Skip to main content
NCBI home page
JAMA Network logoLink to JAMA Network
. 2023 Aug 17;9(10):1371–1379. doi: 10.1001/jamaoncol.2023.2800

Medicaid Expansion of the Patient Protection and Affordable Care Act and Participation of Patients With Medicaid in Cancer Clinical Trials

Joseph M Unger 1,✉, Hong Xiao 1, Riha Vaidya 1, Michael LeBlanc 1, Dawn L Hershman 2
PMCID: PMC10436183  PMID: 37590003

Key Points

Question

What is the association between the Patient Protection and Affordable Care Act (ACA) Medicaid expansion and access to cancer clinical trials for patients younger than 65 years with Medicaid insurance?

Findings

In this cohort study of 51 751 patients enrolled from 1992 to 2020, the ACA Medicaid expansion was associated with an annual 19% increase in the odds of using Medicaid insurance for trial participation and an overall 52% increase in the number of patients using Medicaid insurance. The association was greater in states that implemented the expansion.

Meaning

Findings suggest that socioeconomically vulnerable patients with cancer had better access to the newest treatments in trials after this policy change.

Abstract

Importance

The Patient Protection and Affordable Care Act (ACA) Medicaid expansion resulted in increased use of Medicaid insurance nationwide. However, the association between Medicaid expansion and access to clinical trials has not been examined to date.

Objective

To examine whether the implementation of ACA Medicaid expansion was associated with increased participation of patients with Medicaid insurance in cancer clinical trials.

Design, Setting, and Participants

Data for this cohort study of 51 751 patients were from the SWOG Cancer Research Network. All patients aged 18 to 64 years and enrolled in treatment trials with Medicaid or private insurance between April 1, 1992, and February 29, 2020, were included. Interrupted time-series analysis with segmented logistic regression was used. The monthly unemployment rate and presidential administration were adjusted to reflect potential differences in Medicaid use associated with economic conditions and national administrative policies, respectively. Data analysis was conducted between June 22, 2021, and August 5, 2022.

Exposure

Implementation of Medicaid expansion on January 1, 2014, was the independent exposure variable.

Main Outcomes and Measures

The number and proportion of patients by insurance type enrolled in cancer clinical trials over time were analyzed.

Results

Overall, data for 51 751 patients were analyzed. Mean (SD) age was 50.6 (9.8) years, 67.3% of patients were female, 41.1% were younger than 50 years, and 9.1% used Medicaid. A 19% annual increase (odds ratio [OR], 1.19; 95% CI, 1.11-1.28; P < .001) was identified in the odds of patients using Medicaid after the ACA Medicaid expansion, resulting in a 52% increase (OR, 1.52; 95% CI, 1.29-1.78; P < .001) compared with what was expected in the number of Medicaid patients enrolled over time. The association was greater in states that adopted Medicaid expansion in 2014 to 2015 (OR, 1.26; 95% CI, 1.15-1.38; P < .001) compared with other states (OR, 1.08; 95% CI, 0.96-1.21; P = .20; P = .04 for interaction). By February 2020, the proportion of patients with Medicaid insurance was 17.8% (95% CI, 15.0%-20.8%; P < .001), whereas the expected proportion had ACA Medicaid expansion not occurred was 6.9% (95% CI, 4.4%-10.3%; P < .001).

Conclusions and Relevance

Findings suggest that implementation of ACA Medicaid expansion was associated with increased participation of patients using Medicaid in cancer clinical trials. Improved participation in clinical trials for Medicaid-insured patients is critical for socioeconomically vulnerable patients seeking access to the newest treatments available in trials and for improving confidence that trial findings apply to patients of all backgrounds.

This cohort study assesses participation of patients with Medicaid in cancer clinical trials after Patient Protection and Affordable Care Act Medicaid expansion.

Introduction

The health care insurance market in the United States is rapidly changing. The passage of the Patient Protection and Affordable Care Act (ACA) has led to changes in whether and how patients obtain insurance, has afforded patients new protections, and has been associated with increased numbers of patients with insurance.1,2,3 This shifting landscape of insurance accessibility, affordability, and provisions will influence whether and how individuals are treated for disease.

Cancer clinical trials, which form the backbone of clinical cancer research, are conducted within this uncertain setting. Clinical trials are vital for determining whether new experimental treatments should be adopted as standard care. Patients who choose cancer clinical trial treatment for their care face the same financial and insurance considerations as nontrial patients.4,5 Yet, remarkably, little research exists that characterizes the association between patient insurance status and enrollment in cancer clinical trials despite the recognition that the nature of clinical trial cohorts affects both their results and their interpretation.6,7,8

A key feature of the ACA was the expansion of Medicaid eligibility to individuals earning up to 138% of the federal poverty level.9 Individual states could choose to adopt the measure and receive enhanced federal Medicaid support. Medicaid expansion resulted in increased use of Medicaid insurance nationwide, including for patients with cancer, especially in states that adopted the Medicaid expansion program.3,10 However, the association between ACA Medicaid expansion and access to clinical trials has not been previously examined, to our knowledge.

