Barriers to Timely Lung Cancer Care in Early Stage Non—Small Cell Lung Cancer and Impact on Patient Outcomes

Abstract

Optimal time to treatment of early-stage lung cancer is uncertain. In our study of 204 Veterans who presented with radiographic stage 1 or 2 NSCLC, only 33% of patients received treatment within 14 weeks, which was not associated with improved overall survival or decreased rates of upstaging. Post hoc, 8 weeks was associated with less upstaging. We identified modifiable patient-related and system-related delays in care.

Background:

Optimal time to treatment for early-stage lung cancer is uncertain. We examined causes of delays in care for Veterans who presented with early-stage non–small cell lung cancer (NSCLC) and whether workup time was associated with increased upstaging or all-cause mortality.

Methods:

We performed a retrospective analysis of Veterans referred to our facility with radiographic stage I or II NSCLC between January 2013 to December 2017, with follow-up through October 2021. Patient demographics, tumor characteristics, time intervals of care, and reasons for delays were collected. Guideline concordance (GC) was defined as treatment within 14 weeks of abnormal image. Multivariable analyses were performed to determine association between delays in care, survival, and upstaging.

Results:

Data from 203 Veterans were analyzed. Median time between abnormal imaging to treatment was 17.7 weeks (IQR 12.7-26.6). Only 33% of Veterans received GC care. Most common patient-related delays were: intercurrent hospitalization/comorbidity (23%), no-shows (16%) and inability to reach Veteran (17%). Most common system-related delay: lack of scheduling availability (25%). Delays associated with upstaging: transportation issues, request for coordination of appointments, and unforeseen appointment changes. Rates of upstaging did not differ between GC and discordant groups (P = .6). GC care was not an independent predictor of mortality. Post-hoc, treatment within 8 weeks was associated with lower rates of upstaging (P = .05).

Conclusion:

Although GC care did not impact survival or upstaging for early-stage NSCLC, shorter timeframes may be beneficial. Modifiable delays in care exist which may be addressed at an institutional level to improve timeliness of care.

Introduction

Lung cancer is the leading cause of cancer death in the United States, impacting nearly 1 in 5 United States Veterans.¹ If detected early, it is a potentially curable disease with 65% 5-year survival. Randomized controlled trials^2–4 and microsimulation modeling studies⁵ over the last decade have demonstrated a mortality benefit with lung cancer screening (LCS), leading to expansion of eligibility criteria for low-dose chest CT (LDCT) screening to include individuals between the ages of 50 to 80 years with a minimum of 20 pack year smoking history.⁶ This expansion is intended to improve racial and gender equity of LCS, increasing total US population eligibility from 14% to at least 20%.⁵ Healthcare systems have already noted an increase in incidence of detection of stage I non–small cell lung cancer (NSCLC).⁷ As LCS increases, evidence-based processes are needed to ensure timely work-up and treatment of suspected early-stage lung cancer.

Once a decision is made to pursue invasive testing, what constitutes “timely” care in terms of optimizing outcomes remains unclear, as current recommendations are based on expert opinion only. Heterogeneity in study population and variability in methods across existing studies constrain the ability to perform pooled quantitative analysis, resulting in uncertainty about the optimal time between detection and treatment that would most improve outcomes. Shorter time intervals between detection and treatment have paradoxically been associated with worse outcomes in some reports (i.e., the waiting time paradox),⁸ but this observation may be confounded by the propensity of having patients with the most advanced stages of lung cancer undergo expedited workups. However, there is a relative lack of data regarding early stage (Stage I and II) lung cancer and impact of time on outcomes. Systematic literature reviews have reported 2 common themes - 1) the impact of timeliness on outcomes is likely stage-specific,⁹ suggesting that future guideline development be stratified by radiographic stage at presentation¹⁰ and 2) despite the absence of high-quality evidence, facilities should identify local barriers to timely care.

We therefore conducted a retrospective analysis of the timeliness of care for Veterans referred to our facility for radiographic stage I and II NSCLC, with the objective of investigating whether increased duration of time to treatment was associated with adverse outcomes, including mortality and/or upstaging. We also identified the types of delays that patients experienced during their workup. We defined timely care based upon the RAND guidelines¹¹ (ie, total time of 14 weeks from abnormal image to treatment initiation) which is the most frequently used definition of timely care in recently published literature,¹² based on expert opinion. We also assessed the timeliness of sub-intervals of care, delineating key time points in the care continuum - time from abnormal image to pulmonary referral order, to consultation visit, to date of definitive diagnosis, to the date of treatment initiation. We hypothesized that guideline-concordant (GC) care would be associated with improved survival or decreased rates of upstaging between initial radiographic stage and final pathologic stage. Finally, we investigated whether patient delays and/or system delays contributed to poor outcomes (increased upstaging and decreased survival).

