|Home | About | Journals | Submit | Contact Us | Français|
Many US cancer survivors live years after diagnosis, emphasizing the importance of health care access for survivors. It is not known if having cancer might impact disparities in health care access that are present in the general population. The objective of this study was to examine the prevalence of forgoing care due to financial concern in a representative sample of US adults to determine if cancer history and race/ethnicity are associated with likelihood of forgoing medical care.
The US National Health Interview Survey (NHIS)- 2003–2006 was used to identify 6,602 adult cancer survivors and 104,364 individuals with no history of cancer. We examined self-report of forgoing medical care services because of cost by cancer history and race/ethnicity using multivariate logistic regression.
The prevalence of forgoing care due to cost among cancer survivors was 7.8% for medical care, 9.9% for prescription medications, 11.3% for dental care, and 2.7% for mental health care. Cancer survivors under the age of 65 years were more likely to delay or forgo all types of medical care when compared to adults without a history of cancer. Hispanic and black cancer survivors were more likely to forgo prescription medications and dental care than white survivors. Disparities among cancer survivors were largely reflective of those in the general adult population.
More than two million US cancer survivors did not get one or more needed medical services because of financial concerns. Future research needs to examine the impact of forgoing care on survivors’ quality of life and survival.
In the United States, there are approximately 12 million adults with a history of cancer, commonly referred to as cancer survivors 1. Access to timely cancer and non-cancer related medical care is an important issue for this growing population. Medical needs of cancer survivors include surveillance for primary recurrence and second malignancies, monitoring for chronic and late effects, treatment for other medical comorbidities, mental health services, and general preventive care 2. Hewitt and colleagues 3 found that relative to adults with no history of cancer, survivors are more likely to be in fair or poor health and to have functional limitations. Survivors were also more likely to consult physician specialists and to use supportive medical services.
Despite their great need for medical services, cancer survivors may experience barriers to accessing care. Cancer treatment may result in financial hardship 4;5 and inability to afford medical copayments, prescription medications, and medical services perceived as being non-essential (e.g., dental and mental health care). In addition, cancer-related employment changes 3;6 may result in loss of health insurance coverage, and survivors who qualify for government insurance may lose this coverage once their initial treatment is complete. However, one investigation of insurance coverage7 concluded that cancer survivors 65 years of age and under were no more likely to be uninsured than adults of similar age, sex, and race/ethnicity.
Previous examinations of health care utilization among cancer survivors, have primarily used administrative databases such as the SEER-Medicare database8–10. While informative, these studies are limited in that they only included older cancer survivors and cannot elucidate reasons for forgoing medical care, such as financial concern. There are also differences in health care access between older and younger survivors, suggesting the need to examine financial barriers to care separately by age group.
The purpose of this study was to examine the prevalence of forgoing health care services because of cost among adults with a history of cancer using a nationally representative, population-based sample of US adults. Hispanic and non-Hispanic black survivors were compared with non-Hispanic white survivors to determine if rates of forgoing care differed among ethnic groups. Finally, disparities in forgoing care were compared in cancer survivors and the general adult population to determine if the magnitude of disparities differed as a function of cancer history.
Analyses utilized data from the National Health Interview Survey (NHIS)11, an annual, in person, nationwide survey of approximately 30,000–40,000 households in the civilian, non-institutionalized population, used to track trends in illness and disability in the United States. We combined data from the 2003–2006 surveys to obtain a sample of cancer survivors with sufficient power for cancer versus non-cancer history comparisons. The NHIS utilizes a complex survey design involving clustering, stratification, and multistage sampling 11. Each survey year represents a different sample and blacks and Hispanics are oversampled. The final response rate for the adult sample ranged from 69 to 74.2% in the years studied 12.
We defined cancer survivors as those individuals who reported a history of cancer. Of the 117,881 persons 18 years and older surveyed across the four years, cancer history was unknown for 210 individuals and they were excluded. We also excluded participants reporting only non-melanoma skin cancer (n= 1147) or “unknown” skin cancer (n= 669) and those survivors (n=161) and adults without a history of cancer (n=4,728) who reported non-Hispanic “other” race/ethnicity. This resulted in an analytic sample of 6,602 cancer survivors and 104,364 adults without a history of cancer.