Methods

Data

We examined initial enrollments in clinical trials conducted by the SWOG Cancer Research Network. All patients aged 18 to 64 years and enrolled in cancer treatment trials between April 1, 1992, and February 29, 2020, were included in this cohort study. National unemployment rates by month were extracted from the Bureau of Labor Statistics.11 We did not include enrollments beginning in March 2020 owing to the sudden increase in unemployment after the outbreak of COVID-19, which was considered unrepresentative of temporal associations between unemployment and use of Medicaid insurance. Insurance type was obtained from patients at enrollment. To highlight the association between use of Medicaid insurance and use of private insurance, patients with no insurance and military or Veterans Affairs insurance were excluded.

All trials were previously approved by an institutional review board; patients previously provided written informed consent at trial enrollment. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cohort studies.12

Model Variables

The independent variable was the date of the initial implementation of Medicaid expansion on January 1, 2014. Data on states adopting Medicaid expansion within 2 years after implementation (2014-2015) were identified from published reports.13 Patterns of insurance with Medicaid were examined between groups defined by demographic variables including age (<50 vs 50-64 years), sex, race (Black vs non-Black), ethnicity (Hispanic vs non-Hispanic), and racial and ethnic minority groups (Asian, Black, Hispanic, Native American, or Native Hawaiian or Other Pacific Islander; yes vs no). Non-Black included Asian, Native American, Native Hawaiian or Other Pacific Islander, or White.

Statistical Analysis

We used an interrupted time-series analysis to examine over time the proportion of patients using Medicaid as their insurance type. Proportions were used because the use of counts alone could be confounded by the case mix of active trials.14 In this model, the implementation of ACA Medicaid expansion represented the exposure variable, separating our time series into 2 periods: the pre-Medicaid expansion period (April 1992 to December 2013) vs the post-Medicaid expansion period (January 2014 to February 2020).

Segmented logistic regression was used. Each monthly point estimate was calculated as the number of Medicaid patients divided by the number of total enrollments. Appropriately modeling patterns of Medicaid use during the baseline period was necessary for study validity. We included a linear association of time (modified by covariates) to capture the long-term secular trend, and an interaction between a binary ACA variable (before vs after the ACA) and time to estimate the potential slope change in the proportion.15 Additionally, in accordance with the known association between economic conditions—especially the unemployment rate—and Medicaid enrollment,16,17 we included the unemployment rate as a covariate. Initial explorations of the association between the unemployment rate and the percentage of trial enrollments using Medicaid before implementation of Medicaid expansion in 2014 showed a mean (SD) lag of approximately 12 (4.4) months between each apex or nadir of the proportion of patients in trials with Medicaid insurance and the unemployment rate, respectively (Figure 1). Thus, we included the national unemployment rate with a 12-month lag time. For model parsimony and because state-level unemployment generally tracks with national unemployment trends, national (rather than state-level) unemployment rates were used. Moreover, because Medicaid use may vary by executive administrations, the model included indicator variables representing the presidential administration.18 Newey-West SEs with autocorrelations of up to 3 lags were used to accommodate serial autocorrelation in residual errors in accordance with results of the partial autocorrelation function and the Cumby-Huizinga general test.19

Figure 1. Proportion of Patients Using Medicaid Insurance and US Unemployment Rate Before Implementation of Patient Protection and Affordable Care Act Medicaid Expansion in 2014.

Figure 1.

Open in a new tab

A, The dark blue circles indicate the proportion of patients using Medicaid for their insurance over time. For presentation purposes and for comparison with the national unemployment rate, the dark blue line shows a smoothed regression line of the 6-month moving average. The light blue line indicates the monthly national unemployment rate over time. B, The proportion of trial patients using Medicaid (dark blue line) paralleled the national unemployment rate (light blue line). The mean (SD) lag time between the peak or nadir of the unemployment rate in the US and the corresponding peak or nadir, respectively, of Medicaid use in trials (indicated by the black arrows, with lag times shown in months) was approximately 1 (4.4) year. These patterns are highlighted during recessions, including the recession in 1990 to 1991, the dot-com bubble in 2000 to 2003, and the Great Recession in 2007 to 2009. C, Accounting for this lag time and represented on similar scales, the curves representing Medicaid use in trials (dark blue line) and the national unemployment rate (light blue line) are nearly superimposable.

To estimate the absolute effects of ACA Medicaid expansion on the number of enrollments using Medicaid, we repeated our analysis using negative binomial regression (to account for potential overdispersion) with the monthly number of enrollments as an offset variable to retain the interpretation of the coefficients in terms of enrollment proportions. The overall absolute association was calculated as the difference between the estimated enrollments using Medicaid (ie, factual or “actual” estimate) and the sum of those if ACA Medicaid expansion had not been enacted (ie, counterfactual estimate) during the entire post-ACA Medicaid expansion period. We simulated 10 000 estimates per month under each scenario (actual and counterfactual estimates), using the estimated coefficients and their variance-covariance matrix of a multivariate normal distribution derived from the model (analogous to a parametric bootstrap calculation). The aim was to estimate the cumulative effect of the estimator variable on the outcome variable and account for uncertainties in the estimation in the entire study period.20 We also compared the estimated proportion of patients with Medicaid in February 2020, the end of the study period, with the expected proportion had Medicaid expansion not been implemented.

