Abstract
Study Objective
To compare survival and 5-year mortality, by Medicaid status, in adults diagnosed with 8 select cancers.
Methods
Linking records from the Ohio Cancer Incidence Surveillance System (OCISS) with Ohio Medicaid enrollment data, we identified Medicaid and non-Medicaid patients aged 15–54 years and diagnosed with the following incident cancers in the years 1996–2002: cancer of the testis; Hodgkin’s and non-Hodgkin’s lymphoma; early-stage melanoma, colon, lung, and bladder cancer; or pediatric malignancies (n=12,703). Medicaid beneficiaries were identified in the pre-diagnosis group if they were enrolled in Medicaid at least 3 months before cancer diagnosis, and in the peri/post-diagnosis group if they enrolled in Medicaid upon or after being diagnosed with cancer. We also linked the OCISS with death certificates and data from the U.S. Census. Using Cox and logistic regression analysis, we examined the association between Medicaid status and each of survival and 5-year mortality, respectively, after adjusting for patient covariates.
Results
Nearly 11% of the study population were Medicaid beneficiaries. Of those, 45% were identified in the peri/post-diagnosis group. Consistent with higher mortality, findings from the Cox regression model indicated that compared to non-Medicaid, patients in the Medicaid pre-diagnosis and peri/post-diagnosis groups experienced unfavorable survival outcomes (adjusted hazard ratio (AHR): 1.52, 95% confidence interval (1.27, 1.82), and 2.01 (1.70, 2.38), respectively).
Conclusions
Medicaid status was associated with unfavorable survival, even after adjusting for confounders.
Impact
The findings reflect the vulnerability of Medicaid beneficiaries and possible inadequacies in the process of care.
Keywords: Curable Cancers, Survival Outcomes, Medicaid Status, Linked Databases
Improvement in cancer survival is not shared equally across subgroups of the population1, 2. Racial disparities have been well documented1–7. Although it is generally held that low socio-economic status is associated with disparate outcomes, evidence from population-based databases is lacking, especially for cancers for which 5-year survival is relatively high1, 3, 8–17.
Medicaid beneficiaries constitute a disadvantaged subgroup of the population. In addition to their socioeconomic vulnerabilities, adult beneficiaries who are on Medicaid for reasons other than maternity care, suffer from disabling physical and/or mental ailments.
On the other hand, the Medicaid program may alleviate financial barriers, at least partially, and improve access to health services. In the case of cancer patients, the Medicaid program may serve as a 1) (public) health insurance program when history of enrollment indicates that they had been enrolled in Medicaid for some time prior to cancer diagnosis; or 2) as a safety net program, as may be the case for individuals enrolling in Medicaid upon or after being diagnosed with cancer. Results from previous studies have shown that cancer patients enrolling in Medicaid around the time of cancer diagnosis are significantly more likely to be diagnosed with advanced-stage disease than those enrolled in Medicaid prior to cancer diagnosis18–21.
In this study, we compare survival outcomes for eight potentially curable cancers in adults 15–54 years of age between Medicaid and non-Medicaid populations, using a unique population-based database developed by linking records from the Ohio Cancer Incidence Surveillance System (OCISS), Ohio Medicaid enrollment files, data from the Ohio Death Certificates, and data from the U.S. Census. To distinguish the role of Medicaid as a health insurance versus a safety net program, as described above, we account for the timing of enrollment in Medicaid in relation to cancer diagnosis. We hypothesize that adjusting for patient attributes, as well as for cancer site and stage, Medicaid beneficiaries affected with these cancers experience shorter survival and higher 5-year mortality than their non-Medicaid counterparts. Furthermore, we hypothesize that, among Medicaid beneficiaries, those enrolled in Medicaid prior to cancer diagnosis experience more favorable survival outcomes than those enrolling in Medicaid upon or after being diagnosed with cancer. By identifying survival differentials across subgroups of the population with “curable cancers”, this study will pave the way for future research aimed at identifying specific factors contributing to such disparities, and potentially providing a data-driven foundation for the formulation of interventions to address disparities.
Methods
Data sources
The study uses a population-based database developed by linking records from the Ohio Cancer Incidence Surveillance System (OCISS), Ohio Medicaid enrollment files, and death certificate data, as detailed below. Furthermore, by using geocoded data from the OCISS, we appended data on income and education, at the census tract level, from the U.S. Census. This study was approved by the Case Cancer Institutional Review Board (IRB), the IRB at the Ohio Department of Health, which maintains the OCISS, and the Ohio Department of Job and Family Services, which administers the state Medicaid program.
1) The Ohio Cancer Incidence Surveillance System (OCISS)
Established in 1991, all incident cases of cancer diagnosed in Ohio residents are required to be reported to the OCISS. Exceptions are squamous and basal cell carcinoma of the skin and in situ cancer of the cervix uteri. For our study years, the completeness of the OCISS was reported to be at over 90%22.
