Trends in Healthcare Expenditures among Adults in the United States by Cancer Diagnosis Status, 2008–2016: A Cross-Sectional Study

Cancer survivorship, defined as the period from the point of cancer diagnosis until the end of life, is growing in the United States (1–4). In 2020, it is estimated that the number of cancer survivors in the United States will reach over 18 million (5). By 2026, that number is projected to increase to approximately 20.3 million (6). This increase in cancer survivorship has been attributed to increasing cancer prevalence due to a larger and aging population, as well as advances in medical screening, treatment, and early disease detection (3, 6–9). Because of these factors, expenditures associated with a prior cancer diagnosis are also projected to increase (3). In 2010, the national costs of cancer care totaled over $124 billion (10). By 2020, the estimated cost of cancer care is projected to reach $157 billion (2010 dollars), an increase of 27% (3, 10). As adults with a prior cancer diagnosis are aging and living longer, and with medical advancements emerging, a better understanding of the current trends in healthcare expenditures and incremental expenditures (i.e., expenditures beyond the cost associated with receipt of usual care that occur through time) associated with a prior cancer diagnosis are warranted.

Adults with a prior cancer diagnosis face increased disease, social, and financial burdens when compared with those with no history of malignancies (11–21). While many cancer survivors report overall good health and quality of life, existing research on the effects of treatment shows those with a prior cancer diagnosis face increase rates of financial hardship, reduced workplace and school productivity, physical and/or psychosocial limitations, and increased subsequent healthcare utilization when compared with adults without a prior diagnosis of cancer (2, 11–13). When compared with no prior cancer diagnosis, having a prior cancer diagnosis is further associated with increased healthcare costs to the patient and society. Healthcare expenditures for cancer comprise 5% of all national healthcare expenditures and an estimated 10% of Medicare expenditures (22, 23). According to the Agency for Healthcare Research and Quality, cancer care costs in the United States during 2011 totaled $88.3 billion compared with $56.8 billion in 2001, a trend projected to increase to $173 billion through 2020 (3, 24). In 2011, for those who obtained treatment for cancer, ambulatory visits, inpatient hospital stays, prescription medicine, and home health were drivers of the higher total healthcare expenditures related to their care and management (24).

Previous research has shown an association between a prior cancer diagnosis and increased expenditures (11). Guy and colleagues found that adults with a prior cancer diagnosis faced excess economic burden of $16,213 per survivor aged 18 to 64 years and $16,441 per survivors aged 65 years and older, with additional costs stemming from losses of productivity including health-related missed workdays and employment disability (11). While previous studies have sought to assess the economic burden associated with a prior cancer diagnosis these studies relied on pooled data samples and were limited in their ability to assess temporal trends in cancer-associated costs (11–14, 23–28). Few studies have examined incremental expenditure and trends in direct and indirect medical expenditures among adults with a prior cancer diagnosis in the United States over time, compared with those with no prior cancer diagnosis, using a nationally representative sample (11–14, 23–28). As such, the aim of this study was to examine trends and incremental healthcare expenditures (defined as expenditures beyond the costs associated with receipt of usual care) and indirect costs (defined as health-related missed workdays) by cancer diagnosis status among a nationally representative sample of adults. We hypothesized that adults with a prior cancer diagnosis would have higher direct and indirect costs compared with those without a prior cancer diagnosis.

In this retrospective study, we analyzed a weighted population representing 236,811,875 (unweighted n = 227,110) U.S. adults aged ≥18 years from 2008 to 2016 using the Medical Expenditure Panel Survey Household Component (MEPS-HC). The MEPS provides nationally representative yearly cost estimates of healthcare utilization, expenditures, sources of payment, and health insurance status for the U.S. civilians, noninstitutionalized population (22). The MEPS-HC study participants are a subsample of households that participate in the National Health Interview Survey (NHIS) conducted by the National Center for Health Statistics (29). NHIS participants reside within the District of Columbia and the 50 states of the United States and includes residents of households and noninstitutionalized group settings such as group homes and homeless shelters. NHIS uses geographically clustered sampling techniques to select the sample of dwelling units for NHIS, and units are sampled in a manner such that the sample is nationally representative (30). Comprehensive data about respondent demographics, health status and conditions, access to care, satisfaction with care, and income were collected during in-person and household interviews.