Study Design and Methods

We performed a retrospective electronic medical record review of United States military Veterans presenting with radiographic stage I or II NSCLC between January 1st, 2013, and December 31st, 2017, who had their diagnostic workup and treatment at the Veterans Affairs Ann Arbor Healthcare system (VAAAHS). Patients referred solely for treatment of already diagnosed early-stage lung cancer were excluded (Figure 1). The VAAAHS institutional review board approved the study and waived the requirement of informed consent.

Figure 1.

CONSORT flow diagram.

Three physician reviewers underwent standardized training to perform chart reviews and abstracted data including patient demographics, disease presentation, tumor characteristics, imaging/diagnostic procedures, dates of referrals, pathology, and date of death or last contact as of October 2021, and causes of patient- and system-related delays in care. Data was entered into VA REDCap database with a standardized form.

The date of abnormal image was defined as the date of CT scan demonstrating a suspicious abnormality warranting invasive workup (t0, Figure 2). Date of placing the pulmonary referral was t1. Clinical stage (IA, IB, IIA, IIB) by the 8th edition TNM¹³ stage classification was retrospectively applied and determined by available imaging at the time of initial pulmonary consultation (t2). Date of diagnosis (t3) was defined by the date of the invasive procedure that determined the stage (ie, bronchoscopy, percutaneous biopsy, or surgical biopsy) or the date of the multidisciplinary conference consensus on treatment pathway (ie, direct referral to surgical resection or empiric stereotactic body radiation therapy). Date of treatment (t4) was based upon the surgical resection or start of radiation treatment. Pathologic stage was determined after surgical resection and/or after PET imaging obtained and prior to chemotherapy/radiation treatment.

Figure 2.

Time intervals along lung cancer care continuum.

Veterans who commenced with treatment within 14 weeks of abnormal imaging were defined as receiving guideline concordant (GC) care based upon the RAND guidelines,¹¹ which recommends time to diagnosis of 8 weeks and up to an additional 6 weeks for treatment initiation (T4; Figure 2). Guideline discordant (GD) care was defined as time to initial treatment exceeding 14 weeks. We looked at sub-intervals of care for both the GC and GD groups, including time between abnormal image to pulmonary referral order (T0), referral order placement to consultation visit (T1), consultation visit to definitive diagnosis (T2), and diagnosis to treatment initiation (T3) which together sum to total workup time (T4). Post hoc, we examined whether shorter total workup times were associated with improved outcomes.

Pre-specified delays in care were selected based on local experience with lung cancer care. Frequency of Veterans experiencing patient-related and system-related delays in care were determined by reviewing consult orders, imaging orders, and scheduling or provider documentation. If a patient had a recurring type of delay throughout their care, it was counted once; reasons for delays were not mutually exclusive. The prevalence of patients impacted by each type of delay was calculated. A delay was defined as 1) greater than 2 weeks for CT or 4 weeks for PET imaging from the date provider requested; 2) longer than 2 weeks for provider appointments; or 3) unplanned scheduling changes of at least 7 days.

We evaluated 2 outcome measures - rates of upstaging and survival. Patients were considered upstaged if the final pathologic stage was more advanced than the initial radiologic stage. Survival time was measured from the date of radiographic detection of malignancy (t0) until the date of death/last contact (as of October 1st, 2021), as documented in computerized patient record system (CPRS). Three- and 5-year survival were analyzed.

To identify predictors of GC care, we used multivariable logistic regression with guideline concordance as the outcome. We included age, race, Charlson comorbidity index (CCI) score,¹⁴ smoking status (current or former), reason for radiographic study (ie, lung cancer screening, symptoms of possible malignancy, or incidental detection in surveillance of a different disease), initial radiographic stage, number of encounters for invasive procedures, and treatment type as covariates.