All analyses were conducted in R, version 4.0.2 (R Project for Statistical Computing). A 2-sided P < .05 indicated statistical significance. P values were calculated with permutation tests, t tests, and χ2 tests. Data analysis was conducted from June 22, 2021, and August 5, 2022.

We examined whether patterns of Medicaid use for trial participation differed according to categories of age, sex, and race and ethnicity and whether patients were from states that adopted Medicaid expansion in 2014 to 2015, using interaction tests. We also separately examined aggregate patterns of Medicaid use compared with no insurance over time.

Results

Patient Characteristics

In total, 51 751 patients aged 18 to 64 years were enrolled from April 1992 to February 2020. Mean (SD) age was 50.6 (9.8) years. A total of 16 932 patients were male (32.7%), 34 819 were female (67.3%), and 21 777 were younger than 50 years (41.1%). Overall, 958 patients had unknown or missing race data (1.9%) and 2185 patients had unknown or missing ethnicity data (4.2%). Among patients with known data, 4366 were Black (8.4%) and 2341 were Hispanic (4.7%) (Table 1). Overall, 4709 patients (9.1%) had Medicaid coverage. The proportion of patients using Medicaid insurance was consistently higher among patients from racial and ethnic minority groups (Table 1). Thirty-one states comprising 60.1% of the US population enacted ACA Medicaid expansion between 2014 and 2015. Overall, the proportion of patients using Medicaid insurance increased from 8.6% during the pre-Medicaid expansion period to 11.8% in the postexpansion period (P < .001 by χ2 test).

Table 1. Characteristics of Patientsa.

Characteristic No. (%)
April 1992 to February 2020 April 1992 to December 2013 January 2014 to February 2020
Patients Patients with Medicaid Patients Patients with Medicaid Patients Patients with Medicaid
Total No. 51 751 4709 (9.1) 43 637 3749 (8.6) 8114 960 (11.8)
Sex
Female 34 819 (67.3) 3254 (9.3) 29 387 (67.3) 2576 (8.8) 5432 (66.9) 678 (12.5)
Male 16 932 (32.7) 1455 (8.6) 14 250 (32.7) 1173 (8.2) 2682 (33.1) 282 (10.5)
Age, y
<50 21 277 (41.1) 2290 (10.8) 18 414 (42.2) 1901 (10.3) 2863 (35.3) 389 (13.6)
50-64 30 474 (58.9) 2419 (7.9) 25 223 (57.8) 1848 (7.3) 5251 (64.7) 571 (10.9)
Raceb
Asian 1411 (2.8) 143 (3.0) 1129 (2.6) 102 (2.7) 282 (3.6) 41 (4.3)
Black 4366 (8.6) 1066 (24.4) 3736 (8.7) 918 (24.6) 632 (8.1) 148 (23.4)
Native American 218 (0.4) 65 (1.4) 183 (0.4) 56 (1.5) 35 (0.4) 9 (0.9)
Native Hawaiian or Other Pacific Islander 153 (0.3) 20 (0.4) 130 (0.3) 17 (0.5) 23 (0.3) 3 (0.3)
White 44 645 (87.9) 3217 (68.3) 37 818 (88.0) 2526 (67.4) 6827 (87.6) 691 (72.0)
Missing or unknown 958 NA 641 NA 317 NA
Ethnicityb
Hispanic 2341 (4.7) 711 (30.4) 1741 (4.2) 535 (30.7) 600 (7.6) 176 (29.3)
Non-Hispanic 47 225 (95.3) 3805 (8.1) 39 225 (89.9) 3046 (7.8) 7300 (90.0) 759 (10.4)
Missing or unknown 2185 NA 1971 NA 214 NA
Racial and ethnic minority group
Anyc 8497 (16.5) 2005 (23.6) 6924 (16.0) 1627 (23.5) 1573 (19.6) 378 (24.0)
None 42 941 (83.5) 2682 (6.2) 36 480 (84.0) 2110 (5.8) 6461 (80.4) 571 (8.8)
Missing or unknown 313 NA 233 NA 80 NA
Medicaid expansion state in 2014-2015d
Yes 32 419 (62.6) 2855 (8.8) 27 264 (62.5) 2250 (8.3) 5155 (63.5) 605 (11.7)
No 19 332 (37.4) 1854 (9.6) 16 373 (37.5) 1499 (9.2) 2959 (36.5) 355 (12.0)
Open in a new tab

Abbreviation: NA, not applicable.

a

Statistically significant differences in the proportions of patients with Medicaid were evident between groups in all instances except by sex in the period before Patient Protection and Affordable Care Act (ACA) Medicaid expansion (P = .06) and by Medicaid expansion state in 2014 to 2015 (P = .75).

b

Race and ethnicity data were categorized by self-report. Proportions were calculated among patients with known data.

c

Defined as Asian, Black, Hispanic, Native American, or Native Hawaiian or Other Pacific Islander.

d

Indicates whether the patient was from a state that implemented ACA Medicaid expansion in 2014 or 2015.

