24Y0Z, BB29ZZZ, BB290ZZ, BB2900Z, BB291ZZ, BB2910Z, BB29YZZ, BB29Y0Z, BB28ZZZ, BB280ZZ, BB2800Z, BB281ZZ, BB2810Z, BB28YZZ, BB28Y0Z, BB27ZZZ, BB270ZZ, BB2700Z, BB271ZZ, BB2710Z, BB27YZZ, BB27Y0Z, BP2W0ZZ, BP2W1ZZ, BP2WYZZ, BW24ZZZ, BW240ZZ, BW2400Z, BW241ZZ, BW2410Z, BW24YZZ, BW24Y0Z, BW25ZZZ, BW250ZZ, BW2500Z, BW251ZZ, BW2510Z, BW25YZZ, BW25Y0Z

^dNegative counts and percents set to 0

Figure 3. Incidence rate ratios of lung cancer incidence.

* Incidence rate ratios of lung cancer incidence using 2013 (Pre-LCS implementation) as a comparator (A) and using predicted incidence as a comparator (B)

Figure 4 shows early-stage pre-LCS and post-LCS incidence trends and predicted incidence for 2014-2023. The estimated excess early-stage incidence and counts relative to the predicted incidence are presented in Table 2. A total of 93.3 cases (15.1%) were potentially overdiagnosed using the excess incidence approach while 69.7 cases (11.3%) were potentially overdiagnosed using the excess count approach. Using predicted incidence, IRRs ranged between a low of 0.64 (95% CI, 0.40 to 1.03) in 2014 to the highest IRR of 1.52 (95% CI, 1.09 to 2.13) in 2018 (Figure 3B).

Figure 4. Early-stage lung cancer incidence before and after KPCO LCS implementation, 2005-2023*.

Table 2.

Excess early-stage incidence and estimates of potential overdiagnosis using predicted comparator

						Excess early-stage incidence approach			Excess early-stage count approach
Year	Population [a]	ActualCases [b]	Actual Incidence [c]	Predicted Cases [d]	Predicted Incidence [e]	Incidence Excess [f = c – d]	Count Overdiagnosed [g = f * a / 100,000]	% Overdiagnosed [h = g / b]	Excess Count [i = b – d]	% Overdiagnosed [j = i / b]	% of cases that were screen-detected
2014	9566	28	308.8	46.1	481.4	−172.6	0 a	0 a	0 b	0 b	10.7
2015	10265	57	580.2	50.5	491.6	88.6	9.1	16.0	6.5	11.5	31.6
2016	10900	58	554.7	54.7	502.1	52.6	5.7	9.9	3.3	5.6	41.4
2017	10971	66	624.8	56.3	512.7	112.0	12.3	18.6	9.7	14.8	53.0
2018	10998	84	798.4	57.6	523.6	274.8	30.2	36.0	26.4	31.4	41.7
2019	10682	43	416.8	57.1	534.7	−118.0	0a	0 a	0 a	0 a	76.7
2020	9959	63	659.6	54.4	546.1	113.5	11.3	17.9	8.6	13.7	57.1
2021	12477	69	576.0	69.6	557.7	18.3	2.3	3.3	0 a	0 a	47.8
2022	11949	69	601.6	68.1	569.5	32.1	3.8	5.6	0.9	1.3	58.0
2023	11633	82	741.2	67.7	581.6	159.6	18.6	22.6	14.3	17.4	65.9
TOTAL	--	619	--	--	--	--	93.3	15.1	69.7	11.3	50.0

^aNegative counts set to 0

A total of 383 screened patients were diagnosed with lung cancer between 2014 and 2023; 311 (81%) early stage and 75 (19%) late stage. Early-stage disease incidence increased from 33 cases per 100,000 in 2014 to 488 cases per 100,000 in 2023 (absolute difference (D), 455 cases per 100,000; 95% CI, 320 to 591 cases per 100,000; APC, 19.9; 95% CI, 4.3 to 37.9) (Figure 5). Late-stage disease increased, but not significantly from 33 cases per 100,000 in 2014 to 45 cases per 100,000 in 2023 (D, 12 cases per 100,000; 95% CI, −42 to 67 cases per 100,000; APC, 6.2; 95% CI, −3.4 to 16.9).

