Abstract
Background: The treatment of acute myeloid leukemia (AML) has made significant progress in the last 30 years; however, numerous factors affect outcomes in patients with AML. Well-known risk factors are age, cytogenetics, and treatment intensity. The purpose of our study was to investigate the effects of insurance status on the outcome of AML; age, Carlson comorbidity index, distance travelled to the treatment center, and type of treatment center were adjusted by analyzing data from National Cancer Database (NCDB). In the wake of the Affordable Care Act, and its impact on insurance coverage, evaluating the effect having insurance has on health outcome is urgently necessary. Materials and Methods: Data were analyzed from 67,443 men and women (≥18 years of age), who were registered in the NCDB and diagnosed with AML between 1998 and 2011 with follow-up to the end of 2012. The primary predictor variable was payer status, and the outcome variable was overall survival. Additional variables addressed and adjusted, included: sex, age, race, Charleston Comorbidity index, level of education, income, distance traveled, facility type, diagnosing/treating facility, treatment delay, and chemotherapy. Results: In multivariate analysis, after adjusting for other predictor variables, payer status was a statistically significant predictor of overall survival for AML. Relative to privately insured patients, patients with Medicaid had a 17% increased risk, those without insurance had a 21% increased risk, those with Medicare had a 19% increased risk and those with unknown insurance status had a 22% increased risk of mortality from AML. The percentage of patients surviving from AML after 24 months was 37.6%, 31.4%, 32.3%, 31.8%, and 33.1% for patients with private, unknown, Medicare, uninsured, and Medicaid payer status, respectively. All factors investigated were found to be significant predictors of AML survival except distance traveled. Conclusion: We observed that payer status has a statistically significant relationship with overall survival from AML.
The American Cancer Society estimates there will be about 19,950 new cases and 10,430 deaths from acute myeloid leukemia (AML) in 2016, of which most will affect adults (1). AML is typically an older person's disease, being uncommon before the age of 45 years. In 2008, the World Health Organization (WHO) revised the classification of AML for proper prognostication based on morphology, immunophenotyping, cytogenetic data, and molecular studies (2, 3). AML is also divided based on risk into high, intermediate, and low groups (4-7). Recently Arber et al. published the latest classification of AML based on new clinical, prognostic, morphological, immunophenotypic, and genetic data (8).
Complete remission (CR) rates in older adults with AML are 40-60% with treatment (9-18). The 5-year survival rate has increased from 6.3% in 1975 to 23.9% in 2007 according to Surveillance, Epidemiology and End result program (SEER) data analysis. Overall survival rates for AML decrease as age increases (19). Older adults are also more likely to have comorbidities and poorer performance status. This increases treatment-related morbidity and mortality, and limits intensive treatments such as allogenic hematopoietic cell transplantation.
There exist many risk factors associated with survival of patients with AML. Well-known risk factors are age, cytogenetics, molecular markers, and treatment intensity. Other factors, including access to healthcare, modify treatment outcomes. For example, the type of insurance a patient has might influence the cancer treatment outcome.
Previous studies on solid tumors have identified a link between uninsured and underinsured payer status and shorter survival from several cancer types (20-25). However, this relationship has not been demonstrated in patients with AML in whom disease progression is certain, and in the absence of treatment, prognosis is almost always death. Bradley et al. demonstrated uninsured patients were 4.4 times more likely to be untreated than their privately insured counterparts and had a 29% higher likelihood of death. Once treatment was adjusted in the survival analyses, differences between insurance groups were not statistically significant (26). These findings demonstrate the critical role of health insurance in AML, which is a life-threatening disease that is expensive to treat (26-28).
As healthcare reform in the United States continues to evolve, it is challenging to define the impact of payer status on health outcomes. In the infancy of the Affordable Care Act, many expect a more comprehensive coverage will improve outcomes across the United States (29-33). Yet the effect that this shift will have on patients with cancer is still uncertain.