Change in Medicaid Use Associated With ACA Medicaid Expansion

Before the implementation of Medicaid expansion in 2014, the proportion of trial patients using Medicaid paralleled underlying economic trends as reflected by the national unemployment rate (Figure 1A); that is, use of Medicaid insurance waxed as the unemployment rate increased and waned as the unemployment rate decreased. These trends are highlighted during recessions, including the recession in 1990 to 1991, the dot-com bubble in 2000 to 2003, and the Great Recession in 2007 to 2009 (Figure 1B). In fact, the association between the unemployment rate and Medicaid use in trials followed a very consistent pattern, with a mean lag time of approximately 1 year between the peak or nadir of the unemployment rate in the US and the corresponding peak or nadir of Medicaid use in trials (Figure 1B). Accounting for this lag time and represented on similar scales, the curves representing the national unemployment rate and Medicaid use in trials are nearly superimposable (Figure 1C).

The pattern of parallel trajectories in the proportion of patients in trials who used Medicaid and the US unemployment rate was interrupted by the implementation of ACA Medicaid expansion (Figure 2). As shown, beginning in 2014 and 2015, when 31 states adopted ACA Medicaid expansion, the proportion of individuals in trials who used Medicaid began to increase even as the national unemployment rate continued to decrease.

Figure 2. Proportion of Patients Using Medicaid Insurance and US Unemployment Rate Before and After Implementation of Patient Protection and Affordable Care Act (ACA) Medicaid Expansion.

Figure 2.

Open in a new tab

The dark blue circles indicate the proportion of patients using Medicaid for their insurance over time. For presentation purposes and for comparison with the national unemployment rate, the dark blue line shows a smoothed regression line of the 6-month moving average. The light blue line indicates the monthly national unemployment rate over time. Beginning in 2014, 31 states adopted ACA Medicaid expansion; the vertical lines indicate the date at which states adopted it.

Accounting for expected patterns established in the baseline period, the proportion of patients using Medicaid insurance was much greater than the expected proportion had ACA Medicaid expansion not occurred (Figure 3). Overall, we observed a 19% increase (odds ratio [OR], 1.19; 95% CI, 1.11-1.28; P < .001) per year in the odds of trial patients using Medicaid after ACA Medicaid expansion (Table 2). Our model estimated that at the end of the study period, we would have expected 6.9% of trial patients (95% CI, 4.4%-10.3%; P < .001) to have used Medicaid insurance because national unemployment was low (3.5%). In contrast, by February 2020, the actual rate was 17.8% (95% CI, 15.0%-20.8%; P < .001). Under a different metric, 970 patients with Medicaid insurance according to a model-fitted estimate (12.0% of the total) were enrolled after the implementation of ACA Medicaid expansion, but only 640 Medicaid patients (7.9%) would have been enrolled had prior patterns persisted, a difference of 330 and a 52% increase (OR, 1.52; 95% CI, 1.29-1.78; P < .001) (eTable in Supplement 1) compared with what was expected.

Figure 3. Observed and Fitted Proportion of Patients Using Medicaid.

Figure 3.

Open in a new tab

The dark blue circles indicate the proportion of patients using Medicaid for their insurance over time. The dark blue line indicates the model-fitted line for the observed proportions; the orange line (beginning in 2014) indicates the model-fitted counterfactual proportions assuming Patient Protection and Affordable Care Act (ACA) Medicaid expansion had not occurred. When expected patterns established in the baseline period were accounted for, the proportion of patients using Medicaid for their insurance for trial participation was much greater than the expected proportion had ACA Medicaid expansion not occurred.

Table 2. Estimated Change in the Odds of Patients With Medicaid Associated With Medicaid Expansion of the Patient Protection and Affordable Care Act.

