Abstract
Purpose:
There is a dearth of studies on cancer outcomes in individuals with mental illness. We compared breast cancer outcomes in Medicaid beneficiaries with and without mental illness.
Methods:
Using records from the 1996 to 2005 Ohio Cancer Incidence Surveillance System (OCISS) and Medicaid files, we identified fee-for-service women age < 65 years diagnosed with incident invasive breast cancer who had enrolled in Medicaid ≥ 3 months before cancer diagnosis (n = 2,177). We retrieved cancer stage, patient demographics, and county of residence from the OCISS. From Medicaid claims data, we identified breast cancer treatment based on procedure codes and mental illness status based on diagnosis codes, prescription drugs dispensed, and service codes. We developed logistic regression models to examine the association between mental illness, cancer stage, and treatment for locoregional disease, adjusting for potential confounders.
Results:
Women with mental illness represented 60.2% of the study population. Adjusting for potential confounders, women with mental illness were less likely than those without mental illness to have unstaged or unknown-stage cancer (adjusted odds ratio [OR], 0.61; 95% CI, 0.44 to 0.86; P = .005) or to be diagnosed with distant-stage cancer (adjusted OR, 0.59; 95% CI, 0.40 to 0.85; P = .005). We observed no difference by mental illness status in receipt of definitive treatment (adjusted OR, 1.04; 95% CI, 0.84 to 1.29; P = .08).
Conclusion:
Among Ohio Medicaid beneficiaries, women with mental illness did not experience disparities in breast cancer stage or treatment of locoregional disease. These findings may reflect the equalizing effects of Medicaid through vulnerable individuals' improved access to both physical and mental health care.
Introduction
Patients with mental illness bear increased risk for comorbid chronic physical conditions such as diabetes, cardiovascular disease, and cancer.1–3 Despite the well-documented excess morbidity and premature mortality among such patients,4–13 delayed diagnoses and suboptimal management of these conditions are common.2 Patients with mental illness often present with acute or advanced stages of physical disease. Potential contributors to delayed diagnosis include under-reporting of physical signs and symptoms by the patient and physician misinterpretation of physical complaints as psychosomatic symptoms.2 Once diagnosed with medical comorbidities, patients with mental illness are more likely to experience barriers to adequate treatment.1,14,15 These barriers may be related to the health care system (eg, lack of access to health care, fragmented health care system, or stigmatization by health care providers) and to the patients themselves, as a result of poor communications skills and poor adherence to recommended treatment.1,14,16,17
Higher breast cancer incidence and mortality have been documented in women with severe mental illness compared with their healthier counterparts.18 However, less is known about differences in breast cancer stage at which women with or without mental illness are diagnosed. In addition, to our knowledge, only one study has compared breast cancer treatment patterns between women with and without mental illness.19 Findings from this study documented a lower likelihood to undergo breast-conserving surgery and radiation therapy in women with mental illness versus those without.
This study aimed to compare breast cancer outcomes between Medicaid beneficiaries with and without mental illness. Given the lower rates of screening mammography,20,21 we hypothesized that women with mental illness would be more likely than those without to be diagnosed with advanced-stage cancer. Moreover, because of the barriers to optimum health care, we hypothesized that mental illness would be associated with a decreased likelihood to receive definitive cancer treatment for locoregional disease.
Methods
We conducted a cross-sectional analysis using linked Ohio Cancer Incidence Surveillance System (OCISS) and Medicaid files. This study was approved by the institutional review boards at Case Western Reserve University and the Ohio Departments of Health, as well as by the Ohio Department of Medicaid.
Data Sources
OCISS.
By law, except for in situ cervical cancer and nonmelanoma cancers of the skin, all incident cases of cancer must be reported to the OCISS. In addition to patient identifiers, which were used to link OCISS and Medicaid records, the OCISS carries each patient's county of residence, marital status, anatomic cancer site, date of cancer diagnosis, and cancer stage. Because of high proportions of missing values in the variables documenting tumor size, number of positive lymph nodes, and metastatic status, we used the SEER summary stage instead of the stage classification by the American Joint Commission on Cancer.
Ohio Medicaid enrollment and claims files.
For each Medicaid enrollee, the enrollment file includes a record carrying patient identifiers, demographics, and monthly variables indicating the individual's enrollment in the Medicaid program, as well as the eligibility category. Thus, we summarized beneficiaries' enrollment history in relation to cancer diagnosis and created indicators for eligibility, including those with disabilities and those eligible for dual Medicare-Medicaid and/or spend-down program participation.
The claims files include claims for all services reimbursed by Medicaid. Diagnosis codes—documented according to the International Coding of Diseases (ninth revision), Clinical Modification (ICD-9-CM)—and procedural codes—documented according to the ICD-9-CM, Current Procedural Terminology (ed 4), or Healthcare Common Procedure Coding System—were used to flag the presence of mental illness and/or comorbid conditions and to ascertain receipt of definitive treatment.
Data Linkage
Consistent with prior studies by our group,22–24 we linked the OCISS and Medicaid files on a year-by-year basis, using a multistep deterministic linkage algorithm, as follows:
Step 1: Social Security number (SSN), first name, last name.