The MEPS has three components that include the HC, the Medical Provider Component (MPC), and the Insurance Component (3, 32). The MEPS HC component collects information via self-report on sociodemographic characteristics, health status and conditions, healthcare utilization and expenditures, sources of payment, and insurance coverage status (33). The MPC requests data on medical and financial characteristics from hospitals, clinicians, home healthcare providers, and pharmacies to validate and supplement information collected from MEPS-HC respondents (33). Diagnoses coded according to the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) is also collected for this component.

The panel design of the MEPS survey includes five rounds of interviews that cover 2 full calendar years (34). Each panel collects data from different respondents to allow the merging of panel data for a pooled cross-sectional study sample. We combined 9 years of data to increase sample size and the precision of the estimates for those with cancer diagnosis information. Due to its complex survey design and multistage likelihood examining strategy, weighting sample of families and individuals in selected communities can represent the U.S. populace.

The outcome variables in this analysis were total direct healthcare expenditure and health-related missed workdays for each calendar year for each individual. Healthcare expenditures included those covered by insurance such as office-based medical expenses and hospital outpatient, emergency room (ER), inpatient hospital (including zero-night stays), prescription, dental, home health care, and other medical expenditures. Health-related missed workdays were the number of missed workdays due to injury, mental or physical illness reported by survey participants who were employed during the survey year. Mean expenditures are reported in 2016 U.S. dollars (35).

The primary independent variable was a previous cancer diagnosis. This variable was collected by self-report, based on the question, “Have you ever been told by a doctor or other health professional that you had cancer or a malignancy of any kind?”, consistent with other studies that assessed cost and a prior cancer diagnosis (11, 28, 36).

All covariates in the analysis were based on self-report. Race/Ethnicity groups were categorized as non-Hispanic white, non-Hispanic Black, non-Hispanic Other (a variable that included Asian Americans, Native Hawaiians/Pacific Islanders, American Indians/Alaska Natives), and Hispanic. Education was categorized as less than high school (≤11 years), high school (12 years), and college (≥13 years). Marital status was coded as married, not married (including divorced, widowed, or separated), and never married. Sex was categorized as men or women, and age at the time of the interview was categorized as 18 to 44 years, 45 to 64 years, and 65 years and older. Census region at the time of the interview was categorized as Northeast, Midwest, South, and West. Health insurance was categorized into three groups: private insurance, public insurance, or uninsured. Income level referred to the family income and was defined as a percentage of the federal poverty level as poor/negative income (<100%), low income (≥100% to <200%), middle income (≥200% to <400%), and high income (≥400%) and derived from the entire survey year. The comorbidities included were based on a positive response to the question, “Have you ever been diagnosed with hypertension, stroke, emphysema, joint pain, arthritis, or asthma?” The presence of cardiovascular disease was indicated by a positive response to the question, “Have you ever been diagnosed with coronary heart disease, angina, myocardial infarction, or other heart disease?” Comorbid conditions selected for inclusion were based on previous studies that used the MEPS dataset to assess trends in patient costs associated with chronic health conditions (37–39). Calendar year for survey completion was coded in five categories: 2008/2009, 2010/2011, 2012/2013, 2014/2015, and 2016 for the pooled data.