Characteristic Gradual change per year (January 2014 to February 2020) Patients with Medicaid in February 2020
OR (95% CI) P value Ratio of OR (95% CI) P value Factual estimate (95% CI), % Counterfactual estimate (95% CI), % P value
Total 1.19 (1.11-1.28) <.001 NA NA 17.8 (15.0-20.8) 6.9 (4.4-10.3) <.001
Sex
Female 1.27 (1.16-1.38) <.001 1.17 (1.01-1.36) .04 22.1 (18.0-26.4) 6.3 (3.5-10.3) <.001
Male 1.08 (0.96-1.22) .21 1 [Reference] 1 [Reference] 11.0 (8.2-14.3) 7.4 (3.5-13.3) .11
Age, y
<50 1.23 (1.10-1.36) <.001 1.09 (0.95-1.26) .23 20.6 (16.2-25.6) 6.9 (4.0-11.2) <.001
50-64 1.12 (1.02-1.24) .02 1 [Reference] 1 [Reference] 15.4 (11.5-19.9) 8.2 (5.0-12.6) .003
Race
Black 1.00 (0.83-1.21) .99 0.84 (0.69-1.04) .11 30.4 (19.0-44.2) 31.1 (12.3-56.2) .50
Non-Black 1.18 (1.09-1.28) <.001 1 [Reference] 1 [Reference] 14.9 (12.4-17.7) 5.9 (3.8-8.5) <.001
Minority groupa 1.07 (0.94-1.21) .31 0.86 (0.74-1.01) .06 28.4 (22.2-35.4) 21.3 (12.0-32.9) .13
Nonminority group 1.24 (1.13-1.36) <.001 1 [Reference] 1 [Reference] 14.5 (11.8-17.4) 4.4 (2.7-6.6) <.001
Ethnicity
Hispanic 1.11 (0.90-1.36) .34 0.91 (0.73-1.14) .42 37.3 (26.7-49.4) 25.5 (8.8-51.2) .15
Non-Hispanic 1.21 (1.12-1.31) <.001 1 [Reference] 1 [Reference] 16.0 (13.1-19.2) 5.6 (3.5-8.4) <.001
Medicaid expansion state in 2014-2015b
Yes 1.26 (1.15-1.38) <.001 1.17 (1.01-1.35) .04 16.6 (13.3-20.3) 4.6 (2.7-7.1) <.001
No 1.08 (0.96-1.21) .20 1 [Reference] 1 [Reference] 19.8 (15.0-25.3) 14.0 (6.5-25.5) .12
Open in a new tab

Abbreviations: NA, not applicable; OR, odds ratio.

a

Defined as Asian, Black, Hispanic, Native American, or Native Hawaiian or Other Pacific Islander.

b

Indicates whether the patient was from a state that implemented the Patient Protection and Affordable Care Act Medicaid expansion in 2014 or 2015.

Additional Analyses

When we examined data by key variables, we found a stronger 27% annual increase (OR, 1.27; 95% CI, 1.16-1.38; P < .001) in the odds of Medicaid use among female patients compared with an increase of 8% for male patients (OR, 1.08; 95% CI, 0.96-1.22; P = .21), a relative 17% increase (ratio of OR, 1.17; 95% CI, 1.01-1.36; P = .04 for interaction) (Table 2; Figure 3; eFigure 1A and B in Supplement 1). A total of 683 female patients were enrolled after the implementation of Medicaid expansion, whereas the expected rate was 391, an increase of 76.4% (95% CI, 44.2%-112.4%; P < .001).

Similarly, the annual increase in the odds of Medicaid use among patients from states that adopted Medicaid expansion in 2014 to 2015 was 26% (OR, 1.26; 95% CI, 1.15-1.38; P < .001) compared with 8% for other patients (OR, 1.08; 95% CI, 0.96-1.21; P = .20; P = .04 for interaction), also a relative 17% increase (ratio of OR, 1.17; 95% CI, 1.01-1.35; P = .04 for interaction) (Table 2; Figure 3; eFigure 2A and B in Supplement 1). A total of 610 patients from states that implemented ACA Medicaid expansion in 2014 to 2015 were enrolled, whereas the expected rate was 361, an increase of 70.3% (95% CI, 40.0%-103.9%; P < .001). In contrast, there were no statistically significant differences in patterns by age, race, or ethnicity. Among patients with Medicaid or no insurance after the implementation of ACA Medicaid expansion, as the proportion of Medicaid patients increased, the proportion of patients with no insurance decreased (eFigure 3 in Supplement 1).

Discussion

In this study of 51 751 patients enrolled in cancer clinical trials during nearly 4 decades, the implementation of ACA Medicaid expansion in 2014 was associated with an annual 19% increase in the odds of using Medicaid insurance for trial participation. By early 2020, 17.8% of patients younger than 65 years used Medicaid insurance, whereas an estimated 6.9% of patients would have used Medicaid had ACA Medicaid expansion not occurred based on model assumptions, especially those pertaining to underlying economic conditions at the time. Furthermore, the implementation of Medicaid expansion in 2014 was associated with a 52% increase in the number of patients using Medicaid for insurance through early 2020.

In a key finding, the association of ACA Medicaid expansion with increased use of Medicaid insurance was much greater in states that implemented the expansion in 2014 to 2015. This monotonic association supports the hypothesis that it is the ACA’s Medicaid expansion in particular (rather than the ACA in general) that has resulted in an overall increase of Medicaid-insured patients in cancer clinical trials. Furthermore, the implementation of Medicaid expansion was associated with a larger increase in female patients using Medicaid insurance than male patients. This observation is consistent with known patterns of greater use of Medicaid insurance by adults among women (60%) compared with men (40%), given lower mean incomes and eligibility criteria (especially pregnancy and care of dependents) that more commonly include women.21 By extension, there was likely a larger at-risk population of women than men just above the federal poverty line who benefited from Medicaid expansion.