Step 2: SSN, first name, date of birth (month).
Step 3: SSN, last name, date of birth (month).
Step 4: last name, first name, date of birth (month), date of birth (year).
In any given year, 83% to 89% of women in both the OCISS and Medicaid files were identified through step 1, and ≥ 5% were identified through step 4. The remaining individuals were identified through steps 2 and 3.
Study Population
Our study population consisted of Ohio Medicaid beneficiaries age < 65 years diagnosed with invasive incident female breast cancer during the years 1996 to 2005 who had been enrolled in the Medicaid program ≥ 3 months before cancer diagnosis (n = 2,795). Because of incomplete claims data, we excluded all patients who were enrolled in managed care within 6 months before and 6 months after cancer diagnosis (n = 356). In addition, because of the common co-occurrence of intellectual developmental disabilities (IDDs) with psychiatric conditions, we also excluded individuals identified with IDDs (n = 262), leaving our study population at 2,177. Individuals with IDDs were identified by using the relevant ICD-9-CM diagnosis codes (ie, 299, 315.8, 315.9, 317, 318, 319, 343, 742.4, and 758), category of service codes, and provider specialty codes.25
Outcome Variables
Cancer stage at diagnosis.
We identified cases in which patients had unstaged or unknown-stage cancer or in which there was not sufficient information to assign a cancer stage. We also examined distant-stage cancer, identifying cases in which distant metastasis was diagnosed at presentation.
Definitive cancer treatment.
Limiting our analysis to patients diagnosed with locoregional disease and using procedural codes listed elsewhere,26 we categorized definitive cancer treatment as follows: for local-stage cancer, mastectomy or lumpectomy plus radiation therapy; for regional-stage cancer, mastectomy plus chemotherapy or lumpectomy, plus radiation therapy, plus chemotherapy.
Independent Variables
Mental illness.
We relied on a number of indicators to identify the presence of mental illness and or alcohol or substance abuse, consistent with a previously published study.27 The following codes were selected with input from the Ohio Department of Medicaid and from two of our coauthors (C.V.T., S.L.) with extensive content expertise.
First, we identified patients with ICD-9-CM diagnosis codes for organic psychotic conditions and other psychoses (290 to 299); neurotic disorders, personality disorders, and other nonpsychotic mental disorders (300 to 314, 315.0 to 315.5, and 316); and alcohol or drug abuse (265.2, 357.5, 535.3, 571.0-571.3, 980, V11.3, and V65.42). Next, we determined receipt of mental health services, based category of service code for mental inpatient hospital (03), mental health services (41), psychology services (48), and mental health support (59); provider specialty codes for child psychiatry (21), psychiatry (23), and psychiatry neurology (75); and provider-type codes for mental hospital (02), services received through the Ohio Department of Alcohol and Drug Addiction Services (11), psychologist (individual or group; 42 and 67), clinic (mental, drug, or alcohol; 51), services received through the Ohio Department of Mental Health (84), and mental health support services (91). We then identified records for filled prescriptions pertaining to psychiatric conditions (eg, anxiolytic, antipsychotic, mood stabilizer, antidepressant, attention deficit disorder, and narcolepsy drugs). Because the administration of antidepressants may involve indications other than mental illness, we conducted a sensitivity analysis by identifying mental illness with and without antidepressants. The results did not change in any substantive way.
Once these indicators were created, mental illness was identified by applying the following algorithm: diagnosis for mental illness plus receipt of mental health services or prescription drugs for psychiatric conditions.
To further minimize the possibility of identifying false-positive cases, we included an additional criterion: if a patient had received services from a psychologist, we required that there had been at least one visit to a psychiatrist or at least one filled prescription for a psychiatric condition.
Demographics and county of residence.
We grouped patients into the following age categories: 18 to 39, 40 to 49, 50 to 59, and 60 to 64 years. Because of small racial and ethnic minority representation in the state of Ohio, we identified women as either African American or other. Marital status was categorized as married, nonmarried, or unknown. County of residence at the time of diagnosis was identified as Appalachian, metro, rural, or suburban.28
Comorbidities.
We identified nonpsychiatric comorbid conditions according to the method by Elixhauser et al29 and created a dichotomous variable (yes v no) to indicate the presence of any comorbid condition.
Nursing home.
Because of the strong association between nursing home residence and unstaged or unknown-stage cancer documented in previous studies,30 albeit in the elderly population, we also accounted for nursing home stay in the 6 months before or after cancer diagnosis.
Eligibility categories.
We flagged eligibility for the disabled, those eligible for dual Medicare-Medicaid coverage and spend-down program participation, and nursing home stay in the 6 months before or 6 months after cancer diagnosis. These two timeframes were used in the analyses of cancer stage and cancer treatment, respectively.
Statistical Analysis
In addition to a descriptive analysis, we used multivariable regression models to analyze the association between mental illness and the outcomes of interest, after adjusting for patient covariates. For the analysis of stage, we first developed a regression model for unstaged or unknown-stage cancer versus all others. Next, we excluded cases of unstaged or unknown-stage cancer and developed a model for distant-stage cancer versus locoregional disease. Our analysis of receipt of definitive treatment was limited to patients with locoregional cancer. We used SAS software (version 9.2; SAS Institute, Cary, NC).