Descriptive statistics were calculated for participants with a prior cancer diagnosis and without a prior cancer diagnosis. χ² tests were used to assess differences between independent variables by cancer diagnosis status. Unadjusted means and 95% confidence Intervals (CI) were calculated overall and by cancer diagnosis status and were reported by year for each expenditure category. Medical expenditure and health-related missed workdays were skewed and included a large number of zeroes, meaning there are many respondents who report zero health care expenditures across all years of the study period and zero missed workdays due to health across all years of the study period (40). Due to the skewedness and large number of zeroes, we used two-part generalized linear modeling (GLM) for the expenditures and missed workdays, which allows for mixed discrete-continuous outcomes and provides the ability to assess incremental direct and indirect costs over time (41). The first part was a probit model, which gives an estimate of probability of having zero expenditure versus positive expenditure. The second part, conditional on a positive value, used a GLM with gamma distribution and log link to estimate the expenditures conditional on having a positive expenditure. The modified Park test was used to ensure the appropriate family distribution for the GLM was selected (41, 42). Results showed gamma distribution with log link was the most appropriate for the outcomes. The margins command in STATA (StatCorp LP College Station, TX) was used to calculate the incremental expenditures and missed workdays (42, 43). All models were calculated unadjusted and fully adjusted for all covariates. The weighted, clustered, and stratified model was used to ensure the representation of the U.S. population (34). The Consumer Price Index was used to adjust all individual year expenditures to 2016 dollars. Pooled estimates were not adjusted.

The dataset generated and/or analyzed during the current study is available in the Medical Expenditure Panel Survey at https://www.meps.ahrq.gov/mepsweb/. L.E. Egede received grants from the National Institute of Diabetes and Digestive and Kidney Diseases (K24Dk093699, R01DK118038, and R01DK120861), which partially supported this study.

Table 1 shows the weighted sample demographics by cancer diagnosis status among U.S. adults from 2008 to 2016. Of the weighted population (n = 236,811,875), 11% of adults aged ≥18 years reported a previous cancer diagnosis. Significant differences between adults with a prior cancer diagnosis and adults without a prior cancer diagnosis were found by time, demographic characteristics, and comorbid conditions. A prior cancer diagnosis occurred most often in adults who were aged 65 years and older, women, married, had a college education, had private insurance, resided in the South, reported higher income, and had comorbid conditions. The prevalence of cancer survivors increased significantly between years 2008–2009, 2010–2011, 2012–2013, and 2014–2015.

Figure 1 displays the trends in mean total expenditure by cancer diagnosis status between years 2008 and 2016 and for the pooled sample. When compared with adults with no prior cancer diagnosis, those with a prior cancer diagnosis had higher total mean incremental expenditures for 2008/2009 ($4,239 vs. $11,904), 2010/2011 ($4,466 vs. $12,696), 2012/2013 ($4,435 vs. $11,622), 2014/2015 ($4,762 vs. $12,744), 2016 ($4,720 vs. $13,829), and for the pooled sample ($4,267 vs. $11,794).

Figure 2 shows the unadjusted pooled means of total healthcare expenditures by cancer diagnosis status and expenditure type. Among those with a prior cancer diagnosis, the mean total healthcare expenditures of the pooled sample were $11,794, while the pooled sample of adults without a prior diagnosis of cancer reported a mean expenditure of $4,267. Mean inpatient expenditures among the pooled sample of those with a prior cancer diagnosis was $3,390. Among those with no prior diagnosis of cancer, mean inpatient expenditures for the pooled sample was $1,114. Mean office-based expenditures for the pooled sample of those with a prior cancer diagnosis was $3,088. Among those without a prior cancer diagnosis, the mean office-based expenditure for the pooled sample was $1,026. Those with a prior cancer diagnosis reported higher medication expenditures at $2,533 while those with no prior cancer diagnosis reported medication expenditures of $1,050. Outpatient expenditures for those with a prior cancer diagnosis was over three-fold of that for those without a prior cancer diagnosis, at $1,275 compared with $375 ($348, $401). The pooled sample of those with a prior cancer diagnosis had higher dental mean expenditures at $427 compared with those without a prior cancer diagnosis at $268. ER mean expenditures for the pooled sample were also higher for those with a prior cancer diagnosis at $297 compared with those with no prior cancer diagnosis at $191. The mean home health expenditures in the pooled sample of those with a prior cancer diagnosis were $593, whereas those without a prior cancer diagnosis reported $150. Other expenses for the pooled sample of those with a prior cancer diagnosis were higher at $122 compared with those with no prior cancer diagnosis at $38. This data along with 95% CIs is presented as Supplementary Table S1.

