A patient was classified with psychiatric comorbidity
if a psychiatric diagnosis appeared during any of the patient’s visits. The following psychiatric diagnoses were than included–schizophrenia/psychoses, bipolar disorder, depression, anxiety, and dementia (ICD-9 codes available upon request). Data Analyses T-tests of group means were used to investigate differences in number of ED visits across our substance use categories by psychiatric comorbidity. Logistic regression analysis was used to test the predictive ability of the presence of psychiatric comorbidity on frequency of ED visits, controlling for age, race (Caucasian, African-American, Hispanic, other), Inhibitors,research,lifescience,medical and gender. Interaction effects were also tested between psychiatric comorbidity and age, race, and gender.
Inhibitors,research,lifescience,medical Due to the large sample size, we used a conservative p value of .01. Separate logistic regression models were used for each substance use group. Five categories of “frequent ED use” were created: 4 or more visits (4+), 8 or more visits (8+), 12 or more visits (12 +), 16 or more visits (16+), and 20 or more visits Inhibitors,research,lifescience,medical (20+) across the 4.5-year span of the study. The rationale for using multiple categories was twofold: 1) The literature does not agree on what “frequent use” is, and providing a range of categories allows the data to be comparable to a broader range of previous work. 2) The categories allowed for “sensitivity analyses” to investigate Inhibitors,research,lifescience,medical if the predictive ability of the psychiatric comorbidity would be constant across frequency categories or if its strength as a predictor might level or drop-off after a certain number of visits.
To arrive at these specific categories, the data on ED use were examined. The sample’s mean number of visits across the span of the study was 2.9, Inhibitors,research,lifescience,medical with a standard deviation of 4.8. Based on these data, and the judgment of the clinician co-authors of the manuscript, it was decided that the categories would be based on a count of 4. The first category of frequent use (4+ visits) represents a value just beyond the mean as a lower bound. The next Cilengitide category (8+ visits) captures the number of visits corresponding to the first standard deviation. The remaining categories approximate the next standard deviations. This categorization also reflects the judgment of the clinician co-authors that it would be useful to have categories that correspond to 1+ mean visit per year of the study (4+ visits), 2+ mean visits per year of the study (8+ visits), up to 5+ means visits per year of the study (20+ visits). As well, this grouping corresponds closely to the categories used by one of the only other multi-year studies of repeat users of the ED by persons with psychiatric diagnoses. [16] Results Patient demographic information is presented in Table Table1.1. The sample was predominantly male (72.9%).