The Affordable Care Act and Cancer Stage at Diagnosis Among Young Adults

Abstract

The Affordable Care Act–dependent coverage expansion provision implemented in 2010 allows young adults to be covered under their parents’ health insurance until age 26 years, and millions of young adults have gained insurance as a result. The impact of this policy on cancer patients has yet to be determined. Using 2007 to 2012 data from 18 registries of the Surveillance, Epidemiology, and End Results Program, comparing cancer patients age 19 to 25 years to a control group of patients age 26 to 34 years who were not affected by the provision, we observed a 2.0 (95% confidence interval [CI] = 0.7 to 3.4) percentage point decrease in uninsured rate and a 2.7 (95% CI = 0.6 to 4.8) percentage point increase in diagnosis at stage I disease for patients age 19-25 years. Further analyses by specific cancer site revealed that the statistically significant shifts were confined to carcinoma of cervix (21.2, 95% CI = 9.6 to 32.7 percentage points) and osseous and chondromatous neoplasms (14.4, 95% CI = 0.3 to 28.5 percentage points), which are detectable by either screening or clinical manifestation. These early observations suggest the policy has had positive benefits in cancer outcomes.

Young adults have the highest uninsured rate among all age groups in the United States ( 1 ). In September of 2010, as part of the Patient Protection and Affordable Care Act (ACA), a dependent coverage expansion (DCE) mandate required insurers to allow children to remain as dependents on their parents’ insurance plans until age 26 years. Studies on the impact of the mandate have indicated increases in insurance coverage, access to care, and certain preventive service uptake among eligible young adults ( 2–5 ). Among cancer patients, lack of insurance is known to be associated with advanced stage at diagnosis, especially for cancers that can be detected early through screening or clinical manifestation ( 6 , 7 ). A recent study on cervical cancer using the National Cancer Database (NCDB) identified a shift towards early-stage disease among young women age 21 to 25 years under the ACA-DCE, for whom cervical cancer screening is the only recommended cancer screening service ( 8 , 9 ). However, whether similar patterns exist for other cancers that could be detected through symptom assessment is unknown. This study extends the NCDB study to examine changes in stage at diagnosis for all cancers among young adults under this provision, using a population-based sample from the Surveillance, Epidemiology, and End Results (SEER) Program ( 10 ).

Cancer patients age 19 to 34 years diagnosed between 2007 and 2009 (pre-ACA) and between 2011 and 2012 (post-ACA) were identified from 18 SEER registries ( 10 ). These 18 registries cover 28% of the population ( 11 ). The year 2010 was excluded as a washout/phase-in period. The “AYA Site Recode” scheme, adapted specially for tumors of adolescents and young adults ( 12 ), was used to identify cancer types, except that carcinoma of the cervix was separated from carcinoma of the uterus. Stage at diagnosis was coded according to the American Joint Committee on Cancer (AJCC) 6th edition. Noninvasive cancers (except for bladder cancer as per the cancer registry contention [ 13 ]) and cancers without defined or applicable AJCC staging, such as leukemia and central nervous system tumors, were excluded. The final analyses were based on 44 669 cancer patients diagnosed with first primary stage I-IV or unknown stage.

We used a pre/post design and conducted difference-in-differences (DID) analyses. The study population was described by age group (intervention group, age 19-25 years, and control group, age 26-34 years) and pre/post-ACA-DCE (2007-2009 vs 2011-2012). Characteristic differences between pre- and post-ACA-DCE populations were tested using Chi-square tests. Linear probability models were fitted in the DID analyses to examine the changes in proportion of stage I disease (early stage), and P values were calculated for Wald Chi-square tests. The changes in proportion of stage II-IV diseases were also examined in a supplementary analysis. We conducted both crude and multivariable analyses adjusting for sex, race/ethnicity, age at diagnosis, urban/rural residence, marital status at diagnosis, and county-level education (percentage less than high school). Analyses were conducted for total cancer patients and for specific cancer sites, with the less common cancers combined in the “Other” category. All analyses were conducted using SAS 9.4 (SAS Institute, Cary NC). A P value of less than .05 was considered statistically significant, and all statistical tests were two-sided.