Results
Table 1 summarizes the distribution of the study population by demographic and other attributes. Of the 2,177 Medicaid beneficiaries with incident invasive female breast cancer, we identified 1,310 women with mental illness (60.2%). Their mean age was 52.2 years, compared with 50.8 years for women without mental illness. Women with mental illness were more likely to be non–African American, unmarried, and enrolled in the disabled program within Medicaid. They were also more likely to be diagnosed with medical comorbidities (83.8% v 64.6%; P < .001).
Table 1.
Demographic and Clinical Characteristics of Study Population
| Characteristic | Mental Illness |
Total (N = 2,177) |
||||
|---|---|---|---|---|---|---|
| Yes (n = 1,310) |
No (n = 867) |
|||||
| No. | % | No. | % | No. | % | |
| Age, years* | ||||||
| 18-39 | 85 | 6.5 | 149 | 17.2 | 234 | 10.8 |
| 40-49 | 398 | 30.4 | 175 | 20.2 | 573 | 26.3 |
| 50-59 | 549 | 41.9 | 307 | 35.4 | 856 | 39.3 |
| 60-64 | 278 | 21.2 | 236 | 27.2 | 514 | 23.6 |
| Race* | ||||||
| African American | 298 | 22.8 | 258 | 29.8 | 556 | 25.5 |
| Other | 1,012 | 77.3 | 609 | 70.2 | 1,621 | 74.5 |
| Marital status* | ||||||
| Married | 223 | 17.0 | 223 | 25.7 | 446 | 20.5 |
| Not married | 866 | 66.1 | 514 | 59.3 | 1,380 | 63.4 |
| Unknown | 221 | 16.9 | 130 | 15.0 | 351 | 16.1 |
| Region | ||||||
| Appalachian | 265 | 20.2 | 171 | 19.7 | 436 | 20.0 |
| Metro | 785 | 59.9 | 496 | 57.2 | 1,281 | 58.8 |
| Rural | 111 | 8.5 | 91 | 10.5 | 202 | 9.3 |
| Suburban | 149 | 11.4 | 109 | 12.6 | 258 | 11.9 |
| Comorbidities* | ||||||
| Yes | 1,098 | 83.8 | 560 | 64.6 | 1,658 | 76.2 |
| No | 212 | 16.2 | 307 | 35.4 | 519 | 23.8 |
| Nursing home† | ||||||
| Yes | 49‡ | 3.7 | 19‡ | 2.2 | 68‡ | 3.1 |
| No | 1,261‡ | 96.3 | 848‡ | 97.8 | 2,109‡ | 96.9 |
| Medicare | ||||||
| Yes | 336‡ | 25.7 | 195‡ | 22.5 | 531‡ | 24.4 |
| No | 974‡ | 74.4 | 672‡ | 77.5 | 1,646‡ | 75.6 |
| Disability* | ||||||
| Yes | 1,001‡ | 76.4 | 480‡ | 55.4 | 1,481‡ | 68.0 |
| No | 309‡ | 23.6 | 387‡ | 44.6 | 696‡ | 32.0 |
| Spend-down program | ||||||
| Yes | 210‡ | 16.0 | 128‡ | 14.8 | 338‡ | 15.5 |
| No | 1,100‡ | 84.0 | 739‡ | 85.2 | 1,839‡ | 84.5 |
P < .001.
P < .05. All other comparisons not significant at P < .05.
Entered 6 months before cancer diagnosis.
Table 2 lists the results of our bivariate analysis of stage and receipt of definitive cancer treatment. The proportion of women with unstaged or unknown-stage cancer was significantly lower among those with than those without mental illness (6.5% and 9.9%, respectively; P = .004). Conversely, unstaged or unknown-stage cancer was significantly higher among women receiving nursing home care in the 6 months preceding cancer diagnosis, compared with their community-dwelling counterparts (23.5% v 7.4%).
Table 2.