The OCISS record carries patient identifiers, including name, social security number (SSN), date of birth, gender, race, county, address of residence, and zip code; as well as cancer-specific data, including date of diagnosis, anatomical site, stage at diagnosis, and tumor markers. Because of missing values in data elements needed to determine cancer stage according to the American Joint Committee on Cancer (AJCC), we relied on the Surveillance, Epidemiology, and End Results (SEER) summary stage (i.e., in situ, local, regional, distant, unstaged/unknown stage).
2) The Ohio Medicaid Enrollment Files
These files carry records for each individual enrolled in the Ohio Medicaid program. In addition to patient identifiers, these records also carry enrollment spans, which we used to characterize the individual’s enrollment history in Medicaid in relation to cancer diagnosis.
3) The Ohio Death Certificate Files
For nearly every decedent who was a resident of the State of Ohio, the death certificate record carries the individual’s identifiers, as well as the date and cause of death.
4) Data from the U.S. Census
The census tract in which the patient resided at that time s/he was diagnosed with cancer is provided in the OCISS data. Using publicly available data from the U.S. Census, we retrieved the median household income and the proportion of adults with high school diploma at the census tract level.
Study population
Our study population included Ohio residents 15–54 years of age and diagnosed in years 1996–2002 with incident cancer in the sites and histology types listed below. Testicular, colon, lung, and bladder cancers were identified using site codes while lymphoma, melanoma, and pediatric malignancies with their International Classification of Diseases for Oncology, Second Edition/Third Edition (ICD-O-2/ICD-O-3) site or histology type codes, as follows:
Testicular (C620 – C629), age 15–50, all stages (excluding unstaged/unknown stage cases).
Lymphoma / Hodgkins (9650 – 9667, 9650 – 9667), age 15–50, stages in situ, local, and regional.
Lymphoma / non-Hodgkins (9590 – 9596, 9670 – 9729), age 15–50, stages in situ, local, and regional.
Colon (C180 – C189) age 25–54, stages in situ, local, and regional.
Lung (C340 – C349), age 30–50, stages in situ, local.
Bladder (C670 – C679), age 45–50, stages in situ, local, and regional.
Pediatric malignancies (8900 – 8905, 8910, 8912, 8920, 8921, 8960, 8991, 9071, 9180 – 9187, 9192 – 9195, 9200, 9260, 9310, 9362, 9363, 9364, 9364, 9365, 9470, 9471, 9473, 9474, 9490, 9500, 9506, 9510 – 9514, 9522, 9826, 9835), age 20–50, stages in situ, local, and regional.
Melanoma (8720 – 8790), age 20–50, stages in situ, local.
Data linkage
Records for individuals identified through the OCISS were linked with records from each of the Medicaid enrollment files and death certificate files, employing the following multistep deterministic algorithm used in previous studies21, 23, 24:
Step 1: SSN, Last Name, First Name, Sex
Step 2: SSN, Last Name, Date of Birth (Month), Sex
Step 3: SSN, First Name, Date of Birth (Month), Sex
Step 4: First Name, Last Name, Date of Birth (Month and Year), Sex
Of cases who were successfully identified in both the OCISS and Medicaid enrollment files, 83.9% were identified through Step 1, and 10.7% were identified through Step 4. The remaining 5.4% were identified through Steps 2 and 3.
Variables of interest
Outcome variables
Survival time, defined as the time elapsed between the date of diagnosis and date of death. For cancer specific survival, the outcome was the time between diagnosis and death from one of the curable cancers. Those who did not die by December 31st, 2007 were censored on this date.
Five-year mortality: a binary variable (0/1), defined as 1 if the individual died within 5 years of cancer diagnosis, and 0 otherwise. For cancer specific five-year mortality patients dying of any of the curable cancers were assigned a 1 and 0 otherwise.
Independent variables
Medicaid status, the main independent variable, was ascertained through the linkage of the OCISS and Medicaid enrollment files. Further, using the enrollment spans, we constructed beneficiaries’ enrollment history in Medicaid relative to their date of cancer diagnosis. This strategy was aimed at distinguishing between beneficiaries using the Medicaid program as a health insurance program from those who resort to Medicaid as a safety net program25. Thus, the following variables were developed:
-
-
Medicaid pre-diagnosis, a binary (0/1) variable, defined as ‘1’ if the individual had been enrolled in Medicaid at least 3 months prior to cancer diagnosis, ‘0’ otherwise.