To identify predictors of survival, we performed a Cox proportional hazard analysis with GC as the exposure. Age, race, CCI score, smoking status, reason for radiographic detection of malignancy, initial radiographic stage, histology type (adenocarcinoma, squamous cell carcinoma, large cell/undifferentiated, or no histology), and type of treatment (surgery, radiation therapy, or systemic therapy) were included as covariates. We tested the following interactions: CCI score and smoking status; detection method and smoking status; detection method and radiographic stage; and CCI score and treatment.

For continuous variables, the median and interquartile ranges are reported. For categorical variables, frequency and percentages are reported. For continuous variables, we used Mann-Whitney U test to compare groups. For categorical variables, we used chi-square testing to compare groups. A P-value of ≤.05 was considered significant. All confidence intervals (CI) reported are 95% CI. Analyses were performed with Stata Statistical Software: Release 16 (Stata-Corp. College Station, TX, 2019) or Prism v.9 (GraphPad, San Diego, CA).

Results

Cohort Demographics

Our cohort (N = 203) was primarily white (79%) men (98.5%) with mean age of 68 years (SD 6). Sixty-eight (33%) Veterans received GC care, with 135 (67%) Veterans receiving GD care. No significant demographic differences were identified between GC vs. GD groups (Table 1). All patients currently or formerly used combustible tobacco. The GD group exhibited a trend towards higher proportion of current tobacco users compared to the GC group (53% vs. 38% respectively, P = .05).

Table 1.

Characteristics of Veterans Diagnosed and Treated for Early-Stage Non-Small Cell Lung Cancer, Stratified by Guideline Concordance of the Timeliness of Lung Cancer Care

Characteristic	All Patients (N = 203)	GC (N = 68, 33%)	GD (N = 135, 67%)
Age year, median (IQR)	68 (7)	68 (5.5)	68 (7)
Male	200 (98.5%)	67 (33%)	133 (67%)
Race, N (%)
White	161 (79%)	52 (76%)	109 (81%)
Not White	14 (7%)	5 (7%)	9 (7%)
DeclinedD	28 (14%)	11 (16%)	17 (13%)
Married, N (%)	104 (51%)	37 (54%)	67 (50%)
Current Tobacco usea, N (%)	97 (48%)	26 (38%)	71 (53%)
Distance >75 miles, N (%)	101 (50%)	35 (51%)	66 (49%)
Mental health diagnosis, N (%)	68 (33%)	23 (34%)	45 (33%)
Assisted mobilityb, N (%)	56 (28%)	14 (21%)	42 (31%)
Self-driven, N (%)	81 (40%)	29 (43%)	52 (39%)
Outside facility referral	105 (52%)	36 (53%)	69 (51%)
Charlson Comorbidity Score, median (IQR)	5 (3)	5 (2)	5 (3)

Abbreviations: GC = guideline concordant, GD = guideline discordant; IQR = interquartile range.

^aAll patients in this study were current or former tobacco smokers.

^bDiagnosis of arthritis or documented use of assistive device (ie, cane, walker, or electric scooter)

Disease Presentation, Workup, and Treatment

In our cohort of 203 patients who presented with radiographic stage I and II NSCLC, 89 cases (44%) were detected incidentally, 74 cases (36%) detected for symptoms concerning for malignancy, and 40 cases (20%) by LCS (Table 2). Of 203 patients, 60% (n = 123) were radiologic stage IA. Nearly all patient cases (n = 180, 89%) were discussed at a multidisciplinary lung cancer conference, of which 57 (32%) were deemed unfit for surgery. The number and types of diagnostic procedures are presented in Table 2. Histology was adenocarcinoma (44%), squamous cell carcinoma (42%), and large cell or undifferentiated (2%). Twenty-five patients (12%) had no tissue obtained and were treated empirically, of which 5 patients (20%) were later confirmed to have NSCLC. For treatment, surgical resection was performed in 100 patients (49%), radiation therapy in 85 patients (42%) and chemotherapy/radiation in 18 patients (9%) (definitive therapy in 14 patients [78%] and palliative for those upstaged to stage IV disease in 4 patients [22%]).

Table 2.

