Health insurance coverage and cost barriers to needed medical care among U.S. adult cancer survivors age <65 years

Authors

  • Susan A. Sabatino M.D., M.P.H.,

    Corresponding author
    1. Division of Cancer Prevention and Control, National Centers for Chronic Disease Prevention and Health Promotion, Centers for Disease Control and Prevention, Atlanta, Georgia
    • Centers for Disease Control and Prevention, Division of Cancer Prevention and Control, 4770 Buford Highway (K-52), Atlanta, GA 30341===

    Search for more papers by this author
    • Fax: (770) 488-4760

  • Ralph J. Coates Ph.D.,

    1. Division of Cancer Prevention and Control, National Centers for Chronic Disease Prevention and Health Promotion, Centers for Disease Control and Prevention, Atlanta, Georgia
    Search for more papers by this author
  • Robert J. Uhler M.A.,

    1. Division of Cancer Prevention and Control, National Centers for Chronic Disease Prevention and Health Promotion, Centers for Disease Control and Prevention, Atlanta, Georgia
    Search for more papers by this author
  • Linda G. Alley Ph.D., R.N.,

    1. Division of Cancer Prevention and Control, National Centers for Chronic Disease Prevention and Health Promotion, Centers for Disease Control and Prevention, Atlanta, Georgia
    Search for more papers by this author
  • Lori A. Pollack M.D., M.P.H.

    1. Division of Cancer Prevention and Control, National Centers for Chronic Disease Prevention and Health Promotion, Centers for Disease Control and Prevention, Atlanta, Georgia
    Search for more papers by this author

  • The findings and conclusions in this report are those of the authors and do not necessarily represent the views of the funding agency.

Abstract

BACKGROUND

The health insurance and cost barriers to care among cancer survivors age <65 years were examined.

METHODS

Using the 1998 and 2000 National Health Interview Survey, survivors ages 18 to 64 years (n = 1718) were compared with similarly aged adults without cancer (n = 50,276) to examine health insurance and reported delayed/missed needed medical care within the previous year because of cost. Findings were initially adjusted for age, sex, race, and ethnicity, and further adjusted for employment, income, health status, marital status, and region.

RESULTS

Before adjustment, survivors were less likely to be uninsured (12.4% vs. 18.0%) and more likely to have public insurance (11.2% vs. 6.2%). After initial adjustment, survivors were as likely to lack insurance, less likely to have private insurance, and more likely to have public insurance. After further adjusting, differences in being uninsured were found to be small, differences in having private insurance were eliminated, and differences in having public insurance were reduced. Survivors most likely to lack insurance were younger, female, African-American, or lower income. Survivors, particularly uninsured or publicly insured survivors, were more likely to delay/miss care because of cost. Overall, 20.9% of survivors, including 68% of uninsured survivors, reported delaying/missing needed care.

CONCLUSIONS

Health insurance coverage among cancer survivors age <65 years appears to be comparable to that of adults of similar age, sex, race, and ethnicity, but survivors may more likely be publicly insured. Differences are attributable in part to employment, income, and health status, factors potentially influenced by cancer. Unmet medical care needs because of cost were common among survivors, particularly uninsured survivors. Cancer 2006. © 2006 American Cancer Society.

Cancer survivors are individuals who have received a cancer diagnosis at any time during their life.1 More than 10 million U.S. residents are currently cancer survivors,2 a growing segment of the population. The population is aging, and cancer risk increases with age.3 New technologies increase the ability to detect cancers and to detect cancers earlier, leading to increased years of life with a known cancer.4 For some cancers, earlier detection allows earlier treatment when therapies may be more effective.5 The development of more effective therapies also may improve survival for some cancer patients.6 Approximately 1,372,910 new cancer diagnoses are expected in 2005.7 Recent estimates suggest that 64% of cancer patients survive >5 years after diagnosis.8 Approximately 40% of survivors are estimated to be age <65 years.8

In the national public health agenda, an increasing emphasis has been placed on identifying and addressing the needs of this expanding population.1, 9 The National Action Plan for Cancer Survivorship1 notes the importance of healthcare access in providing quality care to survivors and emphasizes the need for information regarding access to care for this group. Health insurance among survivors has been associated with aggressiveness or appropriateness of care,10, 11 treatment delays,12 and mortality.13, 14 Furthermore, insurance may affect long-term follow-up and care. With a growing population of survivors and increasing survival for some cancers, access to care among survivors is likely to become increasingly important.