-
-
Medicaid peri-/post-diagnosis, a binary (0/1) variable, defined as ‘1’ if the individual enrolled in Medicaid in the 3-month window prior to cancer diagnosis, upon cancer diagnosis, or after being diagnosed with cancer, ‘0’ if enrolled in Medicaid more than 3 months before but within the calendar year of cancer diagnosis.
Demographic variables, retrieved from the OCISS, included age, race, sex, and marital status. Patients were grouped in the following age categories: 15–29, 30–39, 40–49, 50–54. We further identified our subjects as male or female; African American, Caucasians, or ‘All Others’; and as married, non-married, or with unknown marital status.
Cancer stage, based on SEER summary stage, was categorized as in situ, local, and regional or distant stages.
County of residence at the time of diagnosis, as recorded in the OCISS, was categorized as follows: Appalachian, Rural, Metro, and Suburban.
Median household income and educational attainment measures were obtained at the census tract level. Income was grouped in quartiles, based on its distribution across the state. Educational attainment reflected the proportion of adults with a high school diploma. Relying on the distribution of this proportion across the state, we flagged low educational attainment if individuals resided in census tracts in which the proportion of adults with high school was less than 10%.
Data Analysis
In addition to detailed descriptive analysis, proportions were compared across the Medicaid pre, peri-/post, and non-Medicaid groups using chi-square tests. Moreover, Kaplan-Meier curves were produced for the entire study population, as well as for each anatomic cancer site, to compare the survival curves between the Medicaid pre-diagnosis and peri/post-diagnosis groups and non-Medicaid patients. Multivariable survival and logistic regression models were developed to analyze the association between Medicaid status and the outcomes of interest after adjusting for patient demographics, income, geographic area of residence, and cancer site and stage at diagnosis. SAS version 9.2 (Cary, NC) was used in all of our analyses.
Results
Our study population included 11,358 non-Medicaid individuals and 1,345 Medicaid beneficiaries diagnosed with the relevant cancers (Table 1). Of Medicaid beneficiaries, 755 (56%) were identified in the pre-diagnosis group.
Table 1.
Distribution of the Study Population by Medicaid Status, Demographics, Income, Education, County of Residence, and Anatomic Cancer Site/Type
| Variable of Interest | Non- Medicaid N (% of total) |
Medicaid N (% of total) |
Medicaid, pre- cancer diagnosis N (% of Total) |
Medicaid, peri/post cancer diagnosis N (% of Total) |
|---|---|---|---|---|
| Age: | ||||
| 15–29 | 1643 (14.5) | 307 (22.8) | 147 (19.5) | 160 (27.1) |
| 30–39 | 2844 (25.0) | 340 (25.3) | 180 (23.8) | 160 (27.1) |
| 40–49 | 5133 (45.2) | 530 (39.4) | 316 (41.9) | 214 (36.3) |
| 50–54 | 1738 (15.3) | 168 (12.5) | 112 (14.8) | 56 (9.5) |
| Race: | ||||
| African-American | 564 (5.0) | 241 (17.9) | 147 (19.5) | 94 (15.9) |
| Caucasian | 10183 (89.7) | 1076 (80.0) | 593 (78.5) | 483 (81.9) |
| All Other | 611 (5.3) | 28 (2.1) | 15 (2.0) | 13 (2.2) |
| Sex:* | ||||
| Male | 6427 (56.6) | 750 (55.8) | 360 (47.7) | 390 (66.1) |
| Female | 4931 (43.4) | 595 (44.2) | 395 (52.3) | 200 (33.9) |
| Median Household Income | ||||
| Quartile 1 | 2002 (17.6) | 669 (49.7) | 396 (52.5) | 273 (46.3) |
| Quartile 2 | 2765 (24.3) | 362 (26.9) | 215 (28.5) | 147 (24.9) |
| Quartile 3 | 3101 (27.3) | 204 (15.2) | 95 (12.6) | 109 (18.5) |
| Quartile 4 | 3490 (30.7) | 110 (8.2) | 49 (6.5) | 61 (10.3) |
| Education | ||||
| Less than High school | 2047 (18.0) | 624 (46.4) | 369 (48.9) | 255 (43.2) |
| High school and more | 9311 (82.0) | 721 (53.6) | 386 (51.1) | 335 (56.8) |
| Marital Status: | ||||
| Married | 5746 (50.6) | 313 (23.3) | 145 (19.2) | 168 (28.5) |
| Non-Married | 2660 (23.4) | 791 (58.8) | 468 (62.0) | 323 (54.8) |
| Unknown | 2952 (26.0) | 241 (17.9) | 142 (18.8) | 99 (16.8) |
| County of Residence: | ||||
| Appalachian | 1357 (12.0) | 267 (19.9) | 157 (20.8) | 110 (18.6) |
| Metro | 6286 (55.3) | 741 (55.1) | 423 (56.0) | 318 (53.9) |
| Rural | 1554 (13.7) | 156 (11.6) | 74 (9.8) | 82 (13.9) |
| Suburban | 2161 (19.0) | 181 (13.5) | 101 (13.4) | 80 (13.6) |