The association between the implementation of the ACA and insurance status for the general cancer population has been examined. The percentage of patients with no insurance decreased after the ACA implementation, especially in Medicaid expansion states.22,23,24,25,26 Similarly, ACA Medicaid expansion was associated with an increased rate of individuals insured by Medicaid.10,27 Although the influence of insurance status on clinical trial participation has been recognized and addressed through state regulations, evidence on the repercussions of these regulations is mixed. In California, a single-institution study showed that after implementation of California’s clinical trial coverage mandate, a smaller proportion of patients declined trial participation, and none declined due to insurance concerns.28 In contrast, a larger study of several community sites showed no difference in annual trial accrual between states that had coverage mandates and those that did not.29 These studies were limited in the types or phases of trials included and the number of sites participating in the study.

To our knowledge, this is the first study to examine the association between ACA Medicaid expansion and participation of Medicaid patients in cancer clinical trials at a national level. A report from the Centers for Disease Control and Prevention indicated that among individuals aged 18 to 64 years in 2019 (excluding uninsured individuals and those with military insurance), an estimated 16% had Medicaid as their source of insurance.30 Historically, the proportion of patients with Medicaid insurance in cancer clinical trials has been notably lower.31,32 The finding that 17.6% of patients in our data set used Medicaid at the end of the study period suggests that, with respect to this important social determinant of health, trial populations became more representative of the general cancer population after the implementation of ACA Medicaid expansion.

Our study also demonstrated that before the implementation of ACA Medicaid expansion, Medicaid use for trial patients closely tracked with economic conditions. This novel observation reveals how the sociodemographic composition of clinical trial cohorts can be highly variable over time and should be recognized to be, in part, organic manifestations of extant socioeconomic conditions. Such an awareness argues for greater care in interpreting the findings from studies that lack a reliable control group, especially because these social and economic factors have been shown to be determinative of outcomes, even for trial patients with access to protocol-guided care.7,33 Thus, for instance, the interpretation of single-group cohort studies needs to account for the socioeconomic profile of patients in addition to the clinical and treatment variables that are more commonly considered. Conversely, our findings reinforce the critical value of conducting randomized clinical trials when feasible because the process of randomization will balance both known and unknown variables between comparison groups, maintaining the internal validity of study findings regardless of the societal influences that may have been associated with study enrollment.

The findings suggest that targeted policies can act to counterbalance the potential adverse consequences of societal influences on trial composition. According to our findings, the implementation of ACA Medicaid expansion interrupted the typically strong association between the unemployment rate and the participation of Medicaid-insured patients in trials. An outstanding question is whether the participation of Medicaid-insured patients in clinical trials is now wholly divorced from economic trends or may continue to fluctuate. One possibility is that the composition of trial populations may be more susceptible to economic conditions among patients from states that have not adopted Medicaid expansion, as our findings suggested. The recently enacted Cancer Treatment Act, which mandates that state-level Medicaid programs cover the routine care costs of clinical trials for Medicaid patients, may maintain or even further improve access to clinical trials for patients with Medicaid insurance, especially in states that have not adopted ACA Medicaid expansion. Research will be required to further address this question. These considerations are important for the interpretation of trial findings. For instance, 1 study showed that the usual benefits of positive new treatments in trials were attenuated for patients with Medicaid or no insurance.7 Therefore, adequate representation of Medicaid patients in trials would improve confidence that the results of trials are applicable to patients with any insurance type and would further improve the ability to identify why differential benefits of new treatments between socioeconomic patient groups may occur. These considerations have implications for other policies aimed to address the influence of social determinants of health in access to clinical trials. For instance, the US Food and Drug Administration recently provided draft guidance for sponsors to better ensure that prospective strategies are incorporated at the trial design to enhance participation of underrepresented patient groups.34

Limitations

Our study was conducted using data from 1 National Cancer Institute–sponsored cancer network group, which may limit generalizability to other research settings, including pharmaceutical company–sponsored trials, wherein patterns of enrollment disparities are different than for federally sponsored studies.8 Additionally, we used national unemployment rates instead of age- or sex-specific unemployment rates owing to inconsistent reporting by demographic factors. However, age- and sex-specific unemployment rates were generally proportional to the national unemployment rate during the study period.11 Our results do not reveal whether patients insured by Medicaid after the implementation of ACA Medicaid expansion may have otherwise participated in a trial had Medicaid expansion not occurred (for instance, with no insurance). Finally, the reliability of our findings is informed by modeling assumptions, including by the extent to which patterns of Medicaid use before vs after ACA Medicaid expansion were due to Medicaid expansion itself. Although our model provided a good fit to the data, strengthening internal validity, other unknown temporal confounders may have been associated with the results.

Conclusions

Our cohort study found that the expansion of Medicaid under the ACA was associated with increased access to cancer clinical trials for patients with Medicaid insurance. These findings are important because they suggest that socioeconomically vulnerable patients have better access to the newest treatments in trials because of this policy change. This result is also critical for researchers and clinicians because improved participation of socioeconomically vulnerable patients in trials improves confidence that trial findings will apply to all patients.

Supplement 1.

eFigure 1. Observed and Fitted Percentage of Patients Using Medicaid

eFigure 2. Observed and Fitted Proportion of Patients Using Medicaid

eFigure 3. The Proportion of Patients With Medicaid Insurance Among Patients With Medicaid (N=4,709) or No Insurance (N=4,353)

eTable. Estimated Change in the Number of Patients With Medicaid Associated With Medicaid Expansion of the ACA

Click here for additional data file. (299.4KB, pdf)
Supplement 2.