Bivariable Association Between Patient Covariables, Stage at Diagnosis, and Receipt of Definitive Treatment
| Variable | Unstaged or Unknown-Stage Cancer |
Distant-Stage Cancer |
Definitive Treatment for Locoregional Cancer |
|||
|---|---|---|---|---|---|---|
| No. | % | No. | % | No. | % | |
| Mental illness | ||||||
| Yes | 85* | 6.5 | 66† | 5.4 | 683 | 58.9 |
| No | 86 | 9.9 | 80 | 10.2 | 384 | 54.8 |
| Age, years | ||||||
| 18-39 | 16 | 6.8 | 27‡ | 12.4 | 100 | 52.4 |
| 40-49 | 43 | 7.5 | 33 | 6.2 | 272 | 54.7 |
| 50-59 | 58 | 6.8 | 52 | 6.5 | 441 | 59.1 |
| 60-64 | 54 | 10.5 | 34 | 7.4 | 254 | 59.6 |
| Race | ||||||
| African American | 49 | 8.8 | 44 | 8.7 | 245‡ | 52.9 |
| Other | 122 | 7.5 | 102 | 6.8 | 822 | 58.8 |
| Marital status | ||||||
| Married | 32† | 7.2 | 28 | 6.8 | 192* | 49.7 |
| Not married | 93 | 6.7 | 97 | 7.5 | 704 | 59.2 |
| Unknown | 46 | 13.1 | 21 | 6.9 | 171 | 60.2 |
| Region | ||||||
| Appalachian | 39 | 8.9 | 22 | 5.5 | 233‡ | 62.1 |
| Metro | 105 | 8.2 | 85 | 7.2 | 595 | 54.5 |
| Rural | 11 | 5.5 | 19 | 10.0 | 109 | 63.4 |
| Suburban | 16 | 6.2 | 20 | 8.3 | 130 | 58.6 |
| Comorbidities | ||||||
| Yes | 127 | 7.7 | 86† | 5.6 | 879† | 60.8 |
| No | 44 | 8.5 | 60 | 12.6 | 188 | 45.3 |
| Nursing home | ||||||
| Yes | 16†§ | 23.5 | 10†§ | 19.2 | 38‖ | 59.4 |
| No | 155§ | 7.4 | 136§ | 7.0 | 1,029‖ | 57.3 |
| Medicare | ||||||
| Yes | 38§ | 7.2 | 23‡ | 4.7 | 305‖ | 57.4 |
| No | 133§ | 8.1 | 123§ | 8.1 | 762‖ | 57.3 |
| Disability | ||||||
| Yes | 122§ | 8.2 | 94§ | 6.9 | 837†‖ | 61.6 |
| No | 49§ | 7.0 | 52§ | 8.0 | 230‖ | 45.9 |
| Spend down | ||||||
| Yes | 34§ | 10.1 | 18§ | 5.9 | 198‖ | 61.5 |
| No | 137§ | 7.5 | 128§ | 7.5 | 869‖ | 56.5 |
P < .01.
P < .001.
P < .05. All other comparisons not significant at P < .05.
Entered 6 months before cancer diagnosis.
Entered 6 months before or after cancer diagnosis.
With regard to distant-stage cancer at diagnosis, a significantly lower proportion of women with mental illness than those without were diagnosed with metastatic cancer (5.4% v 10.2%). The proportion of distant-stage cancers at diagnosis was also significantly lower among women with comorbidities compared with those without (5.6% v 12.6%), among those with dual Medicare-Medicaid eligibility compared with those not eligible for dual participation (4.7% v 8.1%), and among those with known marital status (married, 7.2%; unmarried, 6.7%; and unknown, 13.1%); however, it was higher among those receiving nursing home care in the 6 months before cancer diagnosis compared with their community-dwelling counterparts (19.2% v 7.0%).
Receipt of definitive treatment by women diagnosed with locoregional cancer did not differ by presence of mental illness (58.9% v 54.8% in women with and without mental illness, respectively; P = .08). We observed a lower proportion of definitive treatment among married women (49.7% v 59.2% in unmarried women or 60.2% in women with unknown marital status). Conversely, women with medical comorbidities and those enrolled in the disabled program were more likely to receive definitive treatment (60.8% v 45.3% and 61.6% v 45.9%, respectively; both P < .001).
Table 3 lists the results of the multivariable analysis. Results from the regression analysis indicated that after adjusting for potential confounders, women with mental illness were less likely than those without to have unstaged or unknown-stage cancer (adjusted odds ratio [OR], 0.61; 95% CI, 0.44 to 0.86; P < .001) and less likely to present with distant-stage cancer (adjusted OR, 0.59; 95% CI, 0.40 to 0.85; P < .001). The likelihood to receive definitive treatment for breast cancer did not differ by the presence of mental illness (adjusted OR, 1.04; 95% CI, 0.84 to 1.29).
Table 3.