Figure 5. Screen-detected lung cancer incidence by stage among KPCO LCS-eligible individuals, 2013-2023.

4. Discussion

To our knowledge, this is the first study to characterize population-level trends in lung cancer incidence and mortality following the introduction of LCS within a large, integrated community-based healthcare system and to employ techniques specifically to evaluate whether these patterns are consistent with plausible estimates of overdiagnosis. We found a modest, non-significant increase in overall incidence of lung cancer and a non-significant decrease in lung cancer mortality. Incidence of early-stage cancers increased 4.9% per year while late-stage incidence decreased 5.2% per year. Together, these patterns are indicative of a maturing LCS program and suggest a limited impact of overdiagnosis. Using multiple estimation approaches, we estimate between 69 (11.3%) and 222 (35.9%) of the 619 early-stage lung cancer cases may represent plausible upper-bound estimates of the potential magnitude of overdiagnosis, with a portion of these excess cases likely reflecting true overdiagnosis. Because overdiagnosis cannot be directly observed and empirical evidence remains limited, our findings should be considered an initial contribution to an evolving discussion regarding the potential of overdiagnosis associated with LCS.

Our real-world analyses show the impact of changes in KPCO program implementation and programmatic changes on lung cancer incidence. At the beginning of KPCO program implementation, late-stage incidence exceeded early-stage incidence, reflecting detection of screening-eligible individuals who initially presented with late-stage disease. As screening uptake increased, early-stage incidence quickly exceeded late-stage incidence. A second decline in early-stage incidence was observed in 2019, coinciding with an IT malfunction that led to a 10-fold increase in LCS orders among eligible patients and a six-fold rise in completed baseline screenings.(39) This surge may have overwhelmed scheduling and imaging capacity. Additionally, the healthcare system may have needed time to adapt to the increased downstream demand for diagnostic evaluation and treatment.(39) Although screening rates declined due to COVID-19 shutdowns, any positive screens (i.e., when an LDCT scan detects an abnormality that requires further evaluation) were prioritized for follow-up diagnostic scans which is reflected in the increased incidence in 2020.

Population-level estimates of secular changes in early-stage incidence may change as uptake increases, given that the current uptake at KPCO is approximately 37% and may differ if all eligible individuals were screened.(39) This underscores the importance of ongoing evaluation of LCS programs to ensure feasibility and effectiveness in a community setting. Moreover, if alternative and less restrictive smoking duration LCS eligibility criteria are adopted in the future,(51) on-going evaluations will be needed to ensure that possible overdiagnosis of indolent cancers is not exacerbated.

4.1. Comparisons to other studies

Research aimed at quantifying LCS-associated overdiagnosis is still in its early stages, with relatively few population-based studies available. Existing studies outside of clinical trials largely examined population-wide LCS implemented among average-risk individuals in China, Korea, and Taiwan,(47, 48, 49, 52) whereas our LCS program targeted those at highest risk. Consequently, comparable community-based overdiagnosis estimates are limited. However, previous research within the PROSPR-Lung Consortium found no difference in recurrence rates between screen-detected and non-screen detected lung cancers, providing complementary evidence to our observed incidence and mortality trends that the population-level impact of overdiagnosis is likely limited.(53)

In RCTs of LDCT-based LCS, the overall estimate of overdiagnosis was 29%, ranging from 12.9% to 67.2%.(54) However, estimates of overdiagnosis from RCTs may not be comparable to our estimates for multiple reasons. First, both arms in most RCTs were screened either with LDCT or chest X-ray. Therefore, the chest X-ray arms do not provide an unbiased baseline incidence of lung cancer in the absence of screening.(1) Second, the follow-up duration of RCT participants may not have been enough to account for the effects of lead time.(1) Third, participants in most trials are different from individuals participating in community-based LCS.(1, 40) As a result, the average risk of lung cancer and mortality may differ.(55, 56)

Estimates from microsimulation modeling suggested that < 10% of lung cancers detected by CT were overdiagnosed in the NLST.(1, 57) Furthermore, an estimated 9.9% of all lung cancers detected by screening are estimated to be overdiagnosed cancers under the screening policy as recommended by the USPSTF.(58) Estimates by microsimulation modeling were less than what we observed in our real-world estimates, which may reflect differences in the comorbidity profile of our study population, screening adherence patterns, and clinical management practices.