Clinical Characteristics of Patient’s Presenting With Radiographic Stage I and II Non-Small Cell Lung Cancer Referred for Work Up and Treatment

Clinical Presentation, Work Up, and Treatment	No. of Patients (N = 203)
Detection, N (%)
Lung cancer screening	40 (20%)
Symptom triggered	74 (36%)
Incidental/surveillance	89 (44%)
Radiographic TNM stage at referral, N (%)
IA	123 (60%)
IB	29 (14%)
IIA	15 (7%)
IIB	36 (18%)
Due to size of nodule (T3)	9 (25%)
Due to lymph node involvement (N1)	27 (75%)
Solitary pulmonary nodule	73 (36%)
Size primary tumor in cm, median (IQR)	2.2 (1.5-3.2)
Procedures, N (%)
Flexible bronchoscopy	39 (19%)
Endobronchial ultrasound (EBUS)	69 (34%)
Navigational bronchoscopy	10 (5%)
CT-guided biopsy	51 (25%)
Mediastinoscopy	8 (4%)
Surgical biopsy (sub-lobar and lobar resection)	73 (36%)
Discussed in multidisciplinary conference	180 (89%)
Number of procedural encounters to diagnosis
0	22 (11%)
1	121 (60%)
2+	60 (29%)
NSCLC type, N (%)
Adenocarcinoma	89 (44%)
Squamous cell carcinoma	84 (42%)
Large cell/undifferentiated	5 (2%)
No tissue obtained	25 (12%)
Treatment, N (%)
Surgery	100 (49%)
Lobar resection	82 (84%)
Sublobar resection	16 (16%)
Radiation	85 (42%)
Chemotherapy and radiation	18 (9%)

Abbreviation: IQR= interquartile range.

Timeliness of Care

Time from abnormal imaging to treatment initiation varied, with a median time of 17.7 weeks (IQR 12.7-26.6 weeks) (Figure 2). All subintervals of care were significantly longer in the GD group compared to the GC group (P < .001) – T1 (3.0 vs. 1.8 weeks), T2 (8.1 vs. 3.1 weeks), and T3 (8.1 weeks vs. 4.4 weeks). Total time to treatment (129 days vs. 124 days, P = .68) and guideline concordance rates (40% vs. 32.6%, P = .46) were similar in the empirically treated group compared those with confirmed NSCLC. In multivariable logistic regression models, independent predictors of GC care included incidental detection of radiographic abnormality (OR= 3.9, CI 1.4-10.5), symptom-triggered radiographic detection (OR= 2.9, CI 1.0-8.2), and the number of encounters for diagnostic procedures (OR= 0.40, CI 0.22-0.71) (Table 3). Smoking status, race, and treatment type were not associated with GC care.

Table 3.

Factors Associated With Guideline-Concordant Lung Cancer Care for Veterans Presenting With Early-Stage Non-Small Cell Lung Cancer and Underwent Complete Diagnostic Work Up and Treatment

Characteristic	Odds Ratio	95% CI	P-Value
Age, by 5 years	0.79	0.58-1.07	.13
Race
White	—	Ref	—
Non-white	1.60	0.43-5.88	.49
Declined	1.60	0.66-4.11	.32
Charlson Comorbidity Index Score	0.99	0.83-1.20	.95
Smoking status
Current	—	Ref	—
Former	1.86	0.94-3.66	.07
Detection method
Lung cancer screening	—	Ref	—
Surveillance/Incidental	3.89	1.43-10.53	.01
Symptom triggered	2.93	1.04-8.20	.04
Radiographic TNM stage
IA	—	Ref	—
IB	0.56	0.20-1.59	.27
IIA	1.00	0.28-3.62	1.00
IIB	1.32	0.50-3.44	.57
Number of visits to medical center for diagnostic procedures	0.40	0.22-0.71	<.01
Treatment
Surgery	—	Ref	—
Radiation	0.49	0.23-1.06	.07
Chemo/radiation	2.34	0.66-8.28	.19

Multivariate logistic regression was used with guideline concordance as the exposure, and adjusted for age, race, Charlson comorbidity index, smoking status, reason for radiographic detection, initial radiographic stage, number of encounters for invasive procedures, and treatment type as covariates.

Delays in Care

All phases of care were impacted more by patient-related factors than system-related delays, including time to pulmonary appointment (24% vs. 22%), time to diagnosis (30% vs. 17%), and time to treatment (44% vs. 27%). The most common patient-related delays included intercurrent hospitalization/comorbidity requiring priority in management, inability to reach patient by phone, patient no-show, difficulty finding transportation, and patient preference/desire for same day testing (Table 4). The most common reasons for system (hospital)-related delays were scheduling availability and need for additional diagnostic testing (eg, imaging, lung perfusion scans or cardiac testing) requested by a provider. Total time to treatment was significantly longer in patients with any of the patient-related delay and those who had a system-related delay due to need for additional diagnostic testing,

Table 4.

