Effect of Clinical Trial Participation on Costs to Payers in Metastatic Non–Small-Cell Lung Cancer

PURPOSE:

The costs associated with clinical trial enrollment remain uncertain. We hypothesized that trial participation is associated with decreased total direct medical costs to health care payers in metastatic non–small-cell lung cancer.

METHODS:

In this retrospective cohort study, we linked clinical data from electronic medical records to sociodemographic data from a cancer registry and claims data from Medicare and two private insurance plans. We used a difference-in-difference analysis to estimate mean per patient per month total direct medical costs for patients enrolled on a second-line (2L) trial versus patients receiving standard-of-care 2L systemic therapy.

RESULTS:

Among 70 eligible patients, the difference-in-difference of mean per patient per month total direct medical costs between 2L trial participants and nonparticipants was –$6,663 (P = .01), for a mean savings of $45,308 per patient for the duration of 2L trial therapy. In a secondary analysis by primary insurance payer, this difference-in-difference was –$5,526 (P = .26) for patients with commercial insurance and –$7,432 (P = .01) for patients with Medicare.

CONCLUSION:

Participation in a 2L trial was associated with a $6,663 per month cost savings to health care payers for the duration of trial participation. Further studies are necessary to elucidate differences in cost savings from trial participation for Medicare and commercial payers. If confirmed, these results support health care payer investment in programs to improve clinical trial access and enrollment.

INTRODUCTION

Only 2%-8% of patients with cancer enroll in a clinical trial because of a combination of structural, clinical, physician-level, and patient-level barriers.^1-5 Financial considerations like insurance coverage contribute at every level and may widen existing disparities in enrollment, compromising the generalizability of trial findings.⁶ Trial enrollees are more likely to have Medicare insurance than the general US population,² and patients with private insurance are less likely to enroll in a trial than those with government-funded plans.⁷ Insurance networks may influence trial access through coverage of cancer centers that offer clinical trials.^6,8 Furthermore, patients have cited concerns about the costs of trial participation⁹ and insurance coverage of trials¹⁰ as barriers to trial enrollment.

There are limited data on costs to payers associated with clinical trial participation. Most studies have taken the patient perspective to assess direct medical costs incurred by trial participants versus nonparticipants within the first 1-5 years after trial enrollment. Three studies found a 3.8%-11.1% increase in direct medical costs for trial participants,^11-13 whereas other studies found that participants incurred similar or slightly lower costs compared with nonparticipants.^14-16 These studies evaluated costs incurred between 1980-2000, did not focus on costs directly attributable to a trial, and are from varying perspectives, with only one from the third-party health care payer perspective.¹³ No studies have evaluated the cost implications of trial participation to payers in the era of expensive immunotherapy and targeted agents.

We sought to assess whether participation in a therapeutic drug clinical trial is associated with changes in costs to health care payers for patients with metastatic non–small-cell lung cancer (NSCLC). Our study is unique in (1) its focus on a specific cancer type, (2) its updated assessment of this research question in the era of novel therapeutics like immunotherapy and targeted therapies, and (3) its analytic approach designed to capture the costs specifically attributable to clinical trials. Given that cancer-directed therapy is a major driver of lung cancer costs^17,18 and that the cost of cancer-directed therapy would likely be covered by trial sponsors, we hypothesized that trial participation is associated with decreased direct medical costs to health care payers.