Following widespread implementation of screening mammography in the US in the 1980s, overdiagnosis estimates varied substantially, with historical projections suggesting it accounted for approximately 31% of all newly diagnosed breast cancers during the height of screening.(59) Current estimates are estimated to be approximately 15%.(60) A similar pattern has been observed in LCS where overdiagnosis estimates have varied widely across RCTs and tend to decline with longer follow-up and improved methodological approaches. This highlights the need for ongoing surveillance of lung cancer incidence trends in community-based healthcare settings to establish a foundation for future investigations.

4.2. Strengths and Limitations

Our study has several strengths. First, our rich data source includes demographic, clinical, screening, and cancer data for patients who receive both primary care and oncology care services within the KPCO healthcare system. This allowed us to identify LCS eligibility on a population level as well as all associated cancer outcomes. Our findings are likely to be generalizable to other LCS programs deployed in large integrated healthcare systems. Our use of a simple and transparent analytic approach supports interpretability, reproducibility, and application in other real-world settings. Lastly, we used multiple established methods to generate a range of plausible estimates regarding the potential magnitude of overdiagnosis, providing initial evidence from a community-based lung cancer screening setting.

Our study was not without limitations. Adequate follow-up time is needed to separate true overdiagnosis from the effects of lead-time bias.(1, 24) Previous studies have estimated sufficient follow-up time to avoid lead time bias at 3.6 years.(2) The median follow-up time among those participating in LCS at KPCO was 3.7 years.(39) Assuming no overdiagnosis occurred prior to KPCO implementing its LCS program may have resulted in underestimation of our estimates, as overdiagnosis likely occurred in 2013 and earlier years.(48) We extrapolated pre-screening incidence to project expected trends. However, this approach relies on assumptions that are not consistently met. Specifically, it assumes that cancer incidence would have remained stable in the absence of screening and that case ascertainment and diagnostic practices were unchanged over time, both of which are unlikely.(50) We are unable to disentangle how competing conditions along with the COVID-19 healthcare system shutdowns affected receipt of LCS. The underlying cause of death may be subject to misclassification as its determination depends on clinician interpretation in the context of multiple coexisting conditions, including cancer. Lastly, trends in mortality over time cannot be fully attributed to LCS due to the introduction of novel immunotherapies and targeted therapies that have improved patient survival.(61, 62)

5. Conclusions

We observed patterns consistent with a maturing screening program, including a non-significant increase in early-stage detection, a significant reduction in late-stage disease and non-significant decline in lung cancer-specific mortality. Together these incidence and mortality trends suggest a limited impact of overdiagnosis. Upper-bound estimates suggested that 11.3%-35.9% of early-stage lung cancer cases could be attributable to overdiagnosis, however many of these excess cases likely represent cancers that would have become clinically apparent over time, with only a subset reflecting true overdiagnosis. These findings provide an important starting point for understanding the potential magnitude of overdiagnosis in community-based LCS programs. As LCS programs mature, accrue longer follow-up and achieve greater uptake, continued evaluation of lung cancer incidence and mortality trends will be essential for refining estimates and informing implications for healthcare resources, survivorship care, and the overall value of screening.

Highlights.

• There are concerns about overdiagnosis of cancer from lung cancer screening

• Not very much is known about overdiagnosis outside of clinical trials

• Trends in incidence and mortality can inform potential overdiagnosis

• We found patterns of a maturing LCS program with limited impact of overdiagnosis

Data Availability Statement:

All data and analytical code generated or analyzed during this study are openly available on the Open Science Framework at osf.io/fvqns.