| Anatomic Cancer Site: | ||||
| Bladder | 577 (5.1) | 66 (4.9) | 39 (5.2) | 27 (4.6) |
| Colon | 2500 (22.0) | 329 (24.5) | 202 (26.8) | 127 (21.5) |
| Hodgkin’s Lymphoma | 739 (6.5) | 148 (11.0) | 64 (8.5) | 84 (14.2) |
| Non-Hodgkin’s Lymphoma | 967 (8.5) | 207 (15.4) | 99 (13.1) | 108 (18.3) |
| Lung | 418 (3.7) | 134 (10.0) | 88 (11.7) | 46 (7.8) |
| Melanoma | 4415 (38.9) | 208 (15.5) | 155 (20.5) | 53 (9.0) |
| Pediatric Malignancies | 118 (1.0) | 36 (2.7) | 11 (1.5) | 25 (4.2) |
| Testis | 1624 (14.3) | 217 (16.1) | 97 (12.9) | 120 (20.3) |
| Cancer Stage: | ||||
| In Situ | 1983 (17.5) | 85 (6.3) | 72 (9.5) | 13 (2.2) |
| Local | 6707 (59.1) | 698 (51.9) | 441 (58.4) | 257 (43.6) |
| Regional or Distant | 2668 (23.5) | 562 (41.8) | 242 (32.1) | 320 (54.2) |
| Five-year Cancer Specific Death | ||||
| Died of cancer in 5 years | 953 (8.4) | 295 (21.9) | 137 (18.2) | 158 (26.8) |
| Did not die of cancer in 5 years | 10405 (91.6) | 1050 (78.1) | 681 (81.9) | 432 (73.2) |
| TOTAL | 11358 (100.0) | 1345 (100.0) | 755 (100) | 590 (100) |
p = 0.56, comparing sex and Medicaid/non-Medicaid.
All other comparisons are significant at p < 0.0001
Across the comparisons groups, we note significant differences in the distribution of the population by the variables of interest. We note a greater representation of younger individuals, African Americans, and non-married patients in the Medicaid than in the non-Medicaid group. In characterizing the income level and educational attainment in our study population, we note that while 41.9% of non-Medicaid patients resided in census tracts with median household incomes in the lowest two quartiles, 81.0% and 71.2% of patients in the Medicaid pre-diagnosis and peri/post-diagnosis groups did so, respectively. As well, the proportion of patients residing in census tracts with low educational attainment was significantly higher in the Medicaid than in non-Medicaid groups. With respect to geographical distribution, we note a greater proportion of Medicaid patients in Appalachian Ohio (20.8% and 18.6% in each of Medicaid pre-diagnosis and Medicaid peri/post-diagnosis, respectively), only 12% of non-Medicaid patients resided in the disadvantaged areas. Conversely, a greater proportion of non-Medicaid patients resided in Suburban counties.
Comparing the distribution of patients by cancer site across Medicaid and non-Medicaid groups, we note greater proportions of patients with Hodgkin’s and non-Hodgkin’s lymphoma, as well as lung cancer among Medicaid than in non-Medicaid patients. With regard to the distribution by stage, we note the significantly higher proportion of patients presenting with regional- or distant-stage cancer in the Medicaid than in the non-Medicaid population (41.8% vs. 23.5%). We also highlight the differences in stage distribution within the Medicaid population, between the pre- and the peri/post-diagnosis groups, with the proportion of patients presenting with advanced-stage disease being much higher in the latter than in the former group (54.2% and 32.1%, respectively). Finally, the proportion of patients deceased by the end of the 5-year follow-up period was 8.4% in non-Medicaid patients, compared to 18.2% and 26.8% among Medicaid pre-diagnosis and Medicaid peri/post-diagnosis patients, respectively. All of the above comparisons were significant at p < 0.0001.
Figures 2A–2H show that the survival curves differ significantly across the study groups for all but pediatric malignancies, with survival being better in non-Medicaid than in Medicaid patients. However, we note interesting differences across the anatomic cancer sites by whether survival is worse for the Medicaid pre-diagnosis or the Medicaid peri/post-diagnosis groups. In patients diagnosed with Hodgkin’s lymphoma, colon cancer, melanoma, or testicular cancer, survival was worse for patients in the Medicaid peri/post-diagnosis group than for patients in the Medicaid pre-diagnosis group. Conversely, among bladder cancer patients we note better survival in the Medicaid peri/post-diagnosis group than in their pre-diagnosis counterparts. In patients with lung cancer or non-Hodgkin’s lymphoma, survival was almost the same for Medicaid pre-diagnosis and for Medicaid peri/post-diagnosis patients.