Frequency of Patients Who Encountered the Patient or System Mediated Delays in Care

Type of Delay Incurred	Total No. Affected (%)	No. Upstaged	No. Not Upstaged	P-Value
Patient related delay
Hospitalized/new comorbidity	46 (23%)	20	26	.52
Current smoking	39 (19%)	15	24	.89
Unable to contact veteran	35 (17%)	16	19	.40
Transportation problem	29 (14%)	17	12	.02
Veteran declined/no show	33 (16%)	17	16	.12
Same day testing/convenience	26 (13%)	15	11	.04
System related delay
Scheduling availability	52 (26%)	24	28	.25
More testing wanted	47 (23%)	24	23	.06
Lack of imaging w/ referral	22 (11%)	10	12	0.54
Administrative changea	10 (5%)	8	2	<0.01

Chi square analysis was used to compare reasons for delay between patients who were upstaged from radiographic to pathologic stage compared to those do did not get upstaged.

^aUnforeseen change in appointment which delayed by > 1 week.

Upstaging

The rate of upstaging from initial radiographic stage at pulmonary consultation to the histopathologic stage at treatment initiation were not significantly different between the GC and GD groups (37% vs. 41% respectively, P = .6). A total of 19 patients (9%) were upstaged from radiographic stage 1 or 2 disease to pathologic 3 or 4. When comparing time intervals of care between patients who were upstaged and not, only the time from pathologic diagnosis to treatment initiation was significantly longer in the upstaged cohort (8.0 weeks vs. 4.7 weeks, P < .001). Additionally, upstaging was more prevalent in patients who experienced a transportation related delay, those who requested coordinated testing with appointments, and/or those who had a system-related delay due to administrative change (unanticipated change in appointment which delayed care by minimum of 1 week).

Given the relatively small proportion of patients who were GC, we assessed the rates of upstaging by quartiles of time from abnormal imaging to treatment initiation (T4) and found no significant difference across quartiles. Our shortest quartile was T4 less than 12.7 weeks. Post hoc, we used an alternative definition of timeliness as a T4 interval of 8 weeks or less, which we chose based upon feasibility, recommendations from other studies and international society guidelines, as well as concerns about natural progression rates.^15–17 When using this cutoff, there was a strong trend towards more upstaging if T4 exceeded 8 weeks (41% vs. 10%, P = .05) (Figure 3).

Figure 3.

Rate of upstaging stratified by total time from radiographic abnormality to treatment initiation.

Survival

Out of 203 patients in our study, 199 patients had at least 3 years of follow up and 143 patients had 5 years of follow up during our study period. Ninety-seven patients had a confirmed date of death. Sixty-one patients (31%) out of 199 were confirmed dead within 3 years, and 91 (64%) of the 143 patients were confirmed dead within 5 years of radiographic detection of lung malignancy. There was no difference in survival in the empirically treated group compared to confirmed NSCLC groups.

In a Cox proportional hazard model evaluating independent predictors of all-cause mortality, CCI score (HR = 1.29, 95% CI, 1.07-1.54), former smoking status (compared to current) (HR = 6.45, 95% CI, 1.46-28.5), symptom-triggered detection (compared to lung cancer screening) (HR = 2.92, 95% CI, 1.14-7.51), radiographic stage IIA NSCLC (compared to stage IA) (HR = 2.31, 95% CI, 1.03-5.18), radiographic stage IIB NSCLC (HR = 1.87, 95% CI, 1.01-3.44) and treatment with radiation therapy (compared to surgical treatment) (HR = 2.43, 95% CI, 1.45- 4.06) or radiation/chemotherapy (HR = 2.40, 95% CI, 1.10-5.28) had significantly increased hazard ratios for all-cause mortality (Table 5). The interaction between smoking status and Charlson score as well as the interaction between smoking status and detection method were significant and included in the final analysis.

Table 5.