Compared with the control group of people age 26 to 34 years, the intervention group of people age 19 to 25 years was more likely to be male or unmarried ( Table 1 ). The uninsured rate in cancer patients age 19 to 25 years decreased from 9.1% pre-ACA-DCE to 7.7% post-ACA-DCE while it increased from 6.6% to 7.2% in patients age 26 to 34 years, resulting in a net change of -2.1 (95% confidence interval [CI] = -3.3 to -1.0) percentage points ( P <  .001) in the unadjusted analysis and -2.0 (95% CI = -3.4 to -0.7) percentage points in the adjusted analysis ( P =  .004). The proportion of stage I disease for all cancers increased from 51.6% to 54.0% in patients age 19 to 25 years whereas it remained unchanged (55.1% to 54.9%) in patients age 26 to 34 years, resulting in a net change of 2.6 (95% CI = 0.5 to 4.7) percentage points ( P =  .02) in the unadjusted analysis and 2.7 (95% CI = 0.6 to 4.8) percentage points ( P =  .01) in the multivariable analysis ( Table 2 ). When the analysis was stratified by specific cancer site, statistically significant changes were confined to carcinoma of cervix and osseous and chondromatous neoplasms, with adjusted net changes of 21.2 (95% CI = 9.6 to 32.7) percentage points ( P < .001) and 14.4 (95% CI = 0.3, 28.5) percentage points ( P =  .05), respectively ( Table 2 ). During the same time interval, the proportion of stage II-IV diseases statistically significantly decreased in the group age 19 to 25 years for total cancer, for carcinoma of cervix, and for osseous and chondromatous neoplasms ( Supplementary Table 1 , available online).

Table 1.

Characteristics of young adult cancer patients diagnosed in 2007-2009 and 2011-2012, SEER 18 registries *

Characteristic		Pre-ACA (2007-2009) (n = 26 174) No. (%)		Post-ACA (2011-2012) (n = 18 495)		Chi-square P (pre- vs post-ACA)
	(n = 44 669) No. (%)	19-25 y (n = 7119)No. (%)	26-34 y (n = 19 055)No. (%)	19-25 y (n = 4897) No. (%)	26-34 y (n = 13 598) No. (%)
Sex						.06
Female	27 216 (60.9)	3802 (53.4)	12 049 (63.2)	2621 (53.5)	8744 (64.3)
Male	17 453 (39.1)	3317 (46.6)	7006 (36.8)	2276 (46.5)	4854 (35.7)
Race/ethnicity						.03
Non-Hispanic white	26 452 (59.2)	4308 (60.5)	11 314 (59.4)	2899 (59.2)	7931 (58.3)
Non-Hispanic black	4064 (9.1)	612 (8.6)	1791 (9.4)	407 (8.3)	1254 (9.2)
Hispanic	9414 (21.1)	1525 (21.4)	3913 (20.5)	1106 (22.6)	2870 (21.1)
Non-Hispanic other	4739 (10.6)	674 (9.5)	2037 (10.7)	485 (9.9)	1543 (11.3)
Residence †						.46
Metropolitan counties	40 744 (91.2)	6534 (91.8)	17 319 (90.9)	4502 (91.9)	12 389 (91.1)
Nonmetropolitan counties	3851 (8.6)	569 (8)	1709 (9)	390 (8)	1183 (8.7)
Unknown	74 (0.2)	16 (0.2)	27 (0.1)	5 (0.1)	26 (0.2)
Marital status ‡						<.001
Married	17 186 (38.5)	1189 (16.7)	9326 (48.9)	621 (12.7)	6050 (44.5)
Not married	23 714 (53.1)	5418 (76.1)	8222 (43.1)	3834 (78.3)	6240 (45.9)
Unknown	3769 (8.4)	512 (7.2)	1507 (7.9)	442 (9)	1308 (9.6)
Insurance status						<.001
Any Medicaid	6963 (15.6)	1173 (16.5)	2693 (14.1)	829 (16.9)	2268 (16.7)
Other insured	31 476 (70.5)	4853 (68.2)	13 789 (72.4)	3393 (69.3)	9441 (69.4)
Uninsured	3254 (7.3)	647 (9.1)	1248 (6.5)	375 (7.7)	984 (7.2)
Unknown	2976 (6.7)	446 (6.3)	1325 (7)	300 (6.1)	905 (6.7)
County % with <High school education §						.94
2.6-10.8	11 214 (25.1)	1793 (25.2)	4796 (25.2)	1245 (25.4)	3380 (24.9)
10.9-14.0	11 184 (25)	1745 (24.5)	4819 (25.3)	1189 (24.3)	3431 (25.2)
14.1-20.7	11 903 (26.6)	1935 (27.2)	5034 (26.4)	1335 (27.3)	3599 (26.5)
20.8-42.6	10 365 (23.2)	1646 (23.1)	4404 (23.1)	1128 (23)	3187 (23.4)
Unknown	3 (0)	0 (0)	2 (0)	0 (0)	1 (0)
Cancer types						<.001
Thyroid carcinoma	8575 (19.2)	1302 (18.3)	3543 (18.6)	1013 (20.7)	2717 (20)
Germ cell and trophoblastic neoplasms of gonads	6160 (13.8)	1425 (20)	2234 (11.7)	957 (19.5)	1544 (11.4)
Melanoma	5609 (12.6)	902 (12.7)	2451 (12.9)	589 (12)	1667 (12.3)
Breast carcinoma in women	5021 (11.2)	220 (3.1)	2661 (14)	136 (2.8)	2004 (14.7)
Hodgkin’s lymphoma	3888 (8.7)	1105 (15.5)	1229 (6.4)	743 (15.2)	811 (6)
Non-Hodgkin’s lymphoma	3179 (7.1)	613 (8.6)	1312 (6.9)	403 (8.2)	851 (6.3)
Carcinoma of cervix	2330 (5.2)	175 (2.5)	1253 (6.6)	108 (2.2)	794 (5.8)
Carcinoma of colon and rectum	1927 (4.3)	191 (2.7)	906 (4.8)	138 (2.8)	692 (5.1)
Soft tissue sarcomas (excluding Kaposi sarcoma)	1256 (2.8)	289 (4.1)	469 (2.5)	187 (3.8)	311 (2.3)
Carcinoma of kidney	1123 (2.5)	88 (1.2)	541 (2.8)	81 (1.7)	413 (3)
Carcinoma of head and neck	1027 (2.3)	160 (2.2)	455 (2.4)	112 (2.3)	300 (2.2)
Osseous and chondromatous neoplasms	668 (1.5)	226 (3.2)	187 (1)	123 (2.5)	132 (1)
Carcinoma of gonads	579 (1.3)	80 (1.1)	253 (1.3)	56 (1.1)	190 (1.4)
Carcinoma of stomach	433 (1)	47 (0.7)	196 (1)	27 (0.6)	163 (1.2)
Carcinoma of trachea, bronchus, and lung	412 (0.9)	37 (0.5)	210 (1.1)	34 (0.7)	131 (1)
Other	2482 (5.6)	259 (3.6)	1155 (6.1)	190 (3.9)	878 (6.5)

