Brief Reports: Elevated Cancer Incidence Among Adults With Serious Mental Illness

The Surgeon General's report Mental Health recognized "the inextricably intertwined relationship between our mental health and our physical health and well-being" ( 1 ). As part of the growing interest in the health status of persons with mental illness, the study presented here examined the incidence of cancer in a statewide population of adults who were receiving services for serious mental illness.

A substantial body of literature has examined cancer-related mortality among adults with serious mental illness without reaching consensus about relative risk or causation ( 2 , 3 , 4 , 5 ). Little attention, however, has been given to the relative incidence of cancer among adults with serious mental illness. In the one study that focused specifically on this issue, Lawrence and colleagues ( 2 ) found elevated cancer mortality but little elevated incidence of cancer diagnoses in a western Australian population of 172,932 adults with a broad range of psychiatric diagnoses. Our study extends that line of inquiry by examining the incidence of cancer diagnoses in a population of adults with serious mental illness in Vermont and comparing this rate with the same rate for the general population of adults in Vermont.

Methods

Two databases were used in this analysis. Extracts from the Vermont Cancer Registry provided information about all adults with a cancer diagnosis. Extracts from the database of Vermont's Division of Mental Health provided information about all individuals served by community programs for adults with serious mental illness. These extracts included only the birth date and gender of the individuals represented in each data set.

Because these databases do not include unique person identifiers, probabilistic population estimation was used to determine the proportion of individuals served by mental health programs who received a cancer diagnosis during the same year. Probabilistic population estimation is a statistical data-mining tool that estimates the number of people represented in databases who do not have unique person identifiers and the number of people shared between these databases. These estimates are based on a comparison of the distribution of dates of birth observed in the data sets with the expected distribution of dates of birth. The validity and reliability of this procedure have been demonstrated elsewhere ( 6 ). The analysis was completed for each of eight years (1994 to 2001), and the results were averaged.

This analysis yielded the proportion of adults who were in treatment for serious mental illness during each of the study years and the proportion of adults who were receiving such treatment and had a cancer diagnosis recorded in the state cancer registry during each of the study years. Relative risk (RR) was used to compare the proportion of recipients of mental health services who received a cancer diagnosis with the proportion of the general population that received a cancer diagnosis.

The institutional review board of the Vermont Agency of Human Services found that this research qualified for exemption from institutional review board review and therefore from informed consent.

Results

During the period covered by this study, community mental health programs in Vermont served about 3,300 individuals each year. Men and women were represented about equally (46 percent men and 54 percent women). About two-thirds (63 percent) were in the 18- to 49-year age group; 37 percent were aged 50 years or older. Up to four diagnoses were reported per person; 1,493 persons (45 percent) had a diagnosis of schizophrenia or other psychotic disorder, 1,260 (38 percent) had a diagnosis of an affective disorder, 530 (16 percent) had a diagnosis of a personality disorder, and 431 (13 percent) had a diagnosis of a substance use disorder. 2,255 persons (68 percent) were covered by Medicaid, and nearly half (1,592 persons, or 48 percent) were covered by Medicare. A total of 298 persons (9 percent) were uninsured. Finally, 862 (26 percent) had been on the caseload of their current agency for 11 years or more, and 895 (27 percent) had been on the case-load for less than three years.

As shown in Table 1 , the annual incidence of cancer for adults who received community mental health services for serious mental illness in Vermont during the study period was 1.6 percent, more than twice that of the general population in Vermont (RR=2.5). Compared with individuals in the same age and gender groups in the general population, persons who received services for serious mental illness and were younger than 50 years had a lower incidence of cancer (.8 percent compared with 3.0 percent) but their RR was much greater (RR of 5.4 compared with an RR of 2.0). Compared with individuals in the same age and gender groups in the general population, men with serious mental illness who were younger than 50 years had the highest elevated risk (RR=6.6) despite a low incidence rate of cancer (.7 percent). Compared with individuals in the same age and gender groups in the general population, women with serious mental illness who were older than 50 years had the lowest elevated risk (RR=1.8); however, this risk was statistically significant.

Enlarge table

Discussion and conclusions

These findings raise important questions about the nature of the relationship between serious mental illness and cancer. Future research should consider the effect of lifestyle factors, such as smoking and poor nutrition, which are prevalent among adults with serious mental illness ( 7 ). In addition, potential effects of first- and second-generation antipsychotic medications should be investigated. The antitumor properties of first-generation antipsychotic medications are well known ( 8 ); however, the antitumor properties of second-generation antipsychotics are not. The difference between our finding of elevated risk of cancer among patients with serious mental illness and an earlier finding that indicated decreased cancer risk for this group ( 3 ) could be due to changing medication patterns. This difference could also be due to the use of different risk measures. Our study examined risk of cancer diagnosis, whereas the earlier study examined cancer-related mortality. Cancer research indicates that an estimated 20 percent of cancer patients die of causes that are not related to cancer ( 9 ).

Fortunately, the data necessary to explore the relationship between serious mental illness and cancer are widely available in public heath registries and administrative databases ( 10 ). Detailed medication histories of mental health service recipients, for instance, are available in Medicaid, Medicare, and other paid-claims databases. Comparison of cancer diagnosis rates for adults with different medication histories (first- compared with second-generation antipsychotics) would shed light on the potential effects of these medications.

This research focused on cancer diagnosis and treatment of serious mental illness that occurred during the same year. Future research should consider the temporal order of these events to better understand their relationship. Definitively establishing their temporal order, however, will be difficult. Both serious mental illness and cancer can exist for long periods without being medically identified. Participation in mental health services before a cancer diagnosis, for instance, does not necessarily mean that the mental illness existed before the cancer.

More broadly, the wide range of administrative databases relevant to the health and well-being of mental health service recipients provides the opportunity to achieve a much greater understanding of the relationship between mental health and general health. These include provider databases, insurance paid claims databases, and public health registries. Because these databases provide information not only about mental health service recipients but also about broad comparison groups, they provide the opportunity to compare the health status of these two groups. In conjunction with information about the general health and mental health services that these groups received, future research can add to our understanding of the relationship between the receipt of services and general health status of adults with serious mental illness. A broadly focused research initiative in this area has the potential to improve health and increase the well-being of adults with serious mental illness.

Acknowledgments

This study was supported in part by data infrastructure grant 5-HR1-SM-54228-03 from the Center for Mental Health Services. The contents are solely the responsibility of the authors and do not necessarily represent the official views of the Substance Abuse and Mental Health Services Administration's Center for Mental Health Services. The authors acknowledge the contribution of the staff and management of Vermont's community mental heath programs and Vermont's health care providers and administrators for their contribution to the ongoing data collection efforts that make this work possible.