Age, gender, race/ethnicity, cancer site, insurance coverage, and comorbid medical conditions were assessed during the individual and household interviews. Insurance coverage was categorized as public (e.g., Medicaid, state health insurance programs, military health care, or Medicare without supplemental coverage) or private coverage (e.g. HMO or preferred provider with or without Medicare coverage) based on the National Center for Health Statistics recommended recoding variables14, with no insurance coverage added for those under 65 years of age. A very small proportion of individuals (0.9 %) aged 65 year or older reported having no health insurance coverage. We selected self-reported non-cancer comorbidities that would likely require medical attention in the previous 12 months (the time frame for forgoing care) from those available on the NHIS. These included hypertension (diagnosed on two or more occasions), coronary artery disease (ever), stroke (ever), emphysema (ever), asthma (current), ulcer (during the past 12 months), diabetes (ever, excluding gestational diabetes), “chronic bronchitis” (during the past 12 months, “weak or failing kidneys” (during the past 12 months, not including kidney stones, bladder infections, or incontinence), “liver conditions” (in the previous 12 months), and arthritis/joint pain (ever diagnosed, includes arthritis, rheumatoid arthritis, gout, lupus, or fibromyalgia). Persons with one or more comorbidities were compared to those with no comorbidities. For the purpose of this manuscript, Hispanic will refer to persons of Hispanic or Latino heritage of any race, white will refer to non-Hispanic white persons and black will refer to non-Hispanic black persons. Self-reported education (highest level completed) was used as a proxy for socioeconomic status because approximately 25% of the cases were missing household income information. The association between self-reported income and education was very strong (Cochran Mantel Hanzel test for trend: F= 4463, p <.0001).
Participants were asked, “During the past 12 months, was there any time when you needed medical care, but did not get it because you couldn’t afford it?” Delayed medical care was assessed via the question, “During the past 12 months, have you delayed seeking medical care because of worry about the cost?” Finally, to assess non-receipt of prescription medicines, mental health care or counseling, and dental care (including check-ups), participants were asked, “During the past 12 months, was there any time when you needed any of the following, but didn’t get it because you couldn’t afford it?” Possible response options for each item included: “yes”, “no”, “don’t know”, and “refused”. The latter two options were considered to be missing.
All analyses were conducted using SUDAAN to account for the complex sampling design of the survey. The sampling strategies used to create the NHIS sample, such as stratification, clustering, and oversampling of certain ethnic groups, result in subjects having different probabilities of selection. Failure to account for this by incorporating sampling weights (as would be the case with simple linear regression) would result in biased estimates of model parameters and variances. We examined models stratified by age (under 65 years compared with 65 years and older). The primary coefficients of interest were race/ethnicity (two dummy codes reflecting the comparison between Hispanic or black and white groups), cancer history, and the interaction between race/ethnicity and cancer history. A statistically significant interaction would indicate that the effect of race/ethnicity was larger or smaller in cancer survivors compared to the general adult population. To illustrate significant interactions, we calculated the odds ratios for Hispanic and black cancer survivors, with white cancer survivors serving as the reference group. All tests of statistical significance were two-sided with the p-value set at .05. In addition, we calculated national estimates for the number of cancer survivors forgoing or delaying different types of care by multiplying the frequency of each forgoing care outcome by the sum of the weights to obtain population estimates.
Self-reported demographic characteristics stratified by age and cancer history are shown in Table 1. Survivors and adults without a cancer history differ on several characteristics because of differences in age and patterns of cancer incidence. Hispanics and blacks were under-represented in the cancer survivor samples, likely due to younger age, lower cancer incidence, and lower cancer survival. For the cancer survivor sample, the most common cancer sites included breast (18.6%), prostate (12.5%), cervix (9.0%), colon (6.9%), melanoma (7.6%), and uterus (5.4%). Multiple cancers were reported by 11.2% of cancer survivors. These do not mirror patterns of cancer incidence because of site differences in survival. Most cancer survivors (58.8%) were more than five years out from their cancer diagnosis, with 5.2% less than a year, and 36.0% one to five years from diagnosis. Survivors diagnosed before the age of 18 years comprised 4.3% of the sample.