Table 2 shows the incremental effects of healthcare expenditure by cancer diagnosis status among U.S. adults from 2008 to 2016. In unadjusted analyses, those with a prior cancer diagnosis had significantly higher total healthcare costs of $7,696 (95% CI, $7,206, $8,186), inpatient expenditures of $2,373 (95% CI, $2,070–$2,676), office-based expenditures at $2,072 (95% CI, $1,942–$2,202), medication expenditures at $1,497 (95% CI, $1,355–$1,638), outpatient expenditures at $928 (95% CI, $809–$1,047), dental expenditures at $156 (95% CI, $130–$181), and Other expenditures at $84 (95% CI, $69–$99) compared with those without a prior cancer diagnosis. After adjusting for demographic characteristics, comorbidity covariates, and time, the significant associations remained for total healthcare costs of $3,522 (95% CI, $3,072–$3,972), inpatient expenditures of $810 (95% CI, $627–$992), office-based expenditures of $1,085 (95% CI, $990–$1,180), medication expenditures of $409 ($95% CI, $295–$523), and outpatient expenditures of $517 (95% CI, $434–$600) compared with those without a prior cancer diagnosis. Significant differences in dental and other expenditures were no longer significant after adjusting for covariates.

In unadjusted analyses, those with a prior cancer diagnosis had statistically significant more numbers of missed workdays of 1.36 (95% CI, 1.05–0.67). After adjusting for demographic characteristics, comorbidity covariates, and time, these differences remained such that those with a prior cancer diagnosis had a statistically significantly greater number of missed workdays of 0.75 (95% CI, 0.45–1.04) compared with those with no prior cancer diagnosis.

In this nationally representative sample of U.S. adults, those with a prior cancer diagnosis had significantly higher healthcare expenditures and health-related missed workdays compared with adults without a prior cancer diagnosis, after adjusting for relevant confounding factors. In unadjusted analyses, adults with a prior cancer diagnosis had higher total healthcare expenditures and higher incremental expenditures associated with inpatient stays; office-based, outpatient, and dental visits; medications; and other services compared with adults without a prior cancer diagnosis. After adjusting for covariates, significant differences in incremental expenditures between adults with and without a prior cancer diagnosis remained for all services except dental visits and other services. Adults with a prior cancer diagnosis also reported more missed workdays in both unadjusted and adjusted models. These findings suggest having a prior cancer diagnosis is associated with higher healthcare expenditures across service types and increased health-related missed workdays compared with not having a prior cancer diagnosis. Our study adds to the literature by assessing temporal changes in cancer-associated direct expenditures, which allows for comparisons through changing healthcare policies for a population experiencing rapid growth. In addition, our inclusion of indirect costs associated with a prior cancer diagnosis provides insight on interventions to ease social burdens among adults with a prior cancer diagnosis.

