Adults with serious mental illness have higher rates of morbidity and mortality than adults without serious mental illness, including higher risk of cardiovascular disease, hypertension, and diabetes (
1). Although cancer is the second-leading cause of death in the overall population (
2), evidence varies on cancer risk among persons with serious mental illness. Mental illness research frequently focuses on schizophrenia, and little is known about cancer incidence among persons with bipolar disorder and other serious mental illness. Although studies have suggested protective effects of schizophrenia on cancer (
3), more recent research shows differences in risk based on cancer site. For example, studies have shown heightened risk of breast and lung cancer for persons with serious mental illness but uncertainty regarding cancer in other sites, such as the colon (
4,
5). In addition, little is known about whether racial differences in cancer incidence in the population with serious mental illness mirror the U.S. trend of higher incidence among African Americans versus whites (
6). The objectives of our study were to determine the incidence of cancer in a large, community-based cohort of adults with schizophrenia or bipolar disorder and to examine incidence by diagnosis, race, and cancer site.
Methods
We conducted a retrospective cohort study of Maryland Medicaid adult beneficiaries with serious mental illness. The cohort included adults aged 21 to 62 between July 1, 1992, and July 1, 1993. Beneficiaries in the cohort had two years of continuous Medicaid enrollment between July 1, 1992, and June 30, 1994. Eligible beneficiaries had a diagnosis of schizophrenia or were disabled (received Supplemental Security Income) and had a diagnosis of bipolar disorder. Participants were residents of Metropolitan Baltimore or the rural Eastern Shore region of Maryland. The institutional review boards of the Johns Hopkins University School of Medicine and the Maryland Department of Health and Mental Hygiene approved the study and waived the requirement for informed consent.
Medicaid administrative claims data files provided information on age, sex, race, and diagnoses, which were established at cohort initiation. We defined incident cancer cases as two or more inpatient or outpatient visits with a diagnosis of cancer within one fiscal year (
7). Diagnoses were identified with
ICD-9 codes 140–208, and 238.6. The cohort was followed through 2004. Incident cancer cases could occur between 1996 and 2004 (
8).
Incidence rates were standardized according to age by using weights from the U.S. 2000 population to make results comparable with estimates from the Surveillance Epidemiology and End Results (SEER) program. The incidence rate is reported as events per 100,000 person-years. The SEER program collects data from population-based registries. The SEER 2002–2006 data—the period closest to the study period for which SEER calculated incidence—cover 28% of the U.S. population, yielding the most comprehensive national incidence estimates available (
6).
Standardized incidence ratios (SIRs) were calculated to compare cancer incidence in the Maryland Medicaid cohort with the SEER population for breast, lung, colorectal, prostate, and total cancers. Total cancers included all cancers except nonmelanoma skin cancer. We estimated all SIRs by race, sex, and diagnosis. We used Cox proportional hazards models, controlling for age and sex, to estimate the relative hazard of developing cancer among participants with schizophrenia or bipolar disorder for African Americans versus whites.
Results
The cohort included 2,315 beneficiaries with schizophrenia and 1,002 beneficiaries with bipolar disorder. The mean±SD ages were 41.5±10.2 and 42.9±10.7, respectively. A total of 1,196 (52%) beneficiaries with schizophrenia and 369 (37%) beneficiaries with bipolar disorder were male. Among persons with schizophrenia, 1,294 (56%) were African American and 990 (43%) were white, compared with 447 (45%) and 542 (54%) among cohort members with bipolar disorder.
The SIRs comparing total cancer incidence between the study cohort with serious mental illness and the SEER population were 2.6 for persons with schizophrenia or bipolar disorder (
Table 1). Elevated risk of cancer compared with the SEER data was observed in all subgroups with the exception of men with bipolar disorder, for whom the SIR estimate was 1.5, but the CI was not significant. Lung cancer incidence among participants with schizophrenia or bipolar disorder was more than four times higher than the SEER population. Incidence of colorectal cancer was similarly elevated, with SIRs of 3.5 for persons with schizophrenia and 4.0 for persons with bipolar disorder. Female participants had heightened risk of breast cancer, with SIRs of 2.9 and 1.9 among women with schizophrenia or bipolar disorder, respectively. For prostate cancer, men with schizophrenia had a SIR of 1.9 compared with the U.S. population, but this did not reach statistical significance. Men with bipolar disorder did not show increased risk of prostate cancer.
With analyses controlling for age and sex, the hazard ratio comparing risk of all cancer among African Americans with schizophrenia versus whites with schizophrenia was .95 (CI=.7−1.3). The adjusted hazard ratio for cancer for cohort members with bipolar disorder was .88 (CI=.6−1.3).
Discussion
This retrospective cohort study used Maryland Medicaid administrative data to examine cancer incidence by diagnosis, race, and cancer site among adults with schizophrenia or bipolar disorder. Cohort members with each diagnosis experienced higher total cancer incidence in addition to heightened risk of lung, breast, and colorectal cancer. Unlike in the overall U.S. population, the study cohort showed no racial differences in cancer risk.
The more-than-double risk of cancer observed in the study cohort is similar in magnitude to the result of a 2006 study by Pandiani and colleagues (
8) that used probabilistic methods. Our study used a longitudinal design and examined incidence by diagnosis and race and, unlike a previous study that compared cancer incidence among participants with schizophrenia and bipolar disorder and found elevated risk of cancer in the group with schizophrenia only (
4), our results show heightened risk of cancer among persons with either diagnosis.
Increased risk of site-specific cancers suggests that risk factors disproportionately prevalent in the population with serious mental illness increase the risk of developing certain types of cancer. High rates of smoking in the population with serious mental illness likely contribute to lung cancer incidence, and research suggests a possible but inconclusive elevated risk of breast cancer due to low rates of childbearing and increased prolactin levels caused by use of particular psychotropic medications (
3). The risk factors contributing to high risk of colon cancer are less understood but may be related to smoking, a sedentary lifestyle, or a diet high in fat and low in fruits and vegetables (
9).
This study was limited by use of Medicaid administrative data. To address concerns about reliability and validity of diagnoses, we defined cases using methods validated by a 2003 study that demonstrated the ability of Ohio Medicaid claims data to identify incident breast cancer cases (
7). We were unable to determine the stage of cancer at diagnosis, and the Medicaid sample did not include all persons with schizophrenia or bipolar disorder in the state; some persons with serious mental illness may have had private insurance or been covered by Social Security Disability Insurance rather than SSI, and this group is likely of higher socioeconomic status than the study population. However, a 2010 study showed that 87% of persons with schizophrenia received government health insurance such as Medicaid, suggesting that our population was likely representative of much of the population of interest (
10). In addition, the Medicaid sample by definition covered an insured population, whereas the SEER data were population based and included persons without insurance.
Conclusions
Adults with either schizophrenia or bipolar disorder had heightened risk of developing cancer, specifically lung, colorectal, and breast cancers, with no differences between African Americans and whites. Better understanding of how behavioral and pharmacological factors increase cancer risk among persons with serious mental illness, and more information on the extent to which the population receives appropriate cancer screening and treatment, are important in order to improve health in this vulnerable group. Clinicians and mental health system administrators, together with primary care providers, should promote appropriate cancer screening and work to reduce modifiable risk factors, such as smoking, among persons with serious mental illness.
Acknowledgments and disclosures
Funding was provided by grant R01MH074070 from the National Institute of Mental Health. The authors thank the Maryland Department of Health and Mental Hygiene for its collaboration.
The authors report no competing interests.