Many survivors may be uninsured,15–18 although whether survivors have higher insurance coverage rates than adults without cancer is uncertain.15, 17, 19 It is also unclear to what extent higher coverage rates among survivors in some studies may reflect their older age and higher rates of Medicare enrollment.20 Aside from studies of childhood cancer survivors,19, 21 to our knowledge only a few studies to date18 have focused on health insurance issues among survivors age <65 years, who are largely not eligible for Medicare. One study suggested that this younger group may have higher rates of being uninsured than survivors of all ages.18

We therefore sought to examine health insurance coverage among cancer survivors age <65 years compared with adults without cancer to examine insurance among different groups of survivors and to describe survivors' reasons for being uninsured. Furthermore, uninsured adults with chronic conditions are more likely than their insured counterparts to have unmet medical care needs because of cost,22, 23 which could contribute to adverse health outcomes. Therefore, we also sought to examine whether cost barriers to receiving needed care existed among survivors, and if so, to describe the extent of these barriers among survivors with different health insurance coverage.

MATERIALS AND METHODS

Data Source

We used data from the National Health Interview Survey (NHIS), a nationally representative sample of the civilian, noninstitutionalized U.S. population.24, 25 The NHIS is an annual survey administered by the National Center for Health Statistics through in-person interviews. We used information from the Sample Adult Care, for which one “sample adult” was selected randomly from each family to provide self-reported information, and from the Family Care, for which all adults present may respond for themselves. We considered information regarding sociodemographics, cancer, healthcare access, health insurance, income, and health status. To obtain an adequate sample size, we combined data from the 1998 and 2000 NHIS. During 1998, 32,440 sample adults were interviewed (for an overall response rate of 73.9%), and during 2000, 32,374 sample adults were interviewed (for an overall response rate of 72.1%).

Sample

We studied respondents ages 18 to 64 years, excluding 243 who were missing health insurance information. Of the remainder, we identified 1718 respondents who reported a history of cancer other than nonmelanoma skin cancer (cancer survivors). Our comparison group included similarly aged adults reporting no prior cancer history or only nonmelanoma skin cancer (n = 50,276).

Outcomes

Health insurance

We categorized health insurance according to coverage at the time of interview (private insurance with or without public insurance, public insurance only, no insurance). Military insurance was categorized with private insurance. Of the 37,251 respondents in this category, 524 also reported public insurance. Public insurance included Medicaid, Medicare, or other government-sponsored or public-sponsored insurance. Other respondents were considered uninsured, including those reporting only Indian Health Service coverage, consistent with NHIS.24 We did not have information regarding who the policyholder was. We also examined reported reasons for being uninsured. In NHIS, uninsured respondents were asked for reasons they stopped being covered or did not have insurance.

Delayed/missed care due to cost

We evaluated reported delayed or missed medical care within the prior year due to cost. Respondents were asked whether during the previous year they delayed getting medical care because of worry about cost, needed medical care but did not get it because they could not afford it, or needed prescription medicines but did not get them because they could not afford them.

Other Independent Variables

Other factors considered included age, sex, race, ethnicity, employment, income, health status, marital status, and geographic region.24, 25 Because of small numbers of survivors in some groups, we categorized race into white, African-American, and other races. Respondents who reported not having a job during the prior week were considered unemployed. Missing data regarding income were imputed.24 For survivors, we examined the most commonly reported cancers (Table 1), age at diagnosis, and time since diagnosis (age at interview minus age at diagnosis). Survivors with multiple cancers were categorized by first diagnosis. Survivors with multiple cancers diagnosed in the same year were categorized separately.

Table 1. Sociodemographic and Health Characteristics of Adults Age < 65 Years by Cancer History
 Cancer Survivors (n = 1718)No Cancer History (n = 50,276)
No.%* (95% CI)No.%* (95% CI)
  • 95% CI: 95% confidence interval.

  • *

    Weighted, poststratified to U.S. Census.

  • Missing income values imputed by National Center for Health Statistics.

  • Survivors with multiple types of cancer initially diagnosed in the same year.