METHODS

Patients

We identified a modern cohort of patients with de novo or recurrent metastatic NSCLC who received treatment at the Seattle Cancer Care Alliance (SCCA) in Seattle, Washington, through retrospective review of electronic medical records (EMRs). The SCCA is a National Cancer Institute–Designated Comprehensive Cancer Center that serves patients from Washington State, Alaska, Idaho, Montana, and Wyoming. The patient population is largely White and middle class, with a high proportion of patients of Asian descent relative to other US academic centers. Inclusion criteria included cytologically or histologically proven NSCLC, diagnosis between January 1, 2007, and December 31, 2015, documentation of metastatic disease, receipt of at least one dose of an antineoplastic agent within 180 days of metastatic disease, first-line (1L) therapy delivered at the SCCA, claims for at least one antineoplastic agent within 180 days of metastatic disease, and insurance enrollment for at least the first 12 months after documentation of metastatic disease. Exclusion criteria included active secondary malignancy, death within 60 days, or referral to hospice within 90 days of metastatic disease. This study was approved by the Fred Hutchinson Institutional Review Board.

Data Extraction

Patient sociodemographic characteristics, smoking history, baseline Eastern Cooperative Oncology Group performance status at diagnosis, tumor histology, date of metastatic disease confirmation, epidermal growth factor receptor (EGFR) and anaplastic lymphoma kinase (ALK) mutation status, presence of brain metastases within 60 days after metastatic disease, systemic treatment history up to five lines, therapeutic drug clinical trial enrollment, and trial characteristics were abstracted directly from EMRs. All patients in this study received 1L and second-line (2L) systemic therapy at the SCCA. Duration of therapy for each treatment line was defined as the time from the date of first administration to the date of last administration of that line's therapeutic regimen, per review of the medication administration record and clinician notes. We linked EMR-derived records to data on patient race, Spanish or Hispanic ethnicity, marital status, census-based median household income, the Area Deprivation Index¹⁹ at the Washington State level, and Rural Urban Commuting Area codes²⁰ from the Washington State Cancer Surveillance System (CSS) database. The CSS database compiles incidence, treatment, and follow-up data on all new cancer cases in 13 counties in western Washington State²¹ as part of the National Cancer Institute SEER Program. Patient background characteristics were defined at the time of metastatic disease confirmation by pathology, imaging, or clinician assessment, whichever came first. Death dates came from the CSS database complemented by EMRs, and we censored patients who were alive at the end of follow-up (March 7, 2019). We linked the data from EMRs and the CSS registry to claims data from Medicare and two private insurance plans (Premera BlueCross and Regence BlueShield). Claims included reimbursed costs for antineoplastic drugs, radiation, emergency department visits, hospitalizations, skilled nursing homes, home health, hospice, and durable medical equipment. We adjusted all costs for 2019 US dollars using the Consumer Price Index.

Study Design and Treatment

This is a retrospective cohort study and cost analysis. The exposure was therapeutic drug clinical trial participation during 2L therapy for metastatic NSCLC. The primary outcome was mean per patient per month (PPPM) total direct medical costs during 2L therapy incurred by health care payers. Secondary outcomes included duration of 2L therapy and mean aggregated total cost for the duration of 2L therapy. All costs were incurred between date of metastatic disease confirmation and date of death or last follow-up.

Statistical Analysis

Statistical analyses were performed using Stata software (version 16.1, StataCorp). No variables exceeded 10% missingness, and missing data were reclassified as unknown for the purposes of analysis. We calculated descriptive statistics including means, medians, standard deviations (SDs), interquartile range (IQR), and CIs for continuous variables and proportions for categorical variables. Differences in background characteristics between trial participants and nonparticipants were assessed using Student's t-test for continuous variables and a chi-squared test for categorical variables.

We used a difference-in-difference approach to estimate the specific effect of trial enrollment on costs, focusing on costs associated with 2L trial participation versus 2L nontrial systemic therapy. The difference-in-difference method allows us to isolate the effect of trial participation on costs from the influence of other patient characteristics on costs (eg, comorbidities and performance status) by having each patient serve as their own control within each group, trial and nontrial. A difference-in-difference approach also accounts for changes in costs from 1L to 2L within each group.²² We excluded patients with less than two lines of therapy and patients who were enrolled in a 1L trial. We calculated mean PPPM total direct medical costs from the payer perspective for 1L and 2L, respectively. Next, we calculated the mean PPPM difference between 2L and 1L costs in trial (Diff_trial) and nontrial enrollees (Diff_nontrial). Finally, we calculated the difference-in-difference: subtracting Diff_nontrial from Diff_trial. We used paired and nonpaired t-tests for statistical comparisons. For our secondary analysis, we repeated this approach with the study population stratified by insurance payer. A P value of < .05 was considered statistically significant.