Results From Multivariable Logistic Regression Analysis
| Factor | Unstaged or Unknown-Stage Cancer |
Distant-Stage Cancer |
Definitive Treatment for Locoregional Cancer* |
|||
|---|---|---|---|---|---|---|
| Adjusted OR | 95% CI | Adjusted OR | 95% CI | Adjusted OR | 95% CI | |
| Mental illness | ||||||
| Yes | 0.61† | 0.44 to 0.86 | 0.59† | 0.40 to 0.85 | 1.04 | 0.84 to 1.29 |
| No‡ | 1.0 | 1.0 | 1.0 | |||
| Age, years | ||||||
| 18-39‡ | 1.0 | 1.0 | 1.0 | |||
| 40-49 | 1.22 | 0.64 to 2.34 | 0.76 | 0.42 to 1.37 | 0.63§ | 0.43 to 0.93 |
| 50-59 | 0.98 | 0.50 to 1.91 | 0.84 | 0.46 to 1.54 | 0.67 | 0.45 to 0.99 |
| 60-64 | 1.57 | 0.80 to 3.09 | 1.01 | 0.53 to 1.92 | 0.68 | 0.44 to 1.03 |
| Race | ||||||
| African American | 1.13 | 0.76 to 1.68 | 1.29 | 0.84 to 1.98 | 0.92 | 0.71 to 1.17 |
| Other‡ | 1.0 | 1.0 | 1.0 | |||
| Marital status | ||||||
| Married‡ | 1.0 | 1.0 | 1.0 | |||
| Not married | 1.00 | 0.65 to 1.57 | 1.34 | 0.84 to 2.13 | 1.39§ | 1.08 to 1.79 |
| Unknown | 2.06† | 1.26 to 3.38 | 1.15 | 0.63 to 2.12 | 1.20 | 0.86 to 1.67 |
| Region | ||||||
| Appalachian | 1.42 | 0.77 to 2.63 | 0.76 | 0.40 to 1.45 | 1.13 | 0.78 to 1.62 |
| Metro | 1.32 | 0.74 to 2.32 | 0.84 | 0.49 to 1.45 | 0.80 | 0.58 to 1.11 |
| Rural | 0.80 | 0.36 to 1.79 | 1.27 | 0.65 to 2.50 | 1.26 | 0.82 to 1.95 |
| Suburban‡ | 1.0 | 1.0 | 1.0 | |||
| Comorbidities | ||||||
| Yes | 0.86 | 0.55 to 1.35 | 0.44‖ | 0.28 to 0.69 | 1.60† | 1.21 to 2.12 |
| No‡ | 1.0 | 1.0 | 1.0 | |||
| Nursing home | ||||||
| Yes | 4.17‖ | 2.25 to 7.71 | 3.91‖ | 1.84 to 8.30 | 0.95 | 0.55 to 1.63 |
| No‡ | 1.0 | 1.0 | 1.0 | |||
| Medicare | ||||||
| Yes | 0.75 | 0.48 to 1.19 | 0.60 | 0.35 to 1.03 | 0.74§ | 0.58 to 0.96 |
| No† | 1.0 | 1.0 | 1.0 | |||
| Disability | ||||||
| Yes | 1.38 | 0.91 to 2.08 | 1.61§ | 1.02 to 2.54 | 1.79‖ | 1.38 to 2.31 |
| No‡ | 1.0 | 1.0 | 1.0 | |||
| Spend-down program | ||||||
| Yes | 1.35 | 0.83 to 2.20 | 0.97 | 0.53 to 1.79 | 1.29 | 0.95 to 1.76 |
| No† | 1.0 | 1.0 | 1.0 | |||
| Length of enrollment in Medicaid before cancer diagnosis, years | 0.92§ | 0.86 to 0.98 | 0.93§ | 0.86 to 0.99 | 1.00 | 0.96 to 1.04 |
| Cancer stage | ||||||
| Local | — | — | 3.06‖ | 2.50 to 3.73 | ||
| Regional‡ | 1.0 | |||||
Abbreviation: OR, odds ratio.
Analysis of receipt of definitive treatment was limited to patients with local- or regional-stage cancer.
P < .01.
Referent.
P < .05.
P < .001. All other statistics not significant at P < .05.
Discussion
In this study, we examined potential differences in breast cancer outcomes between women with and without mental illness. The findings indicated that among Ohio Medicaid beneficiaries, women with mental illness were less likely than those without to have unstaged or unknown-stage cancer and less likely to be diagnosed with metastatic disease, but they were equally likely to receive definitive treatment for locoregional cancer. As such, findings from this study suggest that the higher breast cancer mortality previously reported among women with mental illness18 may not be the result of later stage at diagnosis, especially if the excess cancer mortality in individuals with mental illness, as documented previously in the population at large, also holds true in the Medicaid population. Additional studies need to focus on other causes of higher mortality, such as differences in factors that influence long-term survivorship. These findings also suggest that patients with mental illness possibly undergo physical assessment at a higher frequency than previously thought. Indeed, findings from a qualitative study of 129 patients who were seen for a mental health intake visit indicated that physical health was discussed in 87% of such visits.31
Two important factors need to be borne in mind when interpreting these unexpected findings. First, the findings reflect the equalizing effect32 of the Medicaid program and its ability to reduce financial barriers, at least partially, thus improving vulnerable individuals' access to care. In addition, although the comparison group consisted of women without mental illness, it included Medicaid-insured women who also face important challenges, such as mobility impairments, which have been shown to be associated with numerous barriers to care.33,34
Second, by limiting our study population to Medicaid beneficiaries who had been enrolled in Medicaid at least 3 months before cancer diagnosis, this study excluded a vulnerable subgroup of individuals who are new Medicaid beneficiaries, including women who may lack the self-help skills and/or the informal and formal support networks needed to identify programs for which they may be eligible and in which they can enroll and women whose incomes slightly exceed the Medicaid eligibility threshold, rendering them uninsured or underinsured. In turn, this limited our study population to women who have learned how to navigate the Medicaid program and excluded those with incomes exceeding the Medicaid threshold who are likely to use Medicaid as a safety net program and enroll after being diagnosed with cancer. As shown previously, these women are significantly more likely to present with advanced-stage cancer than those who had been enrolled in Medicaid before being diagnosed with cancer.22,35–37
Although limiting our study population to those enrolled in Medicaid at least 3 months before cancer diagnosis poses a challenge for the generalizability of the findings to those with mental illness at large, we note that this criterion was required to give us a look-back period of adequate length to identify women with mental illness and/or physical comorbid conditions. As such, the use of administrative claims data to identify and study outcomes in individuals with mental illness and/or medical comorbidities poses an important methodologic challenge that may be overcome by linking these data to an external source of data that flags mental illness independently of claims data.