The National Cancer Data Base (NCDB) began collecting information on payer status for all patients in 1996. This dataset provides the opportunity to examine the relationship between payer status and overall survival of patients with cancer. Recent studies using these data found a statistically significant relationship between payer status and survival from breast, colon, and lung cancer (34-36). The impact of payer status on outcomes of AML is unknown. In an analysis of 67,433 patients with AML from the NCDB, we assessed here how payer status affects survival in AML, adjusting for age, comorbidity index, distance travelled from, and type of treatment center.
Materials and Methods
This study examined data regarding 67,443 patients who were diagnosed with AML between 1998 and 2011 and followed-up until December 31, 2012, and they were registered in the de-identified NCDB. Only patients who had complete survival data and complete chemotherapy data (no, single, multiple agent) were included in the analysis. The NCDB captures approximately 70% of all newly diagnosed cases of cancer in the United States at the institutional level. (37) The International Classification of Disease for Oncology, third edition (ICD-O-3), codes (C420, C421, C424) for a diagnosis of AML were used to select patients. The code C420, C421 and C424 represent blood, bone marrow and hematopoetic not otherwise specified. Then we used histological codes 9840, 9861, 9865-9867, 9869, 9871-9874, 9895-9897, 9898, 9910-9911, 9920 for AML.
The outcome variable of overall survival was calculated from date of diagnosis to date of death, date of loss to follow-up, or date of study end (December 31, 2012). The primary predictor variable of payer status was categorized as uninsured, private, Medicaid, Medicare, or unknown. Other variables investigated included sex, age, race, Charlson Comorbidity index, income, education, distance traveled to treating facility, facility type, diagnosing/treating facility, treatment delay, and chemotherapy. Age was grouped as 18-49, 50-64, 65-74, or ≥75 years. Race was categorized as White, Black, or Asian. Charlson Comorbidity Index, a score that indicates the overall health status of a patient, was defined as 0, 1, ≥2, or unknown (38). Income, or median household income at zip code level, was grouped as <$30, $30-34, $35-45, or ≥$46 k. Education measured as the percentage of adults in the patient's zip code who did not graduate from high school was grouped as <14%, 14-19.9%, 20-28.9%, and ≥29%. Education was determined using the 2000 census data. Distance traveled, i.e. the distance from the patient's residential zip code to a medical center, was defined as <30 or ≥30 miles. Facility type was categorized as a community cancer program, a comprehensive cancer program, or an academic or research cancer program. Diagnosing/treating facility was categorized as either the same (diagnosed and treated at the same facility) or different (diagnosed at one facility and treated at another). Treatment delay was grouped as 0-7, 8-30, or ≥31 days. Chemotherapy was categorized as single agent, multiple agents, or not received.
Descriptive statistics are presented for each variable studied. Multivariate Cox regression was used to simultaneously estimate the hazard of death [hazard ratio (HR)) by payer status while adjusting for other factors. Direct adjusted overall survival (OS) was calculated by using Multivariate Cox regression. Statistical Software SAS 9.4 (SAS Inc., Cary, NC, USA) was used for data management, statistical analysis, and modeling. 95% Confidence intervals (CIs) that did not include 1.00 were considered statistically significant at the 5% level.
Results
The demographic characteristics of the patient population are shown in Table I. A total of 67,443 patients were included in the study, 45.12% had Medicare at diagnosis, 40.41% had private insurance, 7.4% had Medicaid, 4.15% were uninsured, and 2.92% had an unknown insurance status.
The overall mean age at diagnosis was 61.1 years. At diagnosis, Medicare patients had, as expected, the highest mean age (73.6 years), while all other privately insured, Medicaid-insured, or uninsured patients had mean ages of 51.9, 44.6, and 46.9 years, respectively. The majority of patients in the study were White (88.1%). Most patients (63.12%) received multiple agent chemotherapy, although 21.49% received none at all.