RESULTS

Primary Analysis

Background characteristics.

Of 70 eligible patients, 22 (31%) were enrolled in a 2L clinical trial (Table 1). Trial participants were younger (mean age 64 v 66 years), more likely to be female (73% v 46%), never smokers (36% v 31%), and less likely to have brain metastases (14% v 31%) relative to nonparticipants. Primary insurance coverage by a commercial payer was similar between trial participants and nonparticipants (36% v 38%).

TABLE 1.

Background Characteristics by Trial Participation for Entire Study Population

There were 19 (86%) deaths among trial participants versus 46 (96%) among nonparticipants. The median overall survival from the start of 2L therapy to date of death or last follow-up was 1.7 years (95% CI, 0.6 to 2.4 years) for trial participants versus 1.0 years (95% CI, 0.8 to 1.4 years) for nonparticipants (Appendix Fig A1, online only).

Treatment and trial characteristics.

The median time from metastatic disease diagnosis to 1L therapy was shorter for trial participants (14 days, IQR: 7-23 days) than for nonparticipants (18 days, IQR: 7-29 days). Likewise, the median time from metastatic disease diagnosis to the end of 2L therapy was shorter for trial participants (13.1 months, IQR: 8.5-17.5 months) than nonparticipants (13.8 months, IQR: 7.8-18.0 months). Both groups received a median of 3 lines of systemic therapy. Trial participants were more likely to have received a programmed cell death protein 1/programmed death ligand-1 checkpoint inhibitor (50% v 21%, P = .01) or another targeted therapy (45% v 8%, P < .001) at any time during treatment.

Among the 22 trial participants, 6 (27%) were enrolled in at least one other trial beyond 2L. Sixteen patients (73%) were enrolled in at least one phase I or II trial, and seven patients (32%) were enrolled in at least one phase III trial. Seven patients (32%) participated in at least one randomized trial, and 10 (45%) participated in at least one trial of an agent later approved by the US Food and Drug Administration.

Difference-in-difference for entire study population.

For 1L therapy, the mean PPPM total direct medical costs to health care payers were similar between 2L trial participants and 2L nonparticipants ($10,393, SD ± $6,250 v $11,473, SD ± $10,858). For 2L therapy, mean PPPM total direct medical costs differed substantially ($4,808, SD ± $3,370 for 2L trial participants v $12,551, SD ± $13,598 for nonparticipants). The difference between 2L and 1L mean PPPM total direct medical costs decreased for patients enrolled on a 2L trial (Diff_trial = –$5,585, SD ± $6,541, P < .001), but increased in nonparticipants (Diff_nontrial = $1,078, SD ± $14,765, P = .62). The mean difference-in-difference (Diff_trial – Diff_nontrial) in PPPM total direct medical costs was –$6,663 (P = .01, Fig 1). For a mean duration of 2L trial therapy of 6.8 months, payers saved a mean of $45,308 per patient on a trial.

FIG 1.

Difference-in-difference in mean PPPM total direct medical costs by 2L trial participation. The difference-in-difference in mean PPPM total direct medical costs between 2L trial participants and nonparticipants is –$6,663 (P = .01) for the duration of trial participation. 1L, first line; 2L, second line; PPPM, per patient per month.

Secondary Analysis

Difference-in-difference for patients with commercial insurance.