The prevalence of forgoing or delaying care due to cost by cancer survivors was 7.8% (95% CI 7.1–8.6) for forgoing medical care, 10.7% (95% CI 9.8–11.6) for delaying medical care, 9.9% (95% CI 9.1–10.8) for forgoing prescription medication, 11.3% (95% CI 10.4–12.2) for forgoing dental care, and 2.7% (95% CI 2.3–3.2) for forgoing mental health care. The prevalence of forgoing one or more medical services was 17.6% (95% CI 16.6–18.8). Prevalence estimates stratified by ethnic groups are shown in Figure 1. In the total sample, Hispanic and black cancer survivors were more likely to forgo prescription medications (Hispanic OR = 2.14, 95% CI 1.52–3.00 & black OR= 1.87, 95% CI 1.38–2.54) and dental care (Hispanic OR = 2.31, 95% CI 1.68–3.17 & black OR= 1.57, 95% CI 1.18–2.10) than white survivors. Hispanic survivors were also more likely to forgo medical care compared to white survivors (OR= 1.55, 95% CI 1.05–2.29). There were no significant differences by ethnic groups for forgoing mental health care or delaying medical care.
Using weights calculated to reflect sampling design and post-stratification adjustment for sex, age, and race/ethnicity, it is possible to obtain national estimates for the outcomes. The frequency count for each outcome was multiplied by the sum of the weights to approximate the total population. Weighted numbers suggest that approximately 893,000 cancer survivors did not get medical care, 1.22 million delayed medical care, 1.12 million did not get prescription medications, 1.28 million did not get dental care, and 310,000 did not get mental health care because of cost. The weighted population estimate is that 2,002,000 cancer survivors did not get one or more medical services because of financial concerns.
Cancer survivors under the age of 65 years were 1.49 to 1.95 times (all ps< .05) more likely to forgo or delay medical services than adults without a history of cancer (see Table 2A). Adjustment for gender, education, health insurance coverage, and comorbid medical conditions reduced, but did not eliminate the effect of cancer history (see Table 2B). In contrast, for the 65 years and older group there were no significant differences between survivors and those without a history of cancer in either the unadjusted or adjusted models (see Table 3).
We examined the interaction between cancer history and race/ethnicity to determine if the effect of race/ethnicity differed in survivors and the general adult population. The comparisons between ethnic minority and white adults without cancer are represented by the race/ethnicity main effects and the comparisons between ethnic minority and white cancer survivors are represented by the interaction effects.
Among adults under the age of 65 without a history of cancer, Hispanics and blacks were significantly more likely to forgo medical care, prescription medications, and dental care, but not mental health care (see Table 2A). None of the interactions were statistically significantly, suggesting a similar pattern of results in cancer survivors. In general the odds ratios comparing Hispanics and blacks to whites were larger in the 65 and older group for all of the health care outcomes(see Table 3A), with ethnic minorities more likely to forgo all types of care. Again, most of the interaction terms were nonsignificant, suggesting that ethnic disparities among survivors are largely reflective of disparities in the general adult population.