Table II displays the HR and the 95% CI for each variable from multivariate Cox regression analysis. Payer status was a significant predictor of overall survival after adjusting for gender, age, race, comorbidity, income, education, distance traveled, facility type, treatment delay, diagnosing/treating facility, and chemotherapy. Relative to privately insured patients, Medicaid patients had a 16% increased risk of mortality from AML. Uninsured and Medicare patients had a 21% and a 19% increased risk of mortality, respectively. After 24 months, the percentage of patients surviving from AML was 37.66%, 31.44%, 32.23%, 31.76%, and 33.11% for private, unknown, Medicare, uninsured, and Medicaid payer status, respectively.
Multivariate analysis also revealed that female patients were 8% less likely to die than their male counterparts. Compared to the 18- to 49-year-old age group, patients aged 75 years and older were 4.14-times more likely to die. Relative to Whites, Asians had an 8% reduced risk while Black patients had an 8% increased risk of death. Patients with two or more comorbidities were 1.49-times more likely to die than those without comorbidities. Those who received multiple agent chemotherapy had a 38% reduced risk of death as compared to those who received no chemotherapy. A finding of particular interest was that patients treated within 7 days and 8-30 days were 26% and 18% more likely to die compared to patients treated 31 days or more after diagnosis. Distance traveled was not a significant predictor of overall survival.
Discussion
Payer status has a significant effect on the overall survival of patients with AML after adjusting for all other predictive factors (see Table II). Medicare patients had a worse outcome compared to those privately insured likely due to the older age of these patients. The important comparison is between uninsured, Medicaid, and privately insured patients. Uninsured patients had worse outcomes compared to Medicaid patients, who had worse outcome compared to private insurance. Medicaid and uninsured patients had an increased risk of dying compared to those with private insurance. Earlier studies in different types of cancer (mainly colon, breast, and lung) showed similar although not independent associations of insurance status and survival (21, 28, 34, 36, 39, 40). The mechanism by which payer status affects survival is not entirely clear; it could be mediated through differences in access to certain treatment types (41). Access to care, transportation, and supportive care may be involved. Further research and mediation analysis is needed.
Using the NCDB database of more than 60,000 patients with AML, we confirm and extend here three smaller studies about the impact of insurance on AML outcomes. Ortiz-Ortiz et al. described insurance-related disparities in 516 patients with leukemia in Puerto Rico (including 159 cases of AML), but did not comment on other potential confounding factors (42). Bradley et al. described insurance-related disparities in the Virginia Cancer Registry based on 523 patients. The highest risk of death was seen in uninsured patients (26). Borate et al. investigated SEER data for 5,541 patients with AML aged 19 to 64 years diagnosed between 2007 and 2011, finding an HR of death of 1.24 for Medicaid patients compared to privately insured patients using a multivariate analysis. Our data are comparable to the data of Borate et al. (43). Two smaller studies did not find any correlation between insurance and survival. Moreover, in a single-institutional study, no difference was found between different insurance categories, but survival was low in all categories (40). In a study from the 2002-2006 New York and California Cancer Registries, the researchers found no survival differences. However, the authors compared Medicaid patients with all other categories, including uninsured patients (44, 45).
Similar to other studies, we found patients with older age, and higher comorbidity index had the worst AML survival (9, 19). As shown in Table II, age-related mortality increases to approximately double in the 50-64 years group compared to 18-49 years. Risk of death is even higher in those aged 65 years and older, which may in part be due to the lack of chemotherapy administered.
Studies have shown that comorbidities are predictive of early death in the elderly, but not a predictor for younger patients with AML. The comorbidity index, as established in the transplant setting, is an independent predictor for early death in elderly patients (46-48). Our findings are consistent with these studies and, as demonstrated in Table II, the findings of mortality significantly increasing as the comorbidity index increased. An important question arises: Does the comorbidity index account for AML survival differences for patients with different insurance status? In a different malignancy, colon cancer, differences in comorbidity level did not account for the association between insurance status and survival (49).