Age, y    
18–3425815.0 (13.1–17.2)19,12539.1 (38.4–39.7)
35–5059934.5 (31.9–37.2)19,29838.3 (37.7–38.9)
50–6486150.5 (47.8–53.1)11,85322.6 (22.2–23.1)
Sex    
Male46530.7 (28.4–33.1)22,81749.6 (49.1–50.2)
Female125369.3 (66.9–71.6)27,45950.4 (49.8–50.9)
Race    
White147289.3 (87.6–90.8)38,58180.7 (80.1–81.3)
African-American1646.7 (5.7–7.9)743012.1 (11.6–12.7)
All other races814.0 (3.0–5.2)41937.2 (6.9–7.5)
Ethnicity    
Hispanic1384.5 (3.6–5.5)925811.5 (10.9–12.0)
Non-Hispanic157995.5 (94.5–96.4)40,99988.5 (88.0–89.1)
Region    
Northeast27817.7 (15.8–19.7)939619.2 (18.6–19.8)
Midwest41225.3 (23.0–27.7)11,61125.6 (24.9–26.4)
South66739.2 (36.4–41.9)17,92235.8 (34.9–36.6)
West36117.9 (15.7–20.3)11,34719.4 (18.6–20.1)
Marital status    
Married99668.6 (66.3–70.8)27,40762.0 (61.2–62.6)
Unmarried72031.4 (29.2–33.7)22,61638.0 (37.4–38.7)
Employment    
Yes105362.7 (60.1–65.2)37,84576.9 (76.4–77.3)
No65737.3 (34.8–39.9)12,17823.1 (22.7–23.6)
Income    
<$20,00046220.2 (18.1–22.4)12,97018.6 (18.0–19.2)
$20,000–$34,99937319.9 (17.8–22.1)10,58818.8 (18.2–19.3)
$35,000–$64,99945228.7 (26.1–31.4)14,44631.1 (30.5–31.7)
≥$65,00043131.3 (28.7–34.1)12,27331.5 (30.8–32.3)
Health status    
Excellent/very good/good122173.4 (70.9–75.7)45,54991.9 (91.5–92.2)
Fair/poor49326.6 (24.3–29.1)46918.1 (7.8–8.5)
Cancer type    
Cervix31717.5 (15.6–19.6)  
Breast27215.5 (13.7–17.6)  
Uterine1608.6 (7.3–10.1)  
Melanoma1117.2 (5.9–8.8)  
Colorectal815.0 (3.9–6.4)  
Prostate583.7 (2.8–4.8)  
Lymphoma523.2 (2.4–4.4)  
Thyroid482.9 (2.2–3.9)  
Ovarian552.7 (2.1–3.6)  
Multiple sites502.7 (2.0–3.7)  
Other51431.0 (28.3–33.7)  
Age at diagnosis, y    
<18835.4 (4.2–6.8)  
18–3463135.0 (32.7–37.4)  
35–4955533.1 (30.7–35.6)  
50–6443926.5 (24.3–28.7)  
Time since diagnosis, y    
0–248428.9 (26.5–31.4)  
3–533820.0 (17.8–22.4)  
6–1031618.6 (16.6–20.9)  
>1057032.4 (30.1–34.9)  

Statistical Analysis

We used SUDAAN software (version 9.0; Research Triangle Institute, Research Triangle Park, NC) for all analyses to account for the complex sample design.24, 25 Sample weights were halved to reflect the combination of 2 years of data.24, 26 Descriptive statistics and bivariable analyses are presented to show the sample characteristics and to compare insurance between survivors and the comparison group. Multinomial logistic regression modeling was used to examine differences in health insurance between groups after adjusting for nonmodifiable demographic characteristics (age, sex, race, and ethnicity). Results are presented as adjusted percentages (predicted marginals). We used general linear contrasts to test differences in adjusted percentages between groups. Because employment, income, health status, and possibly marital status and region of residence could be influenced as a result of cancer diagnosis, they were not initially included in the model. This was to avoid controlling for factors that may differ between survivors and other adults as a consequence of cancer history and therefore may mediate differences in outcomes between survivors and the comparison group.27 To explore the extent of differences that may be attributable to these potential mediators, we further adjusted our model for these factors.