*Year 2010 was excluded from analysis as a washout/phase-in period. Differences between “pre-ACA” and “post-ACA” were tested with two-sided Chi-square tests excluding “Unknown” categories. ACA = Affordable Care Act.

†County-level rural/urban attributes were from United States Department of Agriculture External Web Site Policy.

‡Not married includes those who are divorced, separated, single (never married), unmarried or domestic partner, or widowed.

§County-level attributes were extracted from the 2008-2012 Census American Community Survey.

We found a 2% reduction in uninsured rate among cancer patients age 19 to 25 years following the ACA-DCE, supplementing the evidence of increased coverage among the general population from national surveys ( 2 , 3 , 14 , 15 ). Our finding of a shift toward early-stage carcinoma of cervix in this age group confirms the recent similar finding based on the NCDB ( 16 ) and suggests that the shift may be because of increased access to screening services. This finding remained statistically significant after Bonferroni adjustment ( P =  .002 in the crude analysis and P =  .005 in the multivariate analysis). Moreover, we also found a shift toward early stage for osseous and chondromatous neoplasms, which mostly occur in the oral cavity and are amenable to early detection during dental exams. This coincided with increased dental insurance coverage and dental checkup as spillover effects of the ACA-DCE ( 5 , 17 ). Although not statistically significant, seven of the 14 remaining cancers also showed increases in early-stage diseases with simultaneous decreases in late-stage diagnosis ( Table 2 ; Supplementary Table 1 , available online). Exceptions to this generally favorable pattern are cancers that are fast growing and thus less likely to be detected early through clinical symptom assessment, such as cancers of stomach and lung.

Table 2.

Difference-in-differences analysis of changes in early stage at diagnosis among young adults age 19-34 years, before and after implementation of the Affordable Care Act-dependent coverage expansion, 2007-2009 and 2011-2012, SEER 18 registries *