Figure 2. Disease-Specific Survival of by Medicaid Status – Cancer Site Specific.
Figure 2A
Bladder
Log-Rank = 15.3, p = 0.0005
Figure 2B
Colon
Log-Rank = 47.2, p < .0001
Figure 2C
Hodgkin’s Lymphoma
Log-Rank = 13.1, p = 0.0014
Figure 2D
Non Hodgkin’s Lymphoma
Log-Rank = 24.2, p < .0001
Figure 2E
Lung
Log-Rank = 14.1, p = 0.0009
Figure 2F
Melanoma
Log-Rank = 55.7, p < .0001
Figure 2G
Pediatric Malignancies
Log-Rank = 3.1, p = 0.2139
Figure 2H
Testicular
Log-Rank = 138.7, p < .0001
The results of the multivariable logistic regression analysis and survival models are presented in Table 2. Adjusting for patient demographics, marital status, census-tract income and educational attainment, county of residence, and anatomic cancer site, Medicaid patients were significantly more likely to experience unfavorable survival outcomes. Furthermore, the results were more accentuated for those in the Medicaid peri/post-diagnosis group than for those in the Medicaid pre-diagnosis group. With respect to 5-year mortality, patients in the Medicaid in the pre-diagnosis and peri/post-diagnosis groups were more than 1.58 and 2.43 times as likely as their non-Medicaid counterparts respectively to die within 5 years of diagnosis (adjusted odds ratio (AOR) and 95% Confidence Interval: 1.58 (1.25, 1.99) and 2.43 (1.94, 3.04)). Similar patterns were observed for hazard ratios.
Table 2.
Results from the Multivariable Analysis
| Variable of Interest | 5-year Disease-specific Mortality, Adjusted Odds Ratio (95% Confidence Interval) |
Disease-specific Mortality Through Dec 31st 2007, Adjusted Hazard Ratio (95% Confidence Interval) |
|---|---|---|
| Medicaid Status: | ||
| Non-Medicaid (ref) | - | - |
| Medicaid-Pre cancer diagnosis | 1.58 (1.25 1.99)c | 1.52 (1.27 1.82)c |
| Medicaid-Peri/Post cancer diagnosis | 2.43 (1.94 3.04)c | 2.01 (1.70 2.38)c |
| Age: | ||
| 15–29 (ref) | - | - |
| 30–39 | 1.23 (0.94 1.61) | 1.16 (0.93 1.45) |
| 40–49 | 1.59 (1.23 2.07)b | 1.50 (1.22 1.86)b |
| 50–54 | 1.53 (1.14 2.05)a | 1.52 (1.20 1.93)b |
| Race: | ||
| African-American | 1.01 (0.80 1.26) | 0.98 (0.82 1.16) |
| Caucasian (ref) | - | - |
| All Other | 0.49 (0.31 0.78)a | 0.55 (0.38 0.79)a |
| Sex: | ||
| Male | 1.40 (1.23 1.60)c | 1.36 (1.22 1.51)c |
| Female (ref) | - | - |
| Median Household Income | ||
| Quartile 1 | - | - |
| Quartile 2 | 0.93 (0.77 1.13) | 1.01 (0.87 1.17) |
| Quartile 3 | 0.92 (0.74 1.15) | 0.93 (0.79 1.11) |
| Quartile 4 | 0.77 (0.61 0.97)a | 0.80 (0.67 0.97)a |
| Education | ||
| Less than High school | 1.05 (0.88 1.26) | 1.04 (0.90 1.19) |
| High school and more | - | - |
| Marital Status: | ||
| Married (ref) | - | - |
| Non-Married | 1.24 (1.06 1.44)a | 1.22 (1.08 1.38)a |
| Unknown | 0.95 (0.79 1.14) | 0.94 (0.82 1.09)a |
| County of Residence: | ||
| Appalachian | - | - |
| Metro | 1.18 (0.96 1.45) | 1.16 (0.99 1.37) |
| Rural | 1.31 (1.02 1.69)a | 1.33 (1.09 1.62)a |
| Suburban | 1.24 (0.97 1.58) | 1.18 (0.97 1.43) |
| Anatomic Cancer Site: | ||
| Bladder | 7.79 (5.32 11.42)c | 7.37 (5.35 10.16)c |
| Colon | 4.57 (3.39 6.16)c | 4.63 (3.56 6.02)c |
| Hodgkin’s Lymphoma | 1.16 (0.79 1.72) | 1.41 (1.01 1.98)a |
| Non-Hodgkin’s Lymphoma | 3.96 (2.90 5.39)c | 4.26 (3.24 5.59)c |
| Lung | 23.52 (16.64 33.23)c | 19.28 (14.38 25.85)c |
| Melanoma | 2.69 (1.94 3.74)c | 3.22 (2.43 4.28)c |
| Pediatric Malignancies | 12.06 (7.71 18.86)c | 9.60 (6.70 13.74)c |
| Testis (ref) | - | - |
| Cancer Stage: | ||
| In Situ (ref) | - | - |
| Local | 4.49 (3.07 6.56)c | 3.89 (2.87 5.29)c |
| Regional or Distant | 17.26 (11.58 25.73)c | 12.77 (9.25 17.63)c |
0.01 <= p < 0.05,
0.001 <= p < 0.01,
p < 0.001.