In general, adjustment for gender, education, insurance coverage, and comorbidities reversed or eliminated ethnic disparities in outcomes observed in unadjusted models (see Table 2B). After adjustment, black and Hispanics without a cancer diagnosis were less likely to forgo medical and mental health care and delay medical care. This trend was not statistically significant for the cancer survivors, except for the comparison between black and white survivors for delaying medical care. There were significant interactions between race/ethnicity and cancer history for forgoing prescription medication, delaying medical care due to cost, and forgoing dental care (see Table 2B). Adjusted results for adults under 65 years of age suggest that although in the general population Hispanics are less likely than whites (OR= .75, 95% CI= 0.68–0.82) ) to report forgoing prescription medication due to cost, Hispanic cancer survivors are more likely than white survivors (OR = 1.18, 95% CI = 0.81–1.72) to forgo medication. Similarly, Hispanic adults without cancer are less likely to report forgoing dental care (OR= .74, 95% CI= 0.68–0.80), but Hispanic survivors are more likely to forgo dental care compared to white survivors (OR= 1.33, 95% CI= 0.94–1.87). Finally for delaying care due to cost, Hispanic adults without cancer were less likely than whites to delay medical care due to cost (OR= .55, 95% CI= 0.51–0.60), but there was no significant difference between Hispanic and white survivors (OR= 0.84, 95% CI= 0.58–1.23). The interaction between cancer history and race/ethnicity was not statistically significant for the comparisons between blacks and whites, suggesting similar patterns between adults with and without a history of cancer.
Compared to unadjusted models, the effect of race/ethnicity on forgoing care was diminished in the adjusted model (see Table 3B). Significant ethnic disparities remained for prescription medications and dental care in the general population, with Hispanic and black seniors being more likely to forgo care because of cost. None of the interaction effects were statistically significant in those persons 65 years and older, suggesting that the magnitude of the observed racial/ethnic disparities in forgoing or delaying health care did not differ between cancer survivors and the general adult population. Interestingly, in adjusted models, Hispanic cancer survivors appeared to be less likely to forgo or delay medical care and more likely to forgo mental health services, but the interaction terms did not reach statistical significant. There was also a trend for black survivors to be more likely to forgo mental health care services, but again this interaction did not reach statistical significance.
We were not able to include cancer specific variables in our comprehensive model because there was no equivalent for those individuals without a cancer history. However, in an exploratory analysis we examined the prevalence of forgoing care separately for cancer survivors to describe differences among the five most common cancer sites (breast, cervix, prostate, melanoma, and multiple cancers) and by time since cancer diagnosis. There were significant differences in forgoing or delaying all medical services by cancer site (see Table 4). In all cases, breast and prostate cancer survivors were least likely and cervical cancer survivors were most likely to forgo services. Those survivors with melanoma or multiple cancers reported intermediate levels of forgoing services. After adjusting for age (under 65 vs 65+), education, insurance coverage, and medical comorbidities (0/1 vs 2 or more), cervical cancer survivors were significantly more likely than breast cancer survivors to forgo all health services (ORs ranging from 1.54–3.08, all p <.05). Melanoma survivors were also significantly more likely than breast cancer survivors to forgo medications (OR= 1.68, 95% CI 1.04–2.72). There were no other significant differences by site.
There were also significant differences by time since diagnosis for forgoing and delaying medical care (see Table 4). Survivors who were less than one year since the time of their diagnosis were most likely to forgo or delay medical care, with the next highest levels reported by very long-term (10+ year) survivors. Very long-term survivors were also most likely to forgo mental health care services. There were no differences by time since diagnosis for forgoing medications or dental services. After adjusting for age (under 65 vs 65+), education, insurance coverage, and medical comorbidities (0/1 vs 2 or more), significant differences by time since diagnosis remained only for delaying medical care. Compared to survivors one to five years post-diagnosis, those survivors less than one year (OR= 1.59, 95% CI 1.01–2.50) and ten or more years post-diagnosis (OR= 1.31, 95% CI 1.03–1.66) were more likely to delay medical care because of cost.
We estimate that more than two million of the estimated 10–12 million cancer survivors living in the United States during the time period of the survey (2003–2006) did not get one or more needed medical services because of concerns about cost. The prevalence of forgoing care among cancer survivors (7.8% forgoing medical care and 9.9% forgoing prescription medications) is higher than that reported for the US population in general (5.2% and 7.2%, respectively for the year 2005)13. Lack of access to medical care among cancer survivors is a significant public health concern, given the importance of regular medical care for cancer survivors and their growing numbers.