Cox Proportional Hazard Model Evaluating Predictors of All-Cause Mortality Among Veterans Who Presented With Radiographic Stage 1 or 2 Non–Small Cell Lung Cancer

Characteristic	Hazard Ratio	95% CI	P-Value
Guideline concordant (T4 <14 weeks)	0.87	0.52-1.45	0.60
Age, by 5 years	1.00	0.84-1.20	0.94
Race
White	Ref		—
Non-White	0.62	0.23-1.68	0.35
Declined	0.78	0.40-1.50	0.46
Charlson Comorbidity Index Score	1.29	1.07-1.54	0.007
Smoking status at detection
Current	Ref		—
Former	6.45	1.46-28.52	0.01
Detection method
Lung cancer screening	Ref		—
Surveillance/incidental	2.00	0.77-5.20	0.15
Symptom triggered	2.92	1.13-7.51	0.03
Radiographic TNM stage
IA	Ref		—
IB	1.72	0.90-3.28	0.10
IIA	2.31	1.03-5.18	0.04
IIB	1.87	1.01-3.44	0.046
NSCLC type
Adenocarcinoma	Ref		—
Squamous cell carcinoma	1.05	0.64-1.70	0.85
Large cell/undifferentiated	3.51	1.11-11.09	0.03
Empirically treated	1.08	0.53-2.19	0.84
Treatment
Surgery	Ref		—
Radiation	2.34	1.45-4.06	0.001
Chemo/radiation	2.40	1.09-5.28	0.03

In our Cox model, GC was the binary exposure and age, race, CCI score, smoking status, reason for radiographic detection of malignancy, initial radiographic stage, histology, and treatment were included as covariates as well as the interaction between smoking status and CCI and interaction between smoking status and detection method of malignancy.

Our study was not powered to look at the impact of specific delays on survival; however, in a post hoc analysis, we did note that only 1 of the prespecified delays had a significant impact on odds of survival at 3 years. A patient who suffered a delay due to their request for coordination of care (ie, office visits coordinated with same day testing/imaging) had increased odds of survival at 3 years after radiographic detection of malignancy (OR = 3.27, 95% CI, 1.01-10.6). A trend toward decreased odds of survival at 3 years was suggested (though not statistically significant) in patients who experienced a delay due to being hospitalized or new comorbidity (OR = 0.48, 95% CI, 0.22-1.04, P = .06). System-related delays did not seem to impact odds of survival at 3 years in our study.

Discussion

Significant resources are being expended to increase the uptake of LCS amongst those eligible.^18,19,20 If successful, there will be more detection of early-stage lung cancer,⁷ underscoring the importance of developing evidence-based guidelines for timeliness of care. Healthcare systems can then allocate resources towards meeting metrics of quality that result in measurable improvement in outcomes for early-stage NSCLC.

Most Veterans in our study did not meet the RAND corporation definition of guideline concordant care for NSCLC, consistent with results from prior studies of VA lung cancer populations.²¹ However, GC care defined by time to treatment of 14 weeks did not independently impact survival in our analysis. We hypothesize that this may be due to the small number of Veterans in our cohort who fell into the GC group and/or because a definition of GC care of 14 weeks may be too long to directly impact survival. In a retrospective cohort study of over 9000 Veterans who underwent surgical resection for stage 1 NSCLC, treatment within 12 weeks from radiographic abnormality was associated with improved overall survival compared to those who had treatment greater than 12 weeks,²² providing support for stricter definitions of timely treatment in early-stage lung cancer. We explored alternative definitions of timely care in post hoc analysis examining the subgroup under 8 weeks, given that 8 weeks is feasible and the volumetric doubling time for NSCLC can be as low as 52 days.²³ Although strictly hypothesis generating, our finding that 8 weeks or shorter may decrease upstaging motivated a change in our diagnostic pathway at our institution, to address the modifiable patient-related delays of transportation issues and coordination of care. With the support of a dedicated nurse care navigator, we are focused on consolidating all diagnostic testing over a 2-day period, whereby the Veteran under-goes in-person consultation visit, imaging studies (eg, PET-CT) and bronchoscopy, as indicated after a multidisciplinary panel review of referral for abnormal imaging concerning for thoracic malignancy.