Data Sharing Statement

Click here for additional data file. (16.3KB, pdf)

References

  • 1.Berchick ER, Hood E, Barnett JC. Health insurance coverage in the United States: 2017. Accessed July 10, 2023. https://www.census.gov/content/dam/Census/library/publications/2018/demo/p60-264.pdf
  • 2.National Center for Health Statistics. Health, United States, 2016: With Chartbook on Long-term Trends in Health. National Center for Health Statistics; 2017. Accessed July 10, 2023. https://www.ncbi.nlm.nih.gov/books/NBK453378/ [PubMed]
  • 3.Miller S, Wherry LR. Health and access to care during the first 2 years of the ACA Medicaid expansions. N Engl J Med. 2017;376(10):947-956. doi: 10.1056/NEJMsa1612890 [DOI] [PubMed] [Google Scholar]
  • 4.Shankaran V, Unger JM, Darke AK, et al. S1417CD: a prospective multicenter cooperative group-led study of financial hardship in metastatic colorectal cancer patients. J Natl Cancer Inst. 2022;114(3):372-380. doi: 10.1093/jnci/djab210 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Ramsey S, Blough D, Kirchhoff A, et al. Washington State cancer patients found to be at greater risk for bankruptcy than people without a cancer diagnosis. Health Aff (Millwood). 2013;32(6):1143-1152. doi: 10.1377/hlthaff.2012.1263 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Busch AB, He Y, Zelevinsky K, O’Malley AJ. Predicting participation in psychiatric randomized controlled trials: insights from the STEP-BD. Psychiatr Serv. 2015;66(8):817-823. doi: 10.1176/appi.ps.201300557 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Unger JM, Blanke CD, LeBlanc M, et al. Association of patient demographic characteristics and insurance status with survival in cancer randomized clinical trials with positive findings. JAMA Netw Open. 2020;3(4):e203842. doi: 10.1001/jamanetworkopen.2020.3842 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Unger JM, Hershman DL, Osarogiagbon RU, et al. Representativeness of Black patients in cancer clinical trials sponsored by the National Cancer Institute compared with pharmaceutical companies. J Natl Cancer Inst Cancer Spectr. 2020;4(4):pkaa034. doi: 10.1093/jncics/pkaa034 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.US Department of Health and Human Services . About the Affordable Care Act. Updated March 17, 2022. Accessed July 5, 2022. https://www.hhs.gov/healthcare/about-the-aca/index.html
  • 10.Moss HA, Wu J, Kaplan SJ, Zafar SY. The Affordable Care Act’s Medicaid expansion and impact along the cancer-care continuum: a systematic review. J Natl Cancer Inst. 2020;112(8):779-791. doi: 10.1093/jnci/djaa043 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.US Bureau of Labor Statistics . Databases, tables & calculators by subject. Accessed July 5, 2022. https://www.bls.gov/data/
  • 12.von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP; STROBE Initiative . The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. J Clin Epidemiol. 2008;61(4):344-349. doi: 10.1016/j.jclinepi.2007.11.008 [DOI] [PubMed] [Google Scholar]
  • 13.Kaiser Family Foundation . Status of state Medicaid expansion decisions: interactive map. Published May 24, 2023. Accessed July 8, 2022. https://www.kff.org/medicaid/issue-brief/status-of-state-medicaid-expansion-decisions-interactive-map/
  • 14.Unger JM, Beauchemin M, Hershman DL. Adolescent and young adult enrollment to an NCI-sponsored national clinical trials network research group over 25 years. Cancer. 2021;127(24):4574-4584. doi: 10.1002/cncr.33855 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Ai C, Norton EC. Interaction terms in logit and probit models. Econ Lett. 2003;80:123-129. doi: 10.1016/S0165-1765(03)00032-6 [DOI] [Google Scholar]
  • 16.Benitez J, Perez V, Seiber E. Medicaid access during economic distress: lessons learned from the Great Recession. Med Care Res Rev. 2021;78(5):490-501. doi: 10.1177/1077558720909237 [DOI] [PubMed] [Google Scholar]
  • 17.Cawley J, Moriya AS, Simon K. The impact of the macroeconomy on health insurance coverage: evidence from the Great Recession. Health Econ. 2015;24(2):206-223. doi: 10.1002/hec.3011 [DOI] [PubMed] [Google Scholar]
  • 18.Thompson FJ, Burke C. Executive federalism and Medicaid demonstration waivers: implications for policy and democratic process. J Health Polit Policy Law. 2007;32(6):971-1004. doi: 10.1215/03616878-2007-039 [DOI] [PubMed] [Google Scholar]
  • 19.Cumby RE, Huizinga J. Testing the autocorrelation structure of disturbances in ordinary least squares and instrumental variables regressions. Econometrica. 1992;60(1):185-195. doi: 10.2307/2951684 [DOI] [Google Scholar]
  • 20.King G, Tomz M, Wittenberg J. Making the most of statistical analyses: improving interpretation and presentation. Am J Pol Sci. 2000;44:347-361. doi: 10.2307/2669316 [DOI] [Google Scholar]
  • 21.Kaiser Family Foundation . Medicaid coverage for women. Published February 17, 2022. Accessed August 6, 2022. https://www.kff.org/womens-health-policy/issue-brief/medicaid-coverage-for-women/
  • 22.Han X, Yabroff KR, Ward E, Brawley OW, Jemal A. Comparison of insurance status and diagnosis stage among patients with newly diagnosed cancer before vs after implementation of the Patient Protection and Affordable Care Act. JAMA Oncol. 2018;4(12):1713-1720. doi: 10.1001/jamaoncol.2018.3467 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Moss HA, Havrilesky LJ, Chino J. Insurance coverage among women diagnosed with a gynecologic malignancy before and after implementation of the Affordable Care Act. Gynecol Oncol. 2017;146(3):457-464. doi: 10.1016/j.ygyno.2017.06.012 [DOI] [PubMed] [Google Scholar]
  • 24.Michel KF, Spaulding A, Jemal A, Yabroff KR, Lee DJ, Han X. Associations of Medicaid expansion with insurance coverage, stage at diagnosis, and treatment among patients with genitourinary malignant neoplasms. JAMA Netw Open. 2021;4(5):e217051. doi: 10.1001/jamanetworkopen.2021.7051 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Barnes JM, Harris JK, Brown DS, King A, Johnson KJ. Impacts of the Affordable Care Act dependent coverage provision on young adults with cancer. Am J Prev Med. 2019;56(5):716-726. doi: 10.1016/j.amepre.2018.12.011 [DOI] [PubMed] [Google Scholar]
  • 26.Parsons HM, Schmidt S, Tenner LL, Bang H, Keegan TH. Early impact of the Patient Protection and Affordable Care Act on insurance among young adults with cancer: analysis of the dependent insurance provision. Cancer. 2016;122(11):1766-1773. doi: 10.1002/cncr.29982 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Cannon RB, Shepherd HM, McCrary H, et al. Association of the Patient Protection and Affordable Care Act with insurance coverage for head and neck cancer in the SEER database. JAMA Otolaryngol Head Neck Surg. 2018;144(11):1052-1057. doi: 10.1001/jamaoto.2018.1792 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.McBride G. More states mandate coverage of clinical trial costs, but does it make a difference? J Natl Cancer Inst. 2003;95(17):1268-1269. doi: 10.1093/jnci/95.17.1268 [DOI] [PubMed] [Google Scholar]
  • 29.Ellis SD, Carpenter WR, Minasian LM, Weiner BJ. Effect of state-mandated insurance coverage on accrual to community cancer clinical trials. Contemp Clin Trials. 2012;33(5):933-941. doi: 10.1016/j.cct.2012.06.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Cohen RA, Cha AE, Terlizzi EP, Martinez ME. Demographic Variation in Health Insurance Coverage: United States, 2019: National Health Statistics Reports, No. 159. National Center for Health Statistics; 2021. [PubMed] [Google Scholar]
  • 31.Baquet CR, Ellison GL, Mishra SI. Analysis of Maryland cancer patient participation in National Cancer Institute–supported cancer treatment clinical trials. J Clin Oncol. 2008;26(20):3380-3386. doi: 10.1200/JCO.2007.14.6027 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.St Germain DC, McCaskill-Stevens W. Use of a clinical trial screening tool to enhance patient accrual. Cancer. 2021;127(10):1630-1637. doi: 10.1002/cncr.33399 [DOI] [PubMed] [Google Scholar]
  • 33.Unger JM, Vaidya R, Albain KS, et al. Sex differences in risk of severe adverse events in patients receiving immunotherapy, targeted therapy, or chemotherapy in cancer clinical trials. J Clin Oncol. 2022;40(13):1474-1486. doi: 10.1200/JCO.21.02377 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.US Food and Drug Administration . Diversity plans to improve enrollment of participants from underrepresented racial and ethnic populations in clinical trials: draft guidance for industry: availability. Published April 2022. Accessed August 6, 2022. https://www.fda.gov/regulatory-information/search-fda-guidance-documents/diversity-plans-improve-enrollment-participants-underrepresented-racial-and-ethnic-populations?utm_medium=email&utm_source=govdelivery

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplement 1.

eFigure 1. Observed and Fitted Percentage of Patients Using Medicaid

eFigure 2. Observed and Fitted Proportion of Patients Using Medicaid

eFigure 3. The Proportion of Patients With Medicaid Insurance Among Patients With Medicaid (N=4,709) or No Insurance (N=4,353)

eTable. Estimated Change in the Number of Patients With Medicaid Associated With Medicaid Expansion of the ACA

Click here for additional data file. (299.4KB, pdf)
Supplement 2.

Data Sharing Statement

Click here for additional data file. (16.3KB, pdf)

Articles from JAMA Oncology are provided here courtesy of American Medical Association

RESOURCES