Consistent with earlier research examining health care access in younger survivors using the NHIS7, we found that cancer survivors under the age of 65 years were more likely to delay or forgo all types of medical care when compared to their same age peers without a history of cancer.. This is concerning because individuals who are diagnosed and treated as younger adults, especially those with early stage cancers, are expected to live many years after their cancer diagnosis, making access to appropriate preventive and health maintenance services crucial for their long-term health and well-being14. It is likely that the comparison group of adults without a history of cancer is healthier in the younger age group, increasing the effect of cancer history. In fact, a large percentage of adults under 65 years without cancer reported no comorbid medical conditions (65.5%), compared with only 36.4% of younger cancer survivors.
In contrast, cancer history was not associated with delaying or forgoing care in those persons 65 years and older. The greater incidence of comorbid medical conditions in the older group and the resulting need for more medical services may account for the diminished effect of cancer history in this group. The difference in distribution of comorbidities between survivors and adults without cancer is much less striking in persons 65 years and older compared to the younger age group. Previous research examining receipt of health care services among cancer survivors 65 years and older using the SEER-Medicare database showed mixed results for cancer history. Long-term colorectal cancer survivors were less likely to receive necessary medical care services such as appropriate follow up for hospitalizations and chronic medical conditions such as heart disease and diabetes 8, One longitudinal study of elderly nonmetastatic breast cancer survivors found that survivors were consistently less likely to receive preventive health services compared to matched controls15, but an earlier cross-sectional study found that survivors were more likely to receive preventive health services 9. Importantly, these studies did not focus specifically on financial barriers to care or ethnic disparities among survivors.
Rates of forgoing care were not consistent across cancer sites. Consistently, women with cervical cancer reported higher rates of forgoing care compared to breast, prostate, melanoma, and multiple cancer survivors. Future research is needed to better examine financial barriers to care among this vulnerable group of cancer survivors. Interestingly, few differences were observed by time since diagnosis. Survivors within one year of diagnosis were more likely to delay care because of cost, possibly due to high out of pocket costs during initial cancer treatment.
As found in previous population- based research comparing cancer survivors and adults without cancer 3;16, adults with and without a cancer history from our sample differed in several ways. Adults reporting a history of cancer were more likely to be older, female, and white. This is consistent with SEER cancer registry estimates showing that cancer prevalence increases with age, females comprise the majority of persons ever diagnosed with cancer, the female proportion increases with increasing time since diagnosis, and the vast majority of cancer survivors currently living in the United States are white1. It has also been suggested that underreporting of cancer history on self-report surveys may be more likely to occur among men 17, possibly resulting in a larger proportion of female cancer survivors identified on the NHIS.
In the total sample of cancer survivors, Hispanic and black cancer survivors were more likely to forgo prescription medications and dental care than white survivors and Hispanic survivors were more likely to forgo medical care. This is consistent with previous survey research which found that black and Hispanic cancer survivors were more likely to report that costs of medication, diagnostic tests, and hospitalization were a barrier to needed cancer treatment18, Our study builds on this prior survey by using a national population-based sample of US survivors and including a comparison sample of adults without cancer. Our questions about forgoing medical care were not cancer-specific which may be more relevant to this population of long-term survivors. Other studies have documented high levels of financial stress and concerns about medical costs among minority cancer survivors19, but it is not known whether these concerns resulted in forgone medical care.
Adjusted models suggest that race/ethnicity disparities were largely due to difference in socioeconomic status and medical insurance coverage. The occurrence of medical comorbidities may also contribute to differences in forgoing care. Among survivors and adults without cancer, those adults reporting more comorbidities were more likely to forgo care (data not shown). This is worrisome as cancer treatment can be associated with late and long-term health effects14. If these same conditions exacerbate financial obstacles to treatment, then health disparities are likely to be exacerbated.
Relatively larger disparities between ethnic minority and white persons were observed among those individuals 65 years and older compared to the under 65 group, but these disparities were similar in the survivor and general adult populations. The finding that black older adults (with or without a cancer history) are more likely to forgo medical care is consistent with administrative database research documenting that older African-American women are less likely to receive preventive health services9, black breast cancer survivors are less likely to receive influenza vaccination, bone densitometry, and mammograms15, and non-white colorectal cancer survivors are less likely to receive influenza vaccination20 and other preventive care services20;21. Examining ethnic disparities was not a focus of these studies, so comparisons in disparities were not made between adults with and without cancer.