Cancer type	Age 19-25 y, % (95% CI)			Age 26-34 y, % (95% CI)			DID (age 19-25 y minus age 26-34 y), % (95% CI)
	Pre-ACA	Post-ACA	Pre/post-difference (post-pre)	Pre-ACA	Post-ACA	Pre/post-difference (post-pre)	Unadjusted DID estimate, % (95% CI)	Unadjusted P†	Adjusted DID estimate, % (95% CI)	Adjusted P†,‡
Total	51.6	54.0	2.4	55.1	54.9	−0.2	2.6	.02	2.7	.01
	(50.4 to 52.8)	(52.6 to 55.4)	(0.6 to 4.2)	(54.4 to 55.8)	(54.0 to 55.7)	(-1.3 to 0.9)	(0.5 to 4.7)		(0.6 to 4.8)
Thyroid carcinoma	95.0	96. 5	1.5	95.8	96.5	0.7	0.7	.45	0.8	.58
	(93.8 to 96.2)	(95.3 to 97.6)	(-0.2 to 3.1)	(95.1 to 96.4)	(95.8 to 97.2)	(-0.3 to 1.7)	(-1.2 to 2.6)		(-2.1 to 3.8)
Germ cell and trophoblastic neoplasms of gonads	63.3	62.5	−0.8	69.7	69.1	−0.6	−0.2	.94	0.1	.98
	(60.8 to 65.8)	(59.4 to 65.6)	(-4.8 to 3.2)	(67.8 to 71.7)	(66.8 to 71.4)	(-3.6 to 2.4)	(-5.1 to 4.8)		(-4.8 to 5.0)
Melanoma	75.2	73.0	−2.2	76.4	74.1	−2.3	0.1	.96	−1.8	.49
	(72.4 to 78.0)	(69.4 to 76.6)	(-6.7 to 2.4)	(74.7 to 78.1)	(72.0 to 76.2)	(-5.0 to 0.4)	(-5.2 to 5.4)		(-7.0 to 3.3)
Breast carcinoma in women	26.4	18.4	−8.0	23.2	21.7	−1.5	−6.5	.16	−5.3	.28
	(20.5 to 32.2)	(11.9 to 24.9)	(-16.7 to 0.8)	(21.6 to 24.8)	(19.9 to 23.5)	(-3.9 to 0.9)	(-15.6 to 2.6)		(-14.8 to 4.3)
Hodgkin’s lymphoma	12.1	15.3	3.2	13.8	12.8	−0.9	4.1	.07	4.0	.28
	(10.2 to 14.1)	(12.8 to 17.9)	(-0.0 to 6.4)	(11.8 to 15.7)	(10.5 to 15.1)	(-3.9 to 2.1)	(-0.3 to 8.6)		(-3.3 to 11.3)
Non-Hodgkin’s lymphoma	33.6	32.5	−1.1	34.2	32.1	−2.1	1.0	.79	0.9	.80
	(29.9 to 37.3)	(27.9 to 37.1)	(-7.0 to 4.8)	(31.6 to 36.7)	(28.9 to 35.2)	(-6.1 to 2.0)	(-6.2 to 8.1)		(-6.3 to 8.2)
Carcinoma of cervix	58.3	75.0	16.7	67.3	61.0	−6.3	23.0	<.001	21.2	<.001
	(51.0 to 65.6)	(66.8 to 83.2)	(5.8 to 27.7)	(64.7 to 69.9)	(57.6 to 64.4)	(-10.6 to -2.1)	(11.3 to 34.8)		(9.6 to 32.7)
Carcinoma of colon and rectum	15.2	15.9	0.8	12.7	14.6	1.9	−1.1	.80	−0.3	.95
	(10.1 to 20.3)	(9.8 to 22.1)	(-7.2 to 8.7)	(10.5 to 14.9)	(12.0 to 17.2)	(-1.5 to 5.3)	(-9.8 to 7.5)		(-11.3 to 10.6)
Soft tissue sarcomas (excluding Kaposi sarcoma)	15.6	20.3	4.8	20.3	23.2	2.9	1.9	.70	1.9	.72
	(11.4 to 19.8)	(14.6 to 26.1)	(-2.4 to 11.9)	(16.6 to 23.9)	(18.5 to 27.8)	(-3.0 to 8.8)	(-7.4 to 11.1)		(-8.2 to 12.0)
Carcinoma of kidney	47.7	55.6	7.8	75.6	76.8	1.2	6. 7	.41	5.1	.54
	(37.3 to 58.2)	(44.3 to 66.4)	(-7.2 to 22.9)	(72.0 to 79.2)	(72.7 to 80.8)	(-4.3 to 6.6)	(-9.3 to 22.7)		(-11.3 to 21.4)
Carcinoma of head and neck	30.6	29.5	−1.2	35.4	31.0	−4.4	3.2	.63	1.6	.81
	(23.5 to 37.8)	(21.0 to 37.9)	(-12.2 to 9.9)	(31.0 to 39. 8)	(25.8 to 36.2)	(-11.2 to 2.5)	(-9.8 to 16.2)		(-11.2 to 14.4)
Osseous and chondromatous neoplasms	12.4	18.7	6.3	26.7	18.9	−7.8	14.1	.02	14.4	.05
	(8.1 to 16.7)	(11.8 to 25.6)	(-1.8 to 14.4)	(20.4 to 33.1)	(12.3 to 25.6)	(-17.0 to 1.4)	(1.8 to 26.4)		(0.3 to 28.5)
Carcinoma of gonads	42.5	48.2	5.7	51.4	62.6	11.3	−5.5	.57	−9.0	.50
	(31.7 to 53.3)	(35.1 to 61.3)	(-11.3 to 22.7)	(45.2 to 57.5)	(55.8 to 69.5)	(2.0 to 20.5)	(-24.9 to 13.8)		(-35.2 to 17.2)
Carcinoma of stomach	4.26	3.7	−0.6	9.7	12.3	2.6	−3.1	.59	−1.1	.94
	(-1.5 to 10.1)	(-3.4 to 10.8)	(-9.7 to 8.6)	(5.6 to 13.8)	(7.2 to 17.3)	(-4.0 to 9.1)	(-14.4 to 8.1)		(-31.5 to 29.3)
Carcinoma of trachea, bronchus, and lung	27.0	23.5	−3.5	14.3	17.6	3.3	−6.8	.54	−5.3	.66
	(12.7 to 41.3)	(9.3 to 37.8)	(-23.7 to 16.7)	(9.6 to 19.0)	(11.0 to 24.1)	(-4.8 to 11.3)	(-28.5 to 15.0)		(-28.6 to 18.0)
Other	45.2	47.4	2.2	51.1	2.6	53.6	−0.4	.94	0.4	.74
	(39.1 to 51.2)	(40.3 to 54.5)	(-7.1 to 11.5)	(48.2 to 54.0)	(-1.8 to 6.9)	(50.4 to 56.9)	(-10.7 to 9.9)		(-9.8 to 10.6)