Our definition of mental illness deserves some additional discussion. By including relevant diagnosis codes, codes for mental health services, and codes for prescription drugs, our goal was to be more rather than less encompassing. Our reliance on such a multipronged approach is supported by findings from previously published studies,38,39 as well as from our own analysis, indicating that not all individuals with a psychiatric condition through diagnosis codes are seen by a mental health professional, have visits with mental health providers, or are prescribed medications for their psychiatric condition. A 2010 study by the Substance Abuse and Mental Health Services Administration indicated that the highest use of psychotropic medications was observed among patients with schizophrenia,39 implying that algorithms relying on relevant prescription drugs alone would yield a study population over-represented by those with the most severe mental illnesses and under-represented by those with milder illnesses. Nonetheless, we required mental illness diagnosis codes to be documented simultaneously with receipt of mental health services or prescription drugs for psychiatric conditions. We believe that this strategy, designed to reduce the number of false-positive cases (ie, individuals with mild or no mental illness), provided a balance to a more restrictive approach based solely on concomitant psychiatric diagnosis and psychoactive medication use.
We note that given our algorithm to identify mental illness, our study population was limited to those who actually received treatment for their mental condition. Thus, our findings cannot be generalized to Ohio Medicaid beneficiaries who might have mild, undiagnosed, and/or untreated mental illness or to women with mental illness who are not enrolled in Medicaid, whether in Ohio or elsewhere. Although this remains to be shown in empirical studies, we hypothesize that individuals treated for their mental illness may be more likely than those who are untreated to receive integrated medical and psychiatric care. If true, our results likely project a more favorable picture of cancer outcomes in this vulnerable population than what occurs in the population of persons with mental illness at large, many of whom are untreated. Finally, we note that our sensitivity analysis identifying women in the categories of severe (schizophrenia and bipolar disorders), mild (all other mental disorders), or no mental illness indicated the lack of disparity in outcomes among women with severe mental illness as compared with those with mild or no mental illness.
The principal strength of our study lies in our use of linked data from the OCISS and Medicaid. We also note that in addition to applying stringent criteria to define mental illness, we excluded persons with intellectual and developmental disabilities from our study population, given the high prevalence of psychiatric conditions in this patient population and the likelihood that they will experience uniquely different cancer-related outcomes.
In conclusion, in our population of women who had enrolled in Ohio Medicaid at least 3 months before breast cancer diagnosis, we observed no disparities associated with mental illness. Additional verification of mental illness, by type and severity of psychiatric diagnosis, through state mental health agencies or other external sources of data, would more fully characterize the study population, allowing more precise interpretation of study findings.
Acknowledgment
Supported by National Cancer Institute (NCI) Grant No. R03 CA 134195. The authors were also supported in part by Case Western Reserve University/Cleveland Clinic Clinical and Translational Science Award No. UL1 RR024989 from the National Center for Research Resources (NCRR), a component of the National Institutes of Health (NIH) and NIH roadmap for Medical Research (S.M.K.) and by NCI Grant No. R25CA11898 (C.V.T.). The content is solely the responsibility of the authors and does not represent the official views of the NCI or NCRR. Presented in part at the Annual Meeting of AcademyHealth, Seattle, WA, June 12-14, 2011. We thank Georgette Haydu of the Ohio Department of Health, which maintains the Ohio Cancer Incidence Surveillance System, and James Gearheart of the Ohio Department of Medicaid, which administers the Ohio Medicaid program, for their careful review of an earlier version of the manuscript. Ms. Haydu and Mr. Gearheart were with the respective Departments during the time the study was conducted. We also thank Meatal Patel, MPH, for her assistance in preparing this article.
Authors' Disclosures of Potential Conflicts of Interest
Disclosures provided by the authors are available with this article at jop.ascopubs.org.
Author Contributions
Conception and design: Siran M. Koroukian
Financial support: Siran M. Koroukian
Administrative support: Siran M. Koroukian
Provision of study materials or patients: Siran M. Koroukian
Collection and assembly of data: Siran M. Koroukian
Data analysis and interpretation: All authors
Manuscript writing: All authors
Final approval of manuscript: All authors
AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST
The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or jop.ascopubs.org/site/misc/ifc.xhtml.