This study adds to the body of knowledge on health care access disparities in cancer survivors by examining non-receipt or delays in care due to financial concerns using a representative sample of the US population, however there are several limitations. First, this study relied on self-report of delaying or forgoing care due to financial concerns. A respondent answering “no” to the question “Was there any time when you needed medical care, but did not get it because you couldn’t afford it?” could have two possibilities in mind. One, no care was needed and therefore financial concerns were irrelevant, or two, care was needed and received. Thus, differences in forgoing care may also reflect differences in health and perception of needing care. Cancer survivors may be sensitized to perceive greater health care needs, thereby increasing opportunities to forgo or delay health care. Self-reports of forgoing or delaying care may also be subject to recall bias. Perhaps medical care that is not received is more salient to cancer survivors than to the general population because of vigilance for signs of cancer recurrence. Second, although we stratified by age, differences in the age distribution of the cancer survivors and the general population may have resulted in residual confounding. In addition, our assessment of comorbidities was incomplete. We relied on self-reported comorbidities that were asked for other purposes on the NHIS and were not able to assess the currency or severity of every condition. Thus, additional research among survivors is needed to better understand the role of comorbidities in forgoing medical care due to financial strain. Finally, although the sample is representative of community dwelling adults, it is not representative of the entire population of cancer survivors, especially those in poorest health who might reside in hospices, nursing homes, or hospitals or who were too ill to complete the interview17. Thus, our results might not generalize to survivors diagnosed with the most advanced disease and with sites associated with relatively short survival. NHIS also does not collect information about cancer stage at diagnosis, current treatment for cancer, or cancer recurrence, so we can not characterize these features of our survivor population.
This paper identifies an important issue is survivorship care, namely that financial concerns may prevent cancer survivors, particularly those under the age of 65, from accessing needed medical services. However, given the preliminary nature of this data, further research is needed to answer crucial questions in this area. First, future studies should better characterize the types of health care services that survivors delay and forgo because of financial concerns. It is important to know if the services are cancer-related (e.g., screening for recurrence or second cancers, assessment or treatment of late-effects) or related to other comorbid conditions. Future research should also investigate whether cancer survivors seek a greater number of services overall and/or choose to forgo less crucial or elective services. Information about the type of care that was not received is not available in the NHIS, or any other population-based dataset of which we are aware. Therefore targeted research is needed to ask survivors about their experiences forgoing care because of medical cost and the implications for their health. It is also important to better understand the role of insurance coverage and out of pocket expenses when evaluating financial barriers to needed care, particularly in light of changes in the health insurance industry such as increases in high deductible plans, increased insurance coverage in certain states, and increased coverage for mental health care services. Finally, it is crucial to investigate the impact of forgoing care on survivors’ physical and mental well-being, length of survival, and cost of care.
Although race/ethnicity disparities in forgoing care due to cost among cancer survivors largely reflect those in the general adult population, given their need for monitoring of recurrence and late effects of treatment, increasing access to care for cancer survivors is an important public health goal. To improve physical and mental well-being of cancer survivors, assessment of access to care may need to become a routine part of post cancer treatment follow-up plans, especially for vulnerable groups such as ethnic minority, uninsured or underinsured, and younger survivors. Referral for social work services may be helpful in educating survivors about low cost sources of medical care and financial assistance programs. Ultimately, public policy interventions that improve access to health insurance coverage generally would appear to have the greatest potential impact on reducing health care access disparities among cancer survivors.
We would like to thank Chris Zeruto and Tim McNeel for their assistance with the preparation of the NHIS dataset and analyses. Results from this manuscript were presented at the 2009 AACR Cancer Health Disparities conference. The authors have no potential conflicts of interest to report. Kathryn Weaver is now in the Department of Social Sciences and Health Policy, Division of Public Health Sciences, Wake Forest University School of Medicine, Winston-Salem, NC.
There are no financial disclosures from any of the authors.