Our findings are supported by evidence in the literature that sought to assess the economic burden associated with a prior cancer diagnosis (11, 13, 25, 27). In this sample, we found adults with a prior cancer diagnosis to have significantly higher incremental healthcare expenditures compared with adults without a prior cancer diagnosis. In a study from 2008 to 2011 that estimated direct medical costs and indirect morbidity costs among U.S. adult cancer survivors compared with adults without a history of cancer, Guy and colleagues reported an increased direct cost burden for adult cancer survivors compared with adults with no cancer history, while also reporting increased costs associated with missed workdays and employment disability (11). Similarly, in a study of cancer survivors between 2008 and 2011 that stratified survivors by sex, Ekwueme and colleagues found men and women with a history of cancer had excess mean annual expenditures of $4,187 and $3,293, respectively, compared with adults without a history of cancer (25). The same study found a prior cancer diagnosis associated with increased difficulty with completing physical and mental tasks required by their jobs, and nearly a quarter of survivors reporting feeling less productive at work due to their prior cancer diagnosis status, with men having higher per capita productivity losses compared with women (25). In another study that estimated annual out-of-pocket expenditures and financial hardship among non-senior adults with a prior cancer diagnosis, Ekwueme and colleagues found that cancer survivors had higher mean out-of-pocket expenditures compared with those without a history of cancer ($1,000 vs. $622; ref. 25). A study by Park and Look pooled 5 years of MEPS data to compare annual healthcare expenditures and found that cancer survivors had nearly four times higher mean expenditures per person than adults without a history of cancer ($16,346 vs. $4,484; ref. 27). Regarding indirect costs, a study of non-senior cancer survivors by Dowling and colleagues found that cancer survivors experienced greater losses of productivity compared with those with no prior cancer diagnosis (44). However, there was noted variation by cancer type, with breast cancer survivors experiencing less burden compared with those with multiple cancers or short survival cancers (i.e., cancers with a 5-year survival rate; ref. 44). Similar work by Yabroff and colleagues identified cancer survivors as less likely to be employed and unable to work due to health and found adults with a prior cancer diagnosis experienced greater physical limitations due to health and had more work absenteeism compared with those with no prior cancer diagnosis (17).

By comparison, cancer costs in the United States exceed that of Western European nations, with variation by cancer type and disease site (45). Between 1982 and 2010, breast cancer care costs in the United States exceeded that of Western European nations by $435 billion (44). Similarly, colorectal, lung, and prostate cancer costs in the United States exceeded that of Western Europe by $326 billion, $406 billion, and $435 billion, respectively, during the same time-period (45). Between 1983 and 1999, the United States increased spending on cancer care by nearly 50% (from $47,000 per cancer survivor to $70,000), compared with a 16% increase in Western European countries (from $38,000 per cancer case to $44,000; ref. 46). However, patients with cancer in the United States have experienced higher survival gain, compared with European cancer survivors creating $598 billion in net social value for U.S. cancer survivors diagnosed between 1983 and 1999, with the highest social value gains for breast ($173 billion) and prostate ($627 billion) cancers (46). In this study, we also found the trend in total healthcare expenditures to be higher over time for cancer survivors compared with trends in healthcare expenditures for adults without a history of cancer. Total mean healthcare expenditures for cancer survivors rose initially prior to declining shortly during 2012/2013 and rising to its highest amount for 2016. This suggests that cancer survivors have higher and steadily increasing costs, even after the passage of the Affordable Care Act (ACA) in 2010 (45, 47). In the wake of the passage of the ACA, our sample was largely insured. Despite this, survivors still experienced significantly higher healthcare expenditures compared with those with no history of cancer.

Utilization and receipt of different services has been associated with increased expenditures. In the current study, expenditures associated with inpatient, office-based, and outpatient visits and medications remained significantly higher for cancer survivors when compared with those with no cancer history after adjustment. This is supported by evidence presented by Park and Look, wherein adult cancer survivors had higher mean healthcare expenditures that those with no cancer history, with ambulatory care visits accounting for the largest proportion of healthcare expenditures (41%) followed by hospital inpatient (27%), medications (21%), and other services (10%; ref. 27). Soni compared the percentage distribution of expenditures (ambulatory, inpatient, ER, prescriptions, and home health) for cancer treatment for 2001 and 2011 (24). In 2011, ambulatory expenditures comprised nearly 50% of total expenditures for survivors followed by inpatient (35.1%), medications (11.4%), home health (3.3%), and ER (0.6%) (24).