Siran M. Koroukian
Employment: American Renal Associates (I)
Leadership: American Renal Associates (I)
Stock or Other Ownership: American Renal Associates (I)
Paul M. Bakaki
No relationship to disclose
Negar Golchin
No relationship to disclose
Carl V. Tyler
No relationship to disclose
Cynthia Owusu
No relationship to disclose
Sana Loue
No relationship to disclose
References
- 1.Green AI, Canuso CM, Brenner MJ, et al. Detection and management of comorbidity in patients with schizophrenia. Psychiatr Clin North Am Mar. 2003;26:115–139. doi: 10.1016/s0193-953x(02)00014-x. [DOI] [PubMed] [Google Scholar]
- 2.Jeste DV, Gladsjo JA, Lindamer LA, et al. Medical comorbidity in schizophrenia. Schizophr Bull. 1996;22:413–430. doi: 10.1093/schbul/22.3.413. [DOI] [PubMed] [Google Scholar]
- 3.Bushe CJ, Hodgson R. Schizophrenia and cancer: In 2010 do we understand the connection? Can J Psychiatry. 2010;55:761–767. doi: 10.1177/070674371005501203. [DOI] [PubMed] [Google Scholar]
- 4.Bushe CJ, Taylor M, Haukka J. Mortality in schizophrenia: A measurable clinical endpoint. J Psychopharmacol. 2010;24(suppl 4):17–25. doi: 10.1177/1359786810382468. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Tran E, Rouillon F, Loze JY, et al. Cancer mortality in patients with schizophrenia: An 11-year prospective cohort study. Cancer. 2009;115:3555–3562. doi: 10.1002/cncr.24383. [DOI] [PubMed] [Google Scholar]
- 6.Lawrence D, Kisely S, Pais J. The epidemiology of excess mortality in people with mental illness. Can J Psychiatry. 2010;55:752–760. doi: 10.1177/070674371005501202. [DOI] [PubMed] [Google Scholar]
- 7.Kisely S, Smith M, Lawrence D, et al. Mortality in individuals who have had psychiatric treatment: Population-based study in Nova Scotia. Br J Psychiatry. 2005;187:552–558. doi: 10.1192/bjp.187.6.552. [DOI] [PubMed] [Google Scholar]
- 8.Saha S, Chant D, McGrath J. A systematic review of mortality in schizophrenia: Is the differential mortality gap worsening over time? Arch Gen Psychiatry. 2007;64:1123–1131. doi: 10.1001/archpsyc.64.10.1123. [DOI] [PubMed] [Google Scholar]
- 9.Chang CK, Hayes RD, Broadbent M, et al. All-cause mortality among people with serious mental illness (SMI), substance use disorders, and depressive disorders in southeast London: A cohort study. BMC Psychiatry. 2010;10:77. doi: 10.1186/1471-244X-10-77. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Auquier P, Lançon C, Rouillon F, et al. Mortality in schizophrenia. Pharmacoepidemiol Drug Saf. 2006;15:873–879. doi: 10.1002/pds.1325. [DOI] [PubMed] [Google Scholar]
- 11.Hennekens CH, Hennekens AR, Hollar D, et al. Schizophrenia and increased risks of cardiovascular disease. Am Heart J Dec. 2005;150:1115–1121. doi: 10.1016/j.ahj.2005.02.007. [DOI] [PubMed] [Google Scholar]
- 12.De Hert M, Dekker JM, Wood D, et al. Cardiovascular disease and diabetes in people with severe mental illness position statement from the European Psychiatric Association (EPA), supported by the European Association for the Study of Diabetes (EASD) and the European Society of Cardiology (ESC) Eur Psychiatry. 2009;24:412–424. doi: 10.1016/j.eurpsy.2009.01.005. [DOI] [PubMed] [Google Scholar]
- 13.Van Gaal LF. Long-term health considerations in schizophrenia: Metabolic effects and the role of abdominal adiposity. Eur Neuropsychopharmacol. 2006;16(suppl 3):S142–S148. doi: 10.1016/j.euroneuro.2006.06.005. [DOI] [PubMed] [Google Scholar]
- 14.Lawrence D, Kisely S. Inequalities in healthcare provision for people with severe mental illness. J Psychopharmacol. 2010;24(suppl 4):61–68. doi: 10.1177/1359786810382058. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Druss BG, Bradford WD, Rosenheck RA, et al. Quality of medical care and excess mortality in older patients with mental disorders. Arch Gen Psychiatry. 2001;58:565–572. doi: 10.1001/archpsyc.58.6.565. [DOI] [PubMed] [Google Scholar]
- 16.Goldman LS. Medical illness in patients with schizophrenia. J Clin Psychiatry. 1999;60(suppl 21):10–15. [PubMed] [Google Scholar]
- 17.Inagaki T, Yasukawa R, Okazaki S, et al. Factors disturbing treatment for cancer in patients with schizophrenia. Psychiatry Clin Neurosci. 2006;60:327–331. doi: 10.1111/j.1440-1819.2006.01509.x. [DOI] [PubMed] [Google Scholar]
- 18.Kisely S, Sadek J, MacKenzie A, et al. Excess cancer mortality in psychiatric patients. Can J Psychiatry. 2008;53:753–761. doi: 10.1177/070674370805301107. [DOI] [PubMed] [Google Scholar]