*Early stage was defined as stage I (stage 0 and I for bladder cancer), coded using American Joint Commission on Cancer, 6th Edition. ACA = Affordable Care Act; CI = confidence interval; DID = difference-in-differences.

† P values are for two-sided Wald Chi-square tests.

‡Models were adjusted for sex, race and ethnicity, absolute age at diagnosis, residence, marital status at diagnosis, and county % with less than high school education.

Increasing detection of cancers at early stage with concomitant decreases in late-stage diseases, as suggested by this study, is the first step toward improved survival and quality of life. Receipt of timely and high-quality treatment and follow-up care and mitigating the financial burden of cancer among survivors are other areas that can potentially be addressed by this and other provisions of the ACA ( 18 ). Future studies should monitor changes in these outcomes among young cancer patients and survivors.

There have been concerns about unintended consequences of the ACA-DCE, such as increases in premiums of parents’ insurance and cost of parents’ employers, fewer employers offering insurance coverage to young adult employees ( 19 ), and privacy leakage to parents as the policyholders ( 20 ). Future research should explore these questions, and policy adjustment may be necessary if the concerns hold true.

A strength of our study is that it documents a shift toward early-stage disease for most cancers, though statistically significant for only carcinoma of the cervix and neoplasms of osseous and chondromatous, among young adults following the ACA-DCE using a population-based dataset. Our study is limited by the relatively short time post the ACA-DCE (2 years); however, a previous study among adolescent and young adult patients with common cancer types showed that the lag time from onset of cancer-specific symptoms to diagnosis rarely exceeded one year ( 21 ). Another limitation is the underlying difference between people age 26 to 34 years and people age 19 to 25 years other than the policy change. In particular, the study period overlaps with the economic recession, which affected employment and thus employer-sponsored insurance. However, the employment rate change in those age 46 to 54 years, who were most likely to be the parents of the people age 19 to 25 years, was very similar to that in those age 26 to 34 years, suggesting that our main findings were unlikely to be influenced by changes in employment associated with the economic recession ( 22 ).

In summary, this early observation suggests a positive effect of the ACA-DCE on cancer stage at diagnosis among the targeted young adults. Moving forward, research is needed to continue to monitor the effects of the ACA on cancer care and outcomes among young adults.

Funding

This work was supported by the Intramural Research Department of the American Cancer Society.

Notes

The American Cancer Society had no role in the design or conduct of the study; the collection, management, analysis, or interpretation of the data; the preparation, review, or approval of the manuscript; or the decision to submit the manuscript for publication.