Adults with a prior cancer diagnosis report greater direct and indirect economic hardship when compared with those with no prior cancer diagnosis (25). They report experiences of financial challenges, spending an estimated 20% of their income on medical costs (14). Previous research has identified that female breast cancer survivors and survivors of prostate cancer comprise the largest proportion of cancer survivors due to the high incidence and low mortality of both disease types (4). Further, because cancer survivors are aging and living longer, they are more likely to suffer from additional chronic illnesses associated with age (48). Financial hardship associated with a prior cancer diagnosis may impede their ability to use care. Adults with a prior cancer diagnosis also face increased risk of recurrence and the development of secondary primary cancer, which may further increase their economic burden (49). Policy initiatives are needed to protect survivors from rising costs associated with their cancer diagnosis. Among health practitioners, the promotion of health behaviors to reduce cancer risk is appropriate.

Despite the important findings from this study, there are limitations that must be acknowledged. First, due to limitations in the dataset, we defined a cancer survivor as anyone with a prior cancer diagnosis. Future studies may use a more refined definition of a cancer survivor to include only those in remission or no longer in treatment. Similarly, the results of our study are not presented by cancer type as has been used in previous work; however, our methodology is supported by studies that examined cancer-associated costs among a pooled nationally representative samples (13, 28, 44). We were limited in our ability to assess for all indirect cost variables (i.e., “days spent in bed due to ill-health,” and “missed school days”). “Days spent in bed due to ill-health” is not reported consistently throughout the study years in the dataset, and “missed school days” would have required excluding age from the regression models as only the youngest age groups reported values for this variable. Third, our measure of cancer survivorship and the comorbid conditions are based on self-report, which may introduce sources of bias (50). Because the MEPS dataset contains only the noninstitutionalized U.S. population, it is unclear of how these results may differ with the inclusion of institutionalized U.S. adult cancer survivors.

Our findings are important and provide new information on health care expenditures for adults with a prior cancer diagnosis. In this nationally representative sample of adults, those with a prior cancer diagnosis had significantly higher health care expenditures compared with adults without a prior cancer diagnosis and these excess costs were consistent through time. As cancer survivors in the United States are a rapidly growing population with unique health, social, and economic needs, these findings warrant additional strategies for reducing the cost burden of care among adults with a prior cancer diagnosis. Trends analysis through time, which few studies have done previously, allows for examination of how cancer-related costs and additional burdens respond to changes in healthcare policies and clinical practice.

Our findings have several implications for research and practice. Our findings provide groundwork for additional research that is needed to understand the underlying mechanisms associated with a higher cost burden in adults with a prior cancer diagnosis. Specifically, research is warranted to understand: (i) the types of cancer treatments associated with increased expenditures for adults with a prior cancer diagnosis; (ii) the role of comorbid conditions in increased expenditures among this population, particularly as cancer survivorship becomes more prevalent, and the population continues to age; (iii) the role of costs among survivors by cancer type; and (iv) factors related to cost reduction among healthcare payers. An understanding of these mechanisms will inform implications for clinical practice, which may include a need for increased screening and utilization of primary care services to reduce the burden of comorbid conditions among adults with a prior cancer diagnosis and guidelines for optimizing treatment options and high-quality care for cancer survivorship at costs affordable to those impacted.

L.E. Egede reports grants from NIDDK; and grants from NIMHD during the conduct of the study. No disclosures were reported by the other authors.

S.L. Walker: Conceptualization, formal analysis, writing–original draft, writing–review and editing. J.S. Williams: Conceptualization, formal analysis, writing–review and editing. K. Lu: Data curation, formal analysis, writing–review and editing. A.Z. Dawson: Formal analysis, writing–review and editing. L.E. Egede: Conceptualization, resources, formal analysis, writing–review and editing.

This work was partially supported by the National Institute of Diabetes and Digestive and Kidney Diseases grants K24Dk093699, R01DK118038, and R01DK120861 (principal investigator: L.E. Egede).

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.