- 19.Iezzoni LI, Ngo LH, Li D, et al. Early stage breast cancer treatments for younger Medicare beneficiaries with different disabilities. Health Serv Res. 2008;43:1752–1767. doi: 10.1111/j.1475-6773.2008.00853.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Chochinov HM, Martens PJ, Prior HJ, et al. Does a diagnosis of schizophrenia reduce rates of mammography screening? A Manitoba population-based study. Schizophr Res. 2009;113:95–100. doi: 10.1016/j.schres.2009.04.022. [DOI] [PubMed] [Google Scholar]
- 21.Carney CP, Jones LE. The influence of type and severity of mental illness on receipt of screening mammography. J Gen Intern Med. 2006;21:1097–1104. doi: 10.1111/j.1525-1497.2006.00565.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Koroukian SM. Assessing the effectiveness of Medicaid in breast and cervical cancer prevention. J Public Health Manag Pract. 2003;9:306–314. doi: 10.1097/00124784-200307000-00009. [DOI] [PubMed] [Google Scholar]
- 23.Koroukian SM. Assessment and interpretation of comorbidity burden in older adults with cancer. J Am Geriatr Soc. 2009;57(suppl 2):s275–s278. doi: 10.1111/j.1532-5415.2009.02511.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.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;38:947–960. doi: 10.1111/1475-6773.00155. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Tyler CV, Schramm SC, Karafa M, et al. Chronic disease risks in young adults with autism spectrum disorder: Forewarned is forearmed. Am J Intellect Dev Disabil. 2011;116:371–380. doi: 10.1352/1944-7558-116.5.371. [DOI] [PubMed] [Google Scholar]
- 26.Koroukian SM, Bakaki PM, Schluchter MD, et al. Treatment and survival patterns in relation to multimorbidity in patients with locoregional breast and colorectal cancer. J Geriatr Oncol. 2011;2:200–208. doi: 10.1016/j.jgo.2011.02.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Koroukian SM, Bakaki PM, Golchin N, et al. Mental illness and use of screening mammography among Medicaid beneficiaries. Am J Prev Med. 2012;42:606–609. doi: 10.1016/j.amepre.2012.03.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Koroukian SM. Screening mammography was used more, and more frequently, by longer than shorter term Medicaid enrollees. J Clin Epidemiol. 2004;57:824–831. doi: 10.1016/j.jclinepi.2004.01.004. [DOI] [PubMed] [Google Scholar]
- 29.Elixhauser A, Steiner C, Harris DR, et al. Comorbidity measures for use with administrative data. Med Care. 1998;36:8–27. doi: 10.1097/00005650-199801000-00004. [DOI] [PubMed] [Google Scholar]
- 30.Koroukian SM, Xu F, Beaird H, et al. Complexity of care needs and unstaged cancer in elders: A population-based study. Cancer Detect Prev. 2007;31:199–206. doi: 10.1016/j.cdp.2007.04.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Carson N, Katz AM, Gao S, et al. Assessment of physical illness by mental health clinicians during intake visits. Psychiatr Serv. 2010;61:32–37. doi: 10.1176/appi.ps.61.1.32. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Currie J, Gruber J. Public health insurance and medical treatment: The equalizing impact of the Medicaid expansion. http://economics.mit.edu/files/101.
- 33.Iezzoni LI, Kilbridge K, Park ER. Physical access barriers to care for diagnosis and treatment of breast cancer among women with mobility impairments. Oncol Nurs Forum. 2010;37:711–717. doi: 10.1188/10.ONF.711-717. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Iezzoni LI, Park ER, Kilbridge KL. Implications of mobility impairment on the diagnosis and treatment of breast cancer. J Womens Health (Larchmt) 2011;20:45–52. doi: 10.1089/jwh.2009.1831. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Bradley CJ, Given CW, Roberts C. Late stage cancers in a Medicaid-insured population. Med Care. 2003;41:722–728. doi: 10.1097/01.MLR.0000065126.73750.D1. [DOI] [PubMed] [Google Scholar]
- 36.O'Malley CD, Shema SJ, Clarke LS, et al. Medicaid status and stage at diagnosis of cervical cancer. Am J Public Health. 2006;96:2179–2185. doi: 10.2105/AJPH.2005.072553. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Perkins CI, Wright WE, Allen M, et al. Breast cancer stage at diagnosis in relation to duration of Medicaid enrollment. Med Care. 2001;39:1224–1233. doi: 10.1097/00005650-200111000-00009. [DOI] [PubMed] [Google Scholar]
- 38.Thomas MR, Waxmonsky JA, Gabow PA, et al. Prevalence of psychiatric disorders and costs of care among adult enrollees in a Medicaid HMO. Psychiatr Serv. 2005;56:1394–1401. doi: 10.1176/appi.ps.56.11.1394. [DOI] [PubMed] [Google Scholar]
- 39.Substance Abuse and Mental Health Services Administration. Rockville, MD: Center for Mental Health Services, Substance Abuse, and Mental Health Services Administration; 2010. Mental health and substance abuse services in Medicaid, 2003: Charts and state tables. HHS Publication No. (SMA) 10-4608. [Google Scholar]