All other comparisons are not significant at p < 0.05
Additional notable differences include
First, men were more likely than women to experience unfavorable survival outcomes. Second, compared to married patients, survival is worse among patients who are not married. We also note great variations in survival outcomes across anatomic cancer sites. Finally, compared to patients with testicular cancer, those with other cancers were significantly more likely to experience 5-year mortality and increased hazard of death over a median follow up of 7.6 years.
Discussion
This study documents the presence of survival disparities by Medicaid status in patients diagnosed with cancers for which 5-year survival is relatively high when adequate treatment is received. These disparities persist even after adjusting for patient demographics, marital status, county of residence, and for census tract level income and educational attainment.
Numerous factors can explain these disparities. First, it is possible that Medicaid beneficiaries are initiating treatment late, and/or receiving inadequate treatment. Our data sources in this study preclude us from determining whether treatment-related factors are associated with the differences in survival outcomes between Medicaid and non-Medicaid patients. Second, poor survival associated with Medicaid may be a reflection of the patients’ high level of vulnerability, as adults enrolled in the Medicaid program are likely to be disabled, presenting with psychiatric and/or physical comorbidities. As well, although financial barriers may be somewhat reduced with their enrollment in the Medicaid program, these patients encounter various barriers (e.g., transportation, poor psychosocial support) that may hinder receipt of adequate treatment and follow-up care.
To our knowledge, this is the first study to analyze survival outcomes by Medicaid status for these cancers, many of which have been largely absent from the disparities literature. The major strength of our study lies in our use of linked databases consisting of the Ohio Cancer Incidence Surveillance system, Medicaid enrollment files, and death certificate files. In addition, rather than accounting for Medicaid status in a dichotomous (yes/no) fashion, we identified Medicaid beneficiaries in the Medicaid pre-diagnosis and peri/post-diagnosis groups to distinguish those using the Medicaid program as a health insurance program from individuals resorting to Medicaid as a safety net program, respectively. Prior studies using a similar approach have reported more favorable stage outcomes in patients enrolled in Medicaid prior to cancer diagnosis as compared to those enrolling in Medicaid upon or after being diagnosed with cancer, but only for breast, colorectal, lung, and cervical cancer patients18–21. In addition, longer length of enrollment in Medicaid has been shown to be associated with a greater likelihood to receive screening mammography26. Together, these findings suggest that the benefit of being in the pre-diagnosis group may be associated with a greater ability on the part of the beneficiary to develop a network of providers and an ability to navigate the system. However, this remains to be demonstrated in empirical studies. In this study, adjusting for cancer stage, favorable survival outcomes in the pre-diagnosis group were observed for most (early stage colon cancer and melanoma, Hodgkin’s lymphoma, and all stages of testicular cancer), but not for all cancers. In patients with non-Hodgkin’s lymphoma and early stage lung cancer, outcomes for the pre-diagnosis and the peri/post-diagnosis groups were almost the same (Figures 2D and 2E). In bladder cancer patients, those in the pre-diagnosis group experienced worse survival than those in the peri/post-diagnosis group (Figure 2A). To better inform policy, additional studies should elucidate the factors associated with these variations by anatomic cancer site.
Given the following limitations, our findings should be interpreted with caution
First, our data sources preclude us from determining the adequacy of the treatment received, or whether treatment was received in a delayed fashion. As well, we were unable to determine whether the patient experience any cancer relapse.
Second, our income and education measures are at the census tract, and not the individual level. The availability of individual-level data on income and educational attainment would have been desirable; in the absence of such measures, however, census tract-level measures have been widely used to obtain a proxy of individuals’ socio-economic status 27–29.
Third, we note that since our data are specific to Ohio, our results may not be generalizable to patients elsewhere in the United States.
In closing, our study documented important disparities in survival outcomes for the relevant cancers. These disparities persisted after adjusting for patient demographics, marital status, and county of residence. Future studies should focus on the differentials in the vulnerability of the Medicaid and non-Medicaid populations, and elucidate delays and/or inadequacies in the process of care that could explain unfavorable outcomes among Medicaid beneficiaries.
