The US Surgeon General and the Institute of Medicine have called for health care systems to help reduce the number of suicides.1,2 However, the few assessments of suicide in such systems have examined specific patient groups rather than the entire population of health plan members.3,4 Here we report, to our knowledge, the first information on suicide for the entire membership of a large health maintenance organization (HMO) network.
We identified all suicides among the membership of our HMO network between 1999 and 2010, determining the date and cause of death using official state mortality records. We linked these data to the medical record by name, sex, address, date of birth, and social security number. Using the Centers for Disease Control and Prevention’s external cause of injury mortality matrix, we designated as suicides all deaths with International Statistical Classification of Diseases, Tenth Revision codes X60-X84 or Y87.0. The Centers for Disease Control and Prevention first used this scheme to classify deaths in 1999, the first year of our study period.
We characterized suicides among 5 study groups: all individuals enrolled in our HMO network (members); all members who did (patients) or did not (nonpatient members) access health care network services during that membership year; and patients who did (mental health [MH] patients) or did not (non-MH patients) make at least 1 visit to the network’s specialty MH services department that year. We defined suicide rates in accordance with the State of Michigan’s vital statistics (ie, per 100 000 population), effectively keeping risk time constant based on the nature of the HMO membership. We used standard linear regression, with annual suicide rate as the dependent variable, to evaluate rates over time with statistical significance defined as P < .05.
The institutional review boards at Henry Ford Hospital and the State of Michigan approved this project; patient consent was waived.
During the 11-year study, the annual HMO network membership ranged from 182 183 to 293 228 and was demographically representative of southeast Michigan, with approximately 25% older than the age of 65 years, 55% female, and 40% white. On average each year, approximately 65% of members were patients, and approximately 60% of patients were MH patients. There were 160 suicides among all HMO members (Table 1).
The mean annual suicide rate for members was 5.77 per 100 000 and did not significantly change over the study period (P = .20) (Table 2), whereas the annual suicide rate in the general population of the state of Michigan increased significantly (P < .001) (mean, 10.82 per 100 000).
The mean annual suicide rate among patients (6.38 per 100 000) and nonpatient members (4.47 per 100 000) was similar to that for all members (5.77 per 100 000), and likewise did not change (Table 2). Suicide rates declined for MH patients (P < .04) but increased for non-MH patients (P < .01).
This report presents the first description of suicide rates in the entire membership of a large HMO. To our knowledge, the only other available studies examining suicide in insured populations within the United States are limited to 2 specific patient populations: military veterans4 and patients receiving treatment for depression.3 In our sample, the first to include nonpatient members, the annual suicide rate among all members did not change over time. However, subgroup analysis suggested important differences.
First, the number of suicides decreased among HMO members who received MH specialty services, in association with a targeted suicide prevention effort (previously described).5 Second, suicides increased among HMO members who accessed general medical services but not specialty MH services. This finding reflects a pattern of health care use observed in persons who die of suicide.6 Third, during a time when suicides increased in the general state population, suicides remained stable among HMO members including those who never accessed any of the HMO’s health care services.
These findings apply to a single HMO’s membership and are limited by small sample size in some subgroups and the potential underrecording of suicides in official records. Notwithstanding, this report provides previously unavailable baseline data for other health care systems engaged in important efforts to measure and prevent suicide.
Corresponding Author: M. Justin Coffey, MD, The Menninger Department of Psychiatry and Behavioral Sciences, Baylor College of Medicine, 12301 Main St, Houston, TX 77035 (firstname.lastname@example.org).
Published Online: January 21, 2015. doi:10.1001/jamapsychiatry.2014.2440.
Author Contributions: Dr Ahmedani had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: All authors.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: M. J. Coffey, C. E. Coffey.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Ahmedani.
Administrative, technical, or material support: C. E. Coffey, Ahmedani.
Study supervision: C. E. Coffey, Ahmedani.
Conflict of Interest Disclosures: None reported.
Coffey MJ, Coffey CE, Ahmedani BK. Suicide in a Health Maintenance Organization Population. JAMA Psychiatry. 2015;72(3):294-296. doi:10.1001/jamapsychiatry.2014.2440