Historically, paradoxical results suggest that shorter time to treatment is associated with decreased survival in patient with lung cancer. It is unclear if this is due to patients or providers behaving urgently when imaging suggests more advanced disease. In a recent review by Zuniga et al.,¹⁰ the authors concluded that these paradoxical results are in part due to methodological limitations of prior studies, which inadequately control for final pathologic stage or confounding by acuity of cancer presentation.¹⁰ The authors also concluded that the effect of delay on survival likely varies by stage, with the impact of delay being lowest for subcentimeter nodules, and highest in stage II disease. Our study accounted for these important methodological considerations when exploring timeliness of early-stage disease. Specifically, we controlled for radiographic stage at presentation, comorbidities (via CCI score), acuity of presentation (by reason for radiographic detection), and type of cancer (our study was limited to NSCLC and accounted for histology type) in our analysis. In doing so, we found that Veterans presenting with radiographic stage IIA and IIB had significantly higher hazard for all-cause mortality after accounting for confounders and moderators, adding to the body of evidence that the radiologic stage is an important factor in survival, particularly for patients presenting with radiographic stage II disease.^10,24

Our study has certain limitations. First, our cohort consisted of Veterans who received all testing, diagnosis and treatment within in the VA healthcare system, and who were predominantly White males. Sex²⁵ and insurance coverage²⁶ both impact survival in lung cancer. Other limitations include being a single-center study, unmeasured confounders, and the use of RAND criteria to define timely treatment as expert opinions may differ, particularly in a LCS population. Additionally, our cohort had an average Charlson Comorbidity Index (CCI) score of 5, which has an estimated 10-year survival¹⁴ of 21%, irrespective of the lung cancer diagnosis. Not surprisingly, with a high comorbid burden, the 5-year survival rate in our cohort (39%) is lower than the national average in the United States for patient treated for localized disease (64%).²⁷ Additionally, patient with high CCI may be less likely to be considered candidates for lung cancer screening given competing mortality, therefore impacting the detection method of disease.

Independent of whether a shorter workup time leads to improved oncologic outcomes, achieving diagnosis and treatment in a more rapid timeframe may have other beneficial psychological effects, particularly given the high rates of anxiety and mental health issues in the Veteran population.²⁸ Our study is the first to examine the types, frequency, and impact of specific patient-related and system-related delays on the timeliness of care for radiographic stage I and II NSCLC. We identified several modifiable risks for delays in care such as lack of reliable transportation and request for coordination of testing/visits. Patients who requested coordination of care also had higher odds of survival 3 years after radiographic detection of malignancy, which we hypothesize may be due to selection bias for patients who are engaged in actively managing their healthcare.

Currently, qualitative recommendations call for hospitals to assess local barriers to timely care and stratify guidelines by radiographic stage at presentation.¹⁰ By understanding the types and impact of delays in care, hospital systems can strategically allocate resources at a local level by identifying specific bottlenecks in existing workflows, increasing access/hiring personnel, or creation of new programs. Our study highlights the need for intensive care coordination between patients and providers to maximize efficiency in time to diagnosis and treatment. With the initiatives to improve LCS and early detection, optimizing the care continuum will be important as institutions adapt to increasing volumes of early-stage disease.

Conclusion

In our study, targeting less than 14 weeks from abnormal image to treatment initiation for early-stage lung cancer did not have a meaningful impact on survival. Future studies should evaluate alternative definitions of timely care. Our study provides a framework to study timeliness of early-stage non–small cell lung cancer care and how to analyze local barriers to timely care.

Clinical Practice Points.

• It is unknown how quickly early-stage lung cancer should be treated to improve lung cancer outcomes.

• Our study aimed to understand the timeliness of care at our facility, while specifically analyzing the reasons for patient-related and system related delays in care and how timeliness impacts outcomes of upstaging and overall survival.

• Time to treatment of less than 14 weeks did not improve overall survival or rate of upstaging in our cohort.

• In post hoc analysis, time to treatment of 8 weeks was associated with lower rates of upstaging (P = .05).

• There were higher number of delays due to patient related factors, many of which may be modifiable.

• Although treatment within 14 weeks did not impact survival or upstaging for early-stage NSCLC, our data suggest that shorter timeframes may be beneficial.

• There are modifiable delays in care which may be addressed at an institutional level to improve timeliness of care.

• Our study provides a framework to study timeliness of early-stage non-small cell lung cancer care and how to analyze local barriers to timely care.

Abbreviations:

CCI

Charlson Comorbidity Index

CI

confidence interval

CPRS

centralized patient record system

CT

computed tomography

GC

guideline-concordant

GD

guideline-discordant

EBUS

endobronchial ultrasound

HR

hazard ratio

IQR

interquartile range

LCS

lung cancer screening

LDCT

low dose computer tomography

NSCLC

non-small cell lung cancer

OR

odds ratio

PET

positron emission tomography

SPN

solitary pulmonary nodule

TNM

TNM classification of malignant tumors

VA

Veterans Affairs