Figure 1. Disease-Specific Survival of by Medicaid Status – All Cancer Sites Combined.
KM All
Log-Rank = 323.2, p < .0001
Acknowledgments
The authors wish to thank Ms. Georgette Haydu of the Ohio Department of Health, which maintains the Ohio Cancer Incidence Surveillance System, and Mr. James Gearheart of the Ohio Department of Job and Family Services, which administers the Ohio Medicaid program, for their careful review of the manuscript.
Research Support:
National Cancer Institute (R03 CA136064, to Dr. Koroukian)
Dr. Koroukian is also supported by the Case Western Reserve University/Cleveland Clinic CTSA Grant Number UL1 RR024989 from the National Center for Research Resources (NCRR), a component of the National Institutes of Health and NIH roadmap for Medical Research. Its contents are solely the responsibility of the authors and do not necessarily represent the official view of NCRR or NIH. Dr. Raghavan was supported by the M. Frank Rudy & Margaret Domiter Rudy Chair in Translational Cancer Research.
Footnotes
Results were presented in part at the American Association for Cancer Research (AACR) meeting, in Miami, Florida, October 2010.
References
- 1.Jemal A, Clegg LX, Ward E, et al. Annual report to the nation on the status of cancer 1975–2001, with a special feature regarding survival. Cancer. 2004 Jul 1;101(1):3–27. doi: 10.1002/cncr.20288. [DOI] [PubMed] [Google Scholar]
- 2.Edwards BK, Brown ML, Wingo PA, et al. Annual report to the nation on the status of cancer-1975–2002, featuring population-based trends in cancer treatment. J Natl Cancer Inst. 2005 Oct 5;97(19):1407–1427. doi: 10.1093/jnci/dji289. [DOI] [PubMed] [Google Scholar]
- 3.Jemal A, Siegel R, Xu J, Ward E. Cancer statistics, 2010. CA Cancer J Clin. 2010 Sep-Oct;60(5):277–300. doi: 10.3322/caac.20073. [DOI] [PubMed] [Google Scholar]
- 4.Hayat MJ, Howlader N, Reichman ME, Edwards BK. Cancer statistics, trends, and multiple primary cancer analyses from the Surveillance, Epidemiology, and End Results (SEER) Program. Oncologist. 2007 Jan;12(1):20–37. doi: 10.1634/theoncologist.12-1-20. [DOI] [PubMed] [Google Scholar]
- 5.President’s Cancer Panel 2009–2010 Annual Report: America’s Demographic and Cultural Transformation – Implications for Cancer. US Dept of Health and Human Services, NIH, NCI; 2011. [Google Scholar]
- 6.Goss E, Lopez AM, Brown CL, Wollins DS, Brawley OW, Raghavan D. American society of clinical oncology policy statement: disparities in cancer care. J Clin Oncol. 2009 Jun 10;27(17):2881–2885. doi: 10.1200/JCO.2008.21.1680. [DOI] [PubMed] [Google Scholar]
- 7.Raghavan D. Disparities in cancer care: challenges and solutions. Oncology (Williston Park) 2007 Apr;21(4):493–496. discussion 499, 503, 506. [PubMed] [Google Scholar]
- 8.Garner MJ, Turner MC, Ghadirian P, Krewski D. Epidemiology of testicular cancer: an overview. Int J Cancer. 2005 Sep 1;116(3):331–339. doi: 10.1002/ijc.21032. [DOI] [PubMed] [Google Scholar]
- 9.Mirabello L, Troisi RJ, Savage SA. Osteosarcoma incidence and survival rates from 1973 to 2004: data from the Surveillance, Epidemiology, and End Results Program. Cancer. 2009 Apr 1;115(7):1531–1543. doi: 10.1002/cncr.24121. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Izawa JI, Al-Omar M, Winquist E, et al. Prognostic variables in adult Wilms tumour. Can J Surg. 2008 Aug;51(4):252–256. [PMC free article] [PubMed] [Google Scholar]
- 11.Jawad MU, Haleem AA, Scully SP. Malignant sarcoma of the pelvic bones: treatment outcomes and prognostic factors vary by histopathology. Cancer. 2011 Apr 1;117(7):1529–1541. doi: 10.1002/cncr.25684. [DOI] [PubMed] [Google Scholar]
- 12.Ferrari A, Dileo P, Casanova M, et al. Rhabdomyosarcoma in adults. A retrospective analysis of 171 patients treated at a single institution. Cancer. 2003 Aug 1;98(3):571–580. doi: 10.1002/cncr.11550. [DOI] [PubMed] [Google Scholar]