For patients with commercial insurance coverage, the mean PPPM total direct medical costs to health care payers were lower for eight 2L trial participants relative to 18 nonparticipants during 1L therapy ($15,100, SD ± $6,764 v $19,173, SD ± $13,479) and during 2L therapy ($6,464, SD ± $2,983 v $16,063, SD ± $11,245). The difference between 2L and 1L mean PPPM total direct medical costs decreased for both patients enrolled on a 2L trial (Diff_trial = –$8,635, SD ± $8,485, P = .02) and nonparticipants (Diff_nontrial = –$3,110, SD ± $15,814, P = .42). The mean difference-in-difference (Diff_trial – Diff_nontrial) in PPPM total direct medical costs for patients with commercial insurance was –$5,526 (P = .26, Fig 2). For a mean duration of 2L trial therapy of 8.9 months, commercial insurance payers saved a mean $49,181 per patient on a trial.

FIG 2.

Difference-in-difference in mean PPPM total direct medical costs for 2L trial participation stratified by primary insurance type. (A) The difference-in-difference in mean PPPM total direct medical costs between 2L trial participants and nonparticipants with commercial insurance is –$5,526 (P = .26) for the duration of trial participation. (B) The difference-in-difference in mean PPPM total direct medical costs between 2L trial participants and nonparticipants with Medicare is –$7,432 (P = .01) for the duration of trial participation. 1L, first line; 2L, second line; PPPM, per patient per month.

Difference-in-difference for patients with Medicare.

Mean PPPM total direct medical costs for 14 2L trial participants and 30 nonparticipants were similar during 1L therapy ($7,703, SD ± $4,128 v $6,854, SD ± $5,069) for patients with Medicare. Mean PPPM costs for 2L therapy were lower for trial participants ($3,861, SD ± $3,302) than for nonparticipants ($10,444, SD ± $14,605). Diff_trial was –$3,842 (SD ± $4,620, P < .01), and Diff_nontrial was $3,591 (SD ± $13,760, P = .16). The mean difference-in-difference in PPPM total direct medical costs for patients with Medicare was –$7,432 (P = .01, Fig 2). Patients with Medicare had a mean 5.6 month duration of 2L trial therapy, translating to a mean $41,619 savings to health care payers per patient on a trial.

DISCUSSION

Our study evaluates total direct medical costs associated with 2L therapeutic drug clinical trial participation from the health care payer perspective. Our primary analysis suggests that payers save a mean $6,663 PPPM for the duration of a 2L trial in patients with metastatic NSCLC. To our knowledge, this study is the first to assess the costs specifically attributable to clinical trial participation using a difference-in-difference analysis, which allows patients to serve as their own controls while accounting for time trends in costs. Previous studies assessed total direct medical costs incurred over a 1- to 5-year period for trial participants versus nonparticipants, providing insight into longer-term differences in costs between these two groups without isolating the costs specific to trial enrollment.^11-16 Our study also focuses on a specific cancer type and stage, removing some heterogeneity in drivers of treatment costs that were present in previous studies. Importantly, it provides an updated assessment of this research question in a rapidly evolving oncology treatment landscape, including patients on immunotherapy and targeted therapies. These novel therapeutics are more expensive and may be used for longer durations than chemotherapy was used during the 1980-2000 timeframe of previous studies, particularly as survival improves in patients with metastatic NSCLC.²³