An additional finding is patients treated at academic institutions had better OS compared to those treated at non-academic institutions (such as community cancer centers and comprehensive cancer centers). Patients treated at community cancer programs and comprehensive CCPs were found to have a 4-5% higher risk of dying compared to patients treated at an academic center. Similar results were observed for patients with breast cancer (50). Survival outcomes can be affected by distance travelled to treatment center. One single-center study, including 281 patients receiving induction therapy, found no association between socioeconomic status or distance from treatment center and survival. A study of a pediatric population with AML and acute lymphocytic leukemia showed distance from treatment center had no effect on outcomes in AML. Our data concurs with the other studies (51-53).
The Patient Protection and Affordable Care Act (ACA) will likely improve insurance coverage for most young adults in the United States, but some of these patients will face significant premium increase in the individual market. In a SEER data analysis of 39,447 patients aged 20 to 40 years diagnosed with a malignant neoplasm between 2007 and 2009, an association between insurance coverage and decreased likelihood of presentation with metastatic disease [odds ratio (OR), 0.84; 95% CI=0.75-0.94)] increased receipt of definitive treatment (OR=1.95; 95% CI=1.52-2.50), and reduced death resulting from any cause (HR=0.77; 95% CI=0.65-0.91) was noted. The improved coverage fostered by the ACA may translate into better outcomes among young adults with cancer. Extra consideration must be given to ensure that patients who face premium increases in the individual market can obtain insurance coverage under the ACA (54).
Despite utilizing a large sample population, there are limitations to this study. Firstly, the NCDB database does not collect information on cytogenetic or molecular subtypes of leukemia. This is important because different socioeconomic and ethnic groups may have a different biology of leukemia. Secondly, the treatment information is limited to single-agent versus multi-agent chemotherapy. Different patient groups may have received different treatment intensity. In particular, the NCDB database does not have detailed information about allogeneic transplantation. Allogeneic transplantation has a potential for cure; however, if used in patients with comorbidities or without good social support, transplantation may worsen the survival rate. Different insurance plans may limit access to transplantation or provide incomplete coverage for supportive care. Detailed data are not available from the NCDB on changes in treatment and change in insurance status that might have occurred over time. Thirdly, the socioeconomic status and education level is collected by zip code level not by direct patient or family information. Fourthly, the effect of factors on OS may be different from the effect on cause-specific survival.
In conclusion, insurance status is a significant predictor of OS for patients with AML. This remains true after adjusting for other previously described non-biological predictive factors. As we continue to navigate our way through the dynamically changing system of healthcare reform in the United States, it is important to consider the influence of payer status on health outcomes in future decision-making in order to mitigate disparities.
Acknowledgements
The Authors wish to acknowledge the Commission on Cancer of the American College of Surgeons and the American Cancer Society for making public data available through the NCDB. The data used in this study were derived from a de-identified NCDB file. The American College of Surgeons and the Commission on Cancer have not verified and are not responsible for the analytic or statistical methodology employed or the conclusions drawn by the investigator.
Footnotes
Ethics Statement
With the support from the Chair of Louisiana State University Hospital in Shreveport (currently University Health Shreveport) Cancer program, the corresponding author applied and was awarded the NCDB Participant Use Data File (PUF) for 1998 to 2011 from the Commission on Cancer. The PUF is a Health Insurance Portability and Accountability Act compliant data file containing cases submitted to the Commission on Cancer's NCDB. The PUF contains de-identified patient level data that do not identify hospitals, healthcare providers, or patients as agreed to in the Business Associate Agreement each CoC-accredited program has signed with the American College of Surgeons. The PUFs are designed to provide investigators associated with CoC-accredited cancer programs with a data resource they can use to review and advance the quality of care delivered to cancer patients through analyses of cases reported to the NCDB. NCDB PUFs are only available through an application process to investigators associated with CoC-accredited cancer programs.
Competing Interests
The Authors have no competing interests to disclose.
- Received June 20, 2016.
- Revision received July 7, 2016.
- Accepted July 11, 2016.
- Copyright© 2016 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved