Purpose: To describe the prevalence of comorbid physical disease in a unique population of long-term state hospital inpatients and to examine demographic predictors of such disease.
Method: This study took place in a state hospital in the southern United States. All state hospital inpatients diagnosed with psychotic disorders and admitted under civil commitment between January 1, 2000 and July 1, 2005 were included (n=267). Demographic information and DSM Axis III medical diagnoses were derived from administrative databases. Data analyses consisted of descriptive statistics, contingency tables, and logistic regression. Given the non-probability nature of the sample, analyses primarily focused on estimation using confidence intervals rather than hypothesis testing.
Results: The patients were chronic, with extended hospitalizations (mean length of stay=481.7 days). The mean number of DSM Axis III diagnoses was 3.86 (SD=2.46, range 0-11), with 21.3% (n=57) of patients receiving a Charlson Comorbidity Inventory (CCI) score ≥1. The most common diagnoses were obesity (42.9%, n=116), constipation (33.6%, n=91), hypertension (31.73%, n=86), and hyperlipidemia (22.88%, n=62). A minority of patients were diagnosed with significant medical diseases such as seizure disorder, HIV, or Alzheimer's. Tardive dyskinesia was rarely diagnosed (1.1%, n=3), while extrapyramidal syndrome was more common (11.57%, n=31).
A logistic regression model identified several client characteristics associated with a CCI score of ≥1, indicative of serious medical comorbidity. Ethnic minorities were 1.9 times more likely to have a CCI ≥1 (OR=1.93; 95% CI= 1.03, 3.64). When compared to men, women were over 2.5 times more likely to have a CCI ≥1 (OR=2.60; 95% CI= 1.35, 4.98). Advancing age also increased risk by approximately 5% per year (OR=1.05; 95% CI= 1.03, 1.08).
Given the known limitations of the CCI, this analysis was supplemented by an additional analysis using simple count measurement as the dependent variable. OLS regression was used to test the association of these predictors with the number of Axis III diagnoses. The model was statistically significant (F=12.491, p<.001) and explained 14.7% of the variance in the number of diagnoses (R2=16.0%, adjusted R2=14.7%). Age (β=.006; 95% CI= .043, .09), gender (β=-.873; 95% CI= -1.445, -.301), 60-day antipsychotic polypharmacy (β=.810; 95% CI= .224, 1.396), all predicted the number of Axis III diagnoses. Ethnic minority