3112 — Alternative Methods of Identifying Mental Health Disorders and their Impact on Mortality Following Acute Myocardial Infarction.
Abrams TE (University of Iowa), Sarrazin M
(CRIISP, VAMC, Iowa City, IA), Rosenthal GE
(University of Iowa)
Prior work on the associations between mental health (MH) disorders and outcomes after acute myocardial infarction (AMI) has yielded conflicting results. Thus, we sought to measure associations between MH disorders and AMI mortality, and to determine if associations were influenced by the method of identifying MH disorders.
The sample included 24,879 consecutive VHA hospitalizations in 2005-06 with a principle diagnosis of AMI (ICD-9 code 410), identified from the Patient Treatment File. MH disorders, including depression, anxiety, bipolar disorder, schizophrenia, and PTSD, were identified using ICD-9 codes from: 1) the index hospitalization, or 2) outpatient encounters in the prior 12 months. The primary outcome, 30-day mortality, was determined from the VA Vital Status File. Severity of illness was measured using admission values of 8 common laboratory tests from Decision Support System files. Generalized estimating equations (GEE) were used to adjust mortality for socio-demographics, medical comorbidities, laboratory severity, and cardiac injury (assessed by troponin levels).
MH disorders were identified in 2,503 (10%) patients from inpatient diagnoses and 4,523 (18%) patients from prior outpatient diagnoses. Patients with MH disorders were younger (mean ages, 62 vs. 68 years and 65 vs. 68 years for inpatient and outpatient codes, [p < .001 for each]). Unadjusted mortality was higher in patients with outpatient MH codes (12.4% vs. 11.0%; p=.01), but was lower in patients with inpatient codes (6.5% vs. 11.8%; p < .001). In GEE analyses, adjusted odds of death were higher for veterans with a MH outpatient codes (OR=1.46; 95% CI, 1.33-1.60), but were similar for veterans with inpatient codes (OR=0.87; 95% CI, 0.72-1.06).
Risk-adjusted mortality was higher for patients with prior MH outpatient codes, but not with MH inpatient codes. Thus, associations between comorbid MH disorders and AMI mortality are complex and differ depending on the method of identifying MH disorders.
The differences in associations suggest that the two methods capture unique constructs of MH disorders. Furthermore, studies using claims data should consider using multiple approaches for identifying MH disorders. Lastly, the higher mortality of patients with prior MH outpatient codes indicate greater unmeasured severity or variations in health care delivery of care that may merit further examination.