- 13.Rieken S, Mohr A, Habermehl D, et al. Outcome and Prognostic Factors of Radiation Therapy for Medulloblastoma. Int J Radiat Oncol Biol Phys. 2011 Feb 22; doi: 10.1016/j.ijrobp.2010.12.042. [DOI] [PubMed] [Google Scholar]
- 14.Riffaud L, Saikali S, Leray E, et al. Survival and prognostic factors in a series of adults with medulloblastomas. J Neurosurg. 2009 Sep;111(3):478–487. doi: 10.3171/2009.1.JNS081004. [DOI] [PubMed] [Google Scholar]
- 15.Lai R. Survival of patients with adult medulloblastoma: a population-based study. Cancer. 2008 Apr 1;112(7):1568–1574. doi: 10.1002/cncr.23329. [DOI] [PubMed] [Google Scholar]
- 16.Jemal A, Siegel R, Xu J, Ward E. Cancer Statistics, 2010. CA Cancer J Clin. 2010 Jul 7; doi: 10.3322/caac.20073. [DOI] [PubMed] [Google Scholar]
- 17.Narayanan S, Shami PJ. Treatment of acute lymphoblastic leukemia in adults. Crit Rev Oncol Hematol. 2011 Feb 24; doi: 10.1016/j.critrevonc.2011.01.014. [DOI] [PubMed] [Google Scholar]
- 18.Bradley CJ, Given CW, Roberts C. Late stage cancers in a Medicaid-insured population. Med Care. 2003 Jun;41(6):722–728. doi: 10.1097/01.MLR.0000065126.73750.D1. [DOI] [PubMed] [Google Scholar]
- 19.O'Malley CD, Shema SJ, Clarke LS, Clarke CA, Perkins CI. Medicaid status and stage at diagnosis of cervical cancer. Am J Public Health. 2006 Dec;96(12):2179–2185. doi: 10.2105/AJPH.2005.072553. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Perkins CI, Wright WE, Allen M, Samuels SJ, Romano PS. Breast cancer stage at diagnosis in relation to duration of medicaid enrollment. Med Care. 2001 Nov;39(11):1224–1233. doi: 10.1097/00005650-200111000-00009. [DOI] [PubMed] [Google Scholar]
- 21.Koroukian SM. Assessing the effectiveness of Medicaid in breast and cervical cancer prevention. J Public Health Manag Pract. 2003 Jul-Aug;9(4):306–314. doi: 10.1097/00124784-200307000-00009. [DOI] [PubMed] [Google Scholar]
- 22.The Ohio Department of Health. Ohio Cancer Incidence Surveillance System. Cancer incidence among Ohio residents, 2000. Columbus, Ohio: 2004. Apr, [Google Scholar]
- 23.Koroukian SM, Cooper GS, Rimm AA. Ability of Medicaid claims data to identify incident cases of breast cancer in the Ohio Medicaid population. Health Serv Res. 2003 Jun;38(3):947–960. doi: 10.1111/1475-6773.00155. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Koroukian SM. Linking the Ohio Cancer Incidence Surveillance System with Medicare, Medicaid, and clinical data from home health care and long term care assessment instruments: Paving the way for new research endeavors in geriatric oncology. J Registry Manag. 2008;35:156–165. [PMC free article] [PubMed] [Google Scholar]
- 25.Koroukian SM, Bakaki PM, Beaird H, Earle CC, Cooper GS. Cancer stage comparison between dual Medicare-Medicaid eligibles using Medicaid as a supplemental health insurance program and low-income duals. J Public Health Manag Pract. doi: 10.1097/PHH.0b013e31821a3f8c. In Press. [DOI] [PubMed] [Google Scholar]
- 26.Koroukian SM. Screening mammography was used more, and more frequently, by longer than shorter term Medicaid enrollees. J Clin Epidemiol. 2004 Aug;57(8):824–831. doi: 10.1016/j.jclinepi.2004.01.004. [DOI] [PubMed] [Google Scholar]
- 27.Krieger N. Socioeconomic data in cancer registries. Am J Public Health. 2001 Jan;91(1):156–157. doi: 10.2105/ajph.91.1.156. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Krieger N, Quesenberry C, Jr, Peng T, et al. Social class, race/ethnicity, and incidence of breast, cervix, colon, lung, and prostate cancer among Asian, Black, Hispanic, and White residents of the San Francisco Bay Area 1988-92 (United States) Cancer Causes Control. 1999 Dec;10(6):525–537. doi: 10.1023/a:1008950210967. [DOI] [PubMed] [Google Scholar]
- 29.Krieger N. Overcoming the absence of socioeconomic data in medical records: validation and application of a census-based methodology. Am J Public Health. 1992 May;82(5):703–710. doi: 10.2105/ajph.82.5.703. [DOI] [PMC free article] [PubMed] [Google Scholar]