Differences in coverage among groups of survivors were examined according to cancer-related and general population characteristics. For uninsured survivors and adults without cancer, we examined reported reasons for lacking insurance after adjusting for age. We also examined differences between survivors and other adults in reporting delayed or missed medical care due to cost. For this analysis, we used multinomial logistic regression models to compare differences in delayed or missed care between survivors and other adults within each category of health insurance after adjusting for age, sex, race, and ethnicity. We used a separate model for each type of delayed or missed care.

RESULTS

Of respondents studied, 3.2% were cancer survivors, representing an estimated 5.35 million survivors nationally ages 18 to 64 years (Table 1). As expected, survivors were older than adults without cancer. Survivors were also more likely to be female, white, non-Hispanic, unemployed, in fair or poor health, and married. Cervical, breast, and uterine cancers were reported most often. Few survivors reported childhood cancers, and approximately half reported diagnosis within 5 years.

Based on NHIS, we estimate that 662,861 cancer survivors nationwide had no health insurance. Survivors were less likely to report being uninsured than other adults (12.4% vs. 18.0%), but were more likely to have public insurance (11.2% vs. 6.2%) in unadjusted analyses (Table 2). We observed no substantial difference in private insurance coverage by cancer history. Among all respondents studied, people who were older, female, white, and non-Hispanic were less likely to be uninsured and more likely to have private insurance (except females), compared with younger, male, nonwhite, and Hispanic people, respectively. In contrast, those from the South or West, those who were unemployed, those who were unmarried, and those with lower incomes were more likely to be uninsured and less likely to have private insurance than their counterparts. People who were unemployed, unmarried, and with lower incomes were also more likely to be publicly insured (data not shown).

Table 2. Health Insurance Coverage According to Cancer History
 Health Insurance Coverage
UninsuredPublic OnlyPrivate
%* (95% CI)%* (95% CI)%* (95% CI)
  • 95% CI: 95% confidence interval.

  • *

    Weighted, poststratified to the U.S. Census.

  • Adjusted percentages are predicted marginals resulting from a multinomial logistic regression model adjusting for age, sex, race, and ethnicity.

  • P<.005 (general linear contrast testing differences between survivors and adults without cancer history within insurance category).

  • §

    Adjusted for age, sex, race, ethnicity, employment, income, health status, marital status, and region.

  • ll

    P<.05.

Cancer history   
Unadjusted   
 Cancer survivors12.4 (10.8–14.2)11.2 (9.6–13.1)76.4 (74.0–78.6)
 No cancer history18.0 (17.6–18.5)6.2 (5.9–6.5)75.8 (75.2–76.3)
Model 1   
 Cancer survivors17.2 (15.2–19.5)11.2 (9.6–13.1)c71.5 (69.0–74.0)
 No cancer history17.8 (17.4–18.3)6.2 (5.9–6.5)76.0 (75.4–76.5)
Model 2§   
 Cancer survivors15.6 (13.7–17.7)ll7.4 (6.3–8.6)77.0 (74.9–79.1)
 No cancer history17.8 (17.3–18.3)6.3 (6.0–6.6)76.0 (75.4–76.5)

After adjusting for age, sex, race, and ethnicity, survivors were as likely as other adults to report being uninsured, remained more likely to report public insurance, and became somewhat less likely to report private insurance. When employment, income, health status, marital status, and region were added to the model, survivors were somewhat less likely to be uninsured, marginally more likely to have public insurance, and as likely to be privately insured.

Ovarian, cervical, or uterine cancer survivors were more likely than other survivors to report being uninsured, as were survivors diagnosed at younger ages (Table 3). Survivors who were younger, female, African-American, and of lower income were also more likely to report being uninsured (income not shown).

Table 3. Health Insurance Coverage by Cancer History and Sociodemographics Among Survivors (n = 1718)
Health Insurance
 Uninsured %* (95% CI)Public only %* (95% CI)Private %* (95% CI)
  • 95% CI: 95% confidence interval.

  • *

    Weighted, poststratified to U.S. Census.

