An Update on Perfect Depression Care

To the Editor: Suicide is an important public health issue and the cause of much personal suffering (1). In 2006, an article in this journal described our quality improvement initiative, the Perfect Depression Care program, which received the American Psychiatric Association’s Gold Significant Achievement Award for that year. The initiative was associated with a dramatic reduction in annual suicide rates among members of our large health maintenance organization (HMO) network, who were receiving mental health care from the Division of Behavioral Health Services of the Henry Ford Health System in Detroit, Michigan (2). More recent data suggest that these improvements may have been sustained (3). Although promising, these clinical results were preliminary because they had not been compared with official U.S. mortality data.

To address this issue, we cross-referenced clinical suicide data from the first 11 years (1999–2009) of our ongoing initiative with the most recent finalized State of Michigan mortality records for any member of our HMO network who died by suicide (defined by the State of Michigan with ICD-10 codes X60–X84 and Y.87.0). (Because of a two-year lag, 2009 is the most recent year for which state information is available. Pre-1999 data were recorded by the State of Michigan with a different coding system.) HMO member records were matched to State of Michigan mortality data with a two-step process: first by Social Security Number and then by first and last name, date of birth, address, and sex.

The matched State of Michigan mortality records indicated 27 completed suicide deaths among our patients; however, the match missed four suicides that our internal clinical surveillance system had previously identified. Two of these individuals were not residents of Michigan and thus were not listed in the Michigan records. For the other two, the cause of death was listed as other than suicide (“unintentional self-poisoning”) even though our internal surveillance process clearly identified both deaths as suicides.

On the basis of the combined total of 31 suicides for the 11-year observation period, the rate of suicide among our patients was 97 per 100,000 (N=13) for the two baseline years (the average rate for 1999 and 2000). This rate is similar to that reported for a clinical population (4). For the start-up year (2001), the rate of suicide was 41 per 100,000 (N=3). For the follow-up interval (the average for 2002–2009), the rate was 19 per 100,000 (N=15). Poisson regression analysis showed a statistically significant decrease of 82% in the suicide death rate between the baseline (1999–2000) and intervention (2002–2009) years (rate ratio=.20; 95% confidence interval=.16–.24, p≤.001).

This analysis used official mortality statistics to extend findings reported in the 2006 description of our program. Furthermore, our experience suggests that suicide data obtained from a clinical surveillance system may be useful in driving quality improvement. Although these results also suggest that our Perfect Depression Care program may be associated with a reduction in suicide, this finding remains preliminary given the small number of suicides in our sample and other methodological challenges inherent in suicide research (5) and quality improvement work.