To better account for heterogeneity associated with insurance payer, we performed separate difference-in-difference analyses for patients with a primary commercial payer and those with primary Medicare payer for our secondary analysis. Both 1L and 2L costs were higher for patients with commercial insurance relative to those with Medicare. We found that commercial payers save a mean $5,526 PPPM, whereas Medicare saves a mean $7,432 PPPM for the duration of a 2L trial. Although the magnitude of cost savings was possibly higher for patients with Medicare, there was still a cost savings with commercial insurance plans. The cost savings for commercial insurance plans was not statistically significant, but this may reflect the small sample size. On the basis of a 2001 National Coverage Decision issued under President Bill Clinton's administration, Medicare covers routine care costs and research-related complications for patients on a clinical trial.²⁴ Washington State implemented legislation in 2012 stating that commercial insurances must not restrict coverage of routine patient costs for trial participants, with routine costs considered to be “items and services delivered to the enrollee that are consistent with and typically covered by the plan or coverage for an enrollee who is not enrolled in a clinical trial.”²⁵ Both the commercial insurance plans in this study (Premera BlueCross and Regence BlueShield) cover routine care costs for trial participants.^26,27 Routine care costs do not include the costs of investigational drugs or services, items or services provided only for the purposes of data collection and analysis, or services that are inconsistent with standard-of-care treatment. Premera also does not cover travel or transportation expenses,²⁶ and Regence does not specify whether these costs are covered. Despite general similarities regarding coverage of clinical trial participation between commercially insured and Medicare beneficiaries observed in our study, our data sources do not provide detailed information about individual policy coverage for the patients included in the analysis. Further studies need to evaluate the effect of type of insurance on costs of trial participation from the payer perspective.

Assuming that the magnitude of savings from trial participation is similar across other lines of therapy, both Medicare and commercial insurances should have a financial incentive to keep patients treated on trials for as long as possible. If these results are confirmed in a larger study, the magnitude of cost savings could encourage payers to develop or support programs to improve clinical trial access. This could include ensuring that centers offering trials are in-network, providing consistent coverage of routine care during trial participation, and developing programs to cover indirect costs of trial participation like transportation.⁶

Our study has several limitations. It uses retrospective data and therefore is subject to selection bias. The difference-in-difference analysis accounts for selection bias within each treatment group by allowing each patient to serve as their own control regarding 1L costs. However, there are likely differences between the two treatment groups that we could not account for with this analytic method. The generalizability of our findings may be limited across other tumor types or practice settings given that this is a single-institution study. As an academic center, our institution charges more for oncologic care relative to regional community practices and the magnitude of cost savings associated with trial participation in metastatic NSCLC may be lower in the community. We selected patients who had a minimum of 12 months of continuous insurance coverage after the date of metastatic disease confirmation, but it is possible that some patients had missing or incomplete claims data between the end of this 12-month interval and completion of 2L therapy. However, we anticipate this to be a small fraction of patients and we do not expect that this would bias our results. Our data also do not allow for further investigation of whether the cost savings associated with trial participation are driven by trial sponsor coverage of cancer-directed therapies or by other factors, such as differences in emergency department use or hospital admissions during the trial period. We also had insufficient numbers of trial participants in other lines of therapy to confirm whether similar cost savings are seen outside of 2L trial participation.

In summary, we found that 2L clinical trial participation in patients with metastatic NSCLC is associated with a mean cost savings of $6,663 PPPM and a mean savings of $45,308 per patient for the duration of 2L trial therapy to health care payers. Although this is a retrospective single-institution study, these data provide an updated analysis of the costs of trial participation in an evolving oncology treatment landscape. If our results hold in larger confirmatory studies, health care payers should have a financial incentive to improve clinical trial access and enrollment.

Appendix

FIG A1.

Kaplan-Meier survival curves by 2L trial participation. For nonparticipants, the median overall survival from start of 2L therapy was 1.0 years (95% CI, 0.8 to 1.4 years). The median overall survival was 1.7 years (95% CI, 0.6 to 2.4 years) for 2L trial participants. 2L, second line.

AUTHOR CONTRIBUTIONS

Conception and design: Cristina Merkhofer, Renato G. Martins, Bernardo H. L. Goulart

Collection and assembly of data: Shasank Chennupati, Qin Sun

Data analysis and interpretation: Cristina Merkhofer, Shasank Chennupati, Keith D. Eaton, Renato G. Martins, Scott D. Ramsey, Bernardo H. L. Goulart

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Effect of Clinical Trial Participation on Costs to Payers in Metastatic Non–Small-Cell Lung Cancer

The following represents disclosure information provided by the authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/op/authors/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).