Cancer type   
 Cervix20.3 (15.9–25.4)12.4 (8.7–17.3)67.4 (61.4–72.8)
 Breast6.3 (3.8–10.5)5.8 (3.6–9.3)87.8 (83.3–91.2)
 Uterine20.1 (13.4–29.0)20.9 (14.8–28.8)59.0 (49.5–67.8)
 Melanoma4.3 (1.6–10.9)5.2 (1.9–13.4)90.5 (81.8–95.2)
 Colorectal9.7 (4.4–20.0)16.1 (8.8–27.6)74.2 (60.1–84.6)
 Prostate5.2 (2.0–13.2)7.9 (3.6–16.7)86.8 (76.7–93.0)
 Ovarian33.0 (20.0–49.2)6.1 (2.5–14.4)60.9 (45.2–74.6)
Age at diagnosis, y   
 <1819.4 (11.1–31.6)14.4 (8.8–22.8)66.2 (54.6–76.2)
 18–3417.4 (14.3–21.1)12.6 (9.9–15.8)70.0 (65.7–73.9)
 35–4911.0 (8.6–13.9)10.2 (7.6–13.6)78.8 (74.7–82.3)
 50–645.2 (3.4–7.9)9.7 (7.3–12.9)85.0 (81.3–88.2)
Time since diagnosis, y   
 0–211.1 (8.3–14.6)12.7 (9.8–16.2)76.3 (71.9–80.2)
 3–515.0 (11.3–19.8)9.2 (6.6–12.7)75.8 (70.5–80.4)
 5–1012.0 (8.7–16.1)10.5 (7.4–14.8)77.5 (72.0–82.2)
 >1011.5 (8.9–14.8)11.4 (8.9–14.4)77.1 (73.1–80.7)
Age at interview, y   
 18–3425.3 (19.4–32.1)14.0 (10.0–19.3)60.7 (53.7–67.4)
 35–5012.4 (9.7–15.6)10.9 (8.3–14.1)76.8 (72.6–80.4)
 50–648.6 (6.8–10.8)10.7 (8.8–12.8)80.8 (78.0–83.3)
Sex   
 Male8.6 (6.2–11.9)11.3 (8.2–15.3)80.1 (75.2–84.3)
 Female14.1 (12.1–16.3)11.2 (8.2–15.3)74.7 (71.2–77.3)
Race   
 White12.1 (10.3–14.0)9.6 (8.0–11.5)78.4 (75.8–80.7)
 African-American19.0 (12.4–27.9)24.9 (18.4–32.6)56.2 (48.1–63.9)
 All other races9.0 (4.7–16.5)25.1 (15.7–37.7)65.9 (52.6–77.1)
Ethnicity   
 Hispanic13.2 (8.4–20.1)18.1 (12.4–25.8)68.7 (59.7–76.5)
 Non-Hispanic12.4 (10.7–14.3)10.9 (9.3–12.8)76.7 (74.3–79.0)

After age-adjusting, approximately 50% of uninsured survivors reported they were uninsured because of cost (Table 4), a finding similar to that of adults without cancer. Losing or changing jobs by the policyholder or working for employers not offering insurance were also common reasons in both groups; however, survivors were more likely than other adults to report being uninsured because they lost Medicaid or medical plan coverage.

Table 4. Most Commonly Reported Reasons* for Not Having Health Insurance According to Cancer History
 Cancer Survivors (n = 200)No Cancer History (n = 9272) 
Reason% (95% CI)% (95% CI)P
  • 95% CI: 95% confidence interval.

  • *

    Not mutually exclusive.

  • Age-adjusted. Weighted, poststratified to U.S. Census.

  • General linear contrast testing for differences between survivors and adults without cancer history.

Cost is too high49.7 (40.5–58.9)51.7 (50.3–53.2).66
Policy holder lost job or changed employers32.1 (24.6–40.7)28.6 (27.4–29.8).40
Got divorced or separated/death of spouse or parent7.0 (4.1–11.5)4.1 (3.6–4.6).11
Became ineligible because of age/left school6.8 (3.0–14.4)7.7 (7.0–8.3).74
Employer not offer coverage/not eligible for coverage13.5 (8.7–20.4)15.1 (14.2–16.0).60
Insurance company refused coverage1.8 (0.6–5.4)1.2 (1.0–1.6).55
Lost Medicaid (for any reason)14.3 (9.6–20.7)7.8 (7.2–8.5).02
Medicaid/medical plan stopped after pregnancy ended2.4 (0.9–6.2)2.3 (2.0–2.6).94
Lost Medicaid/medical plan because of new job or increased income5.1 (2.5–10.0)2.8 (2.4–3.2).21
Lost Medicaid for other reason7.7 (4.4–13.2)2.8 (2.5–3.2).03

Greater than 20% of survivors reported delaying or missing needed care within the prior year because of cost (Table 5). These unmet medical care needs were greatest among uninsured survivors, with 42.8% of this group reporting that they did not obtain needed prescription medicines because of cost and 67.8% reporting any delayed or missed care due to cost. Among insured survivors, those publicly insured were more likely to report cost barriers than those privately insured. A similar relation between health insurance and cost barriers was observed in the comparison group. However survivors, particularly uninsured or publicly insured survivors, were more likely than similarly insured adults without cancer to report delayed or missed care due to cost. When adjusted for age, sex, race, and ethnicity, the results were similar, with slight attenuation among survivors, mostly among uninsured survivors (50.2% [95% confidence interval (95% CI), 42.7-57.6] for delays, 41.8% [95% CI, 34.3-49.7] for not getting medical care, 38.9% [95% CI, 31.2-47.1] for not getting prescriptions, and 64.5% [95% CI, 56.1-72.2] for any delay or missed care) and small increases among uninsured adults without cancer.

Table 5. Reported Delays in or Missed Needed Medical Care According to Health Insurance and Cancer History
 Delayed Medical Care Due To Worry About CostDid Not Get Needed Medical Care Due to CostDid Not Get Needed Prescription Medicines Due to CostAny Reported Delay or Missed Care Due to Cost/Worry About Cost
Cancer Survivors %* (95% CI)Adults Without Cancer %* (95% CI)Cancer Survivors %* (95% CI)Adults Without Cancer %* (95% CI)Cancer Survivors %* (95% CI)Adults Without Cancer %* (95% CI)Cancer Survivors %* (95% CI)Adults Without Cancer %* (95% CI)
  • 95% CI: 95% confidence interval.

  • *

    Weighted, poststratified to U.S. Census.

Total15.2 (13.4–17.2)8.6 (8.3–9.0)10.8 (9.3–12.5)5.9 (5.7–6.2)11.9 (10.2–13.9)5.8 (5.5–6.1)20.9 (18.6–23.3)12.3 (11.9–12.7)
Uninsured53.9 (46.3–61.4)23.4 (22.3–24.5)45.1 (37.4–53.1)18.7 (17.8–19.7)42.8 (34.7–51.3)15.8 (14.9–16.7)67.8 (59.5–75.0)32.0 (30.8–33.2)
Publicly insured20.3 (15.4–26.2)11.4 (10.3–12.7)16.7 (12.4–22.1)9.8 (8.8–11.0)23.2 (17.8–29.7)12.7 (11.5–14.0)33.9 (27.8–40.7)20.2 (18.7–21.8)
Private/military insurance8.2 (6.6–10.0)4.9 (4.6–5.2)4.4 (3.2–5.8)2.6 (2.4–2.7)5.2 (3.9–6.9)2.9 (2.7–3.1)11.3 (9.4–13.6)7.0 (6.7–7.3)

DISCUSSION

To our knowledge, the current study is the first to use nationally representative data to compare health insurance coverage among cancer survivors in this age range with a group from the general population. The findings suggest that many survivors in this age group lack insurance, with an estimated 660,000 of these survivors uninsured nationwide. This may be an underestimation because cancer history may be underreported in NHIS.28 A comparison of 1992 NHIS cancer history estimates with national estimates based on Connecticut Tumor Registry data found that NHIS estimates were 3% lower for women and 20% lower for men, possibly due to self-report or exclusion of institutionalized individuals in NHIS, or refusals to participate due to poor health. Any underreporting in the current study would likely bias towards the null, leading to possibly underestimated differences between survivors and other adults.

To our knowledge, in studies of survivors without comparison groups or including survivors of all ages, few national estimates are available regarding lack of health insurance among survivors in this age group, and the findings of the current study are in agreement with those estimates. Using Medical Expenditure Panel Survey data, Thorpe and Howard18 estimated that 11% of survivors age <65 years lacked insurance primarily near diagnosis. Studies using NHIS data have estimated that 5% to 13% of survivors of all ages are uninsured,15, 17, 20 a finding that is similar to our results (12.4%). Whether being uninsured is less common among survivors than in the general population has been uncertain. Whereas some evidence suggests that survivors of all ages are less likely than adults without cancer to report being uninsured,17, 20 1 study found no significant difference in coverage between survivors and adults of all ages among groups that were comparable with regard to age, sex, and education.15 The current study findings, combined with those of other studies, indicate that although overall survivors are less likely to be uninsured, among groups of comparable age, sex, race, and ethnicity, survivors are as likely to be uninsured as other adults and have a comparable need for health insurance coverage.

The results of the current study also suggest that survivors may be more likely than other adults to have public insurance. Furthermore, the proportion of survivors with public insurance in the current study is lower than other estimates. Yabroff et al.15 estimated that 20% of survivors had public insurance, similar to adults without cancer of similar age, sex, and education. However, Yabroff et al. included survivors of all ages, and the higher proportion of publicly insured survivors in that study compared with ours could reflect higher Medicare enrollment. Similarly, Thorpe and Howard18 reported that 19% of survivors age <65 had public insurance. It is unclear to what extent differences between this estimate and ours may reflect changes in insurance over time, differences in defining public insurance, or differences in study samples.

When we included employment, income, health status, marital status, and region in the model, differences in private insurance were eliminated and differences in public insurance were substantially reduced. Furthermore, the difference between survivors and adults without cancer in being uninsured was small. Many of these changes were due to employment, income, and health status, factors that could be influenced by cancer diagnosis.

For some survivors, the relations between cancer history and insurance status could be mediated through changes in employment and income, which could be consistent with our findings. Survivors may lose their jobs, be unable to work, or choose not to work due to poor health or other factors related to cancer. Although many survivors continue or return to work after diagnosis,29 evidence suggests that survivors are less likely to have worked in the prior month15 and more likely be unable to work15, 17 or limited in work due to health than adults without cancer, with differences persisting over time.15 Furthermore, nationally, 16% of working age survivors report being unable to work because of their health, with an additional 7% limited in work (compared with 5% and 3% of the general population, respectively).17 Given that most adults age <65 years old receive health insurance through an employer,13 shifts in employment among survivors or caregivers could result in loss of private insurance,13 which could be reflected in the lower rates of private insurance among survivors in the current study before adjusting for employment.

In addition, survivors without private coverage may more likely meet eligibility criteria for Medicaid or public health insurance than some adults without cancer. Factors considered in determining eligibility often include age, income and resources, medical needs, and disability.30 It is possible that the increased public health insurance coverage of survivors may reflect reduced income, reduced resources, increased medical needs, or disability related to cancer. Consistent with this, after adjusting for employment, income, and health status, differences between survivors and adults without cancer were substantially reduced. Relocation or change in marital status after cancer diagnosis could also affect health insurance coverage (e.g., move to a state with different Medicaid eligibility criteria, or lose a spouse who was the policyholder). However, because NHIS data are cross-sectional, we cannot determine if cancer diagnosis led to changes in these factors. More research using longitudinal data would be needed to examine these relations.

Our finding that being uninsured is a particular problem for certain groups of survivors is consistent with a small body of literature. Other studies have concluded that survivors who were younger,19 had cervical cancer,18 lower education,31 and lower income18 were less likely to be insured. Moreover, similar to Thorpe and Howard,18 we found that female survivors in this age group were less likely to be insured than males, although in a study of pediatric bone tumor survivors, women more often reported having insurance.31 This may reflect differences in age18 within this age group, given differences in age at diagnosis for sex-specific cancers. Others have found, as we did, that a greater proportion of Hispanic and African-American survivors were uninsured or more likely to have some form of public insurance than other survivors.16, 18, 32 These differences are consistent with differences in the general population13, 14, 33, 34 that may contribute to disparities in health care,14, 34 including cancer care.14

To our knowledge, few studies to date have examined reasons survivors report being uninsured. In a review of employment and insurance issues among adult survivors of childhood cancers, cost was a common reason for both survivors and adults without cancer,19 similar to our findings, and findings in the general population.13 Furthermore, we also found that many uninsured survivors cited job-related reasons. Although most survivors can continue working or resume working without limitations,29 previous studies have reported changes or limitations in employment among survivors.15, 17, 21, 35 However, in the current study, uninsured adults in the comparison group similarly cited these reasons. Previous findings also suggest that survivors sometimes are denied or lose health insurance.16, 19, 21, 28, 35, 36 We found that, compared with other uninsured adults, survivors were marginally more likely to attribute being uninsured to refused coverage, although this was not commonly reported and differences were not significant. However, survivors were somewhat more likely to report losing Medicaid as a reason that they were uninsured. We are not aware of previous studies reporting this finding, and did not have enough information to determine what factors were driving it. It is possible that increased Medicaid eligibility and coverage after a cancer diagnosis is transient, and as adults with cancer survive longer after diagnosis, changes in employment, resources, and medical needs lead to loss of Medicaid eligibility. Although not significant, survivors were more likely to report loss of Medicaid due to a new job or increased income, possibly consistent with this hypothesis. Further research is needed to examine these relations in more detail.

Our findings also provide important information regarding the relationship between insurance coverage and unmet medical care needs due to cost among cancer survivors. The unfavorable medical consequences of lacking health insurance for the general population and for those with other chronic diseases have been described by others,13, 22, 23, 37 including unmet care needs because of cost. Our results indicate that cost barriers to receiving care are prevalent among cancer survivors as well, particularly for uninsured or publicly insured survivors, with approximately two-thirds of uninsured survivors reporting recently delayed or missed care because of cost. Such delays may lead to missed opportunities to address medical problems or risk factors early, and interfere with the timely receipt of care.13 Furthermore, >40% of uninsured survivors reported not receiving needed prescription medications for financial reasons, perhaps because uninsured people in general pay for almost all of the cost of prescription medications.13 However, we were unable to determine to what extent medications not obtained by survivors were prescribed for cancer therapy. It may not be surprising that survivors were more likely to report cost barriers than other adults, given that they may have more medical care needs and costs,16, 18, 21, 28, 35, 36 and therefore more opportunities to miss needed care. However, this analysis reveals the extent of unmet medical care needs due to cost among survivors, and expands the evidence regarding the importance of health insurance coverage in providing access to affordable and timely care for this group.20

The current study results suggest that health care providers and policy makers should recognize that costs may pose considerable barriers for many cancer survivors. Moreover, these findings suggest a possible need to expand health insurance coverage and benefits among survivors to reduce cost barriers. Such changes may prove particularly beneficial for cancer survivors, who are both more likely to miss needed care and more likely to develop recurrence or conditions requiring care.17, 38, 39

Our findings must be interpreted in light of several limitations. Because NHIS data are cross-sectional, we cannot determine whether receiving a cancer diagnosis led to changes in health insurance or cost barriers to receiving care. Second, we do not know who the policyholder was or how long an individual was covered or uncovered. Studies of cancer patients and their families could help address these issues. Third, we used self-reported data; therefore, some recall bias or misclassification may have occurred. Furthermore, we excluded some respondents missing information regarding health insurance. Also, NHIS does not include institutionalized people, which may affect generalizability among survivors. We were unable to evaluate differences in insurance among racial or ethnic groups other than white, African-American, and Hispanic due to small numbers of survivors in other groups in our study. Finally, not all uninsured participants responded to the question concerning why they did not have insurance.

Among cancer survivors age <65 years, 1 in 8 was uninsured. Although survivors more often reported being insured than adults without cancer overall, among groups of similar age, sex, race, and ethnicity, the lack of health insurance was similar between survivors and other adults. Among adults with insurance coverage, survivors may be more likely than other adults to be publicly insured. Among survivors, some groups were more likely to lack insurance than others. More research using longitudinal data is needed to assess whether receiving a cancer diagnosis leads to changes in health insurance. Unmet medical care needs because of cost were common among survivors overall, with findings particularly striking for uninsured survivors. Healthy People 2010, the nation's health promotion and disease prevention agenda, recognizes the central role of health insurance in providing access to quality health care, targeting universal coverage by the year 2010 in objective 1-1.40 Efforts to expand health insurance coverage are likely to increase access to needed health care for cancer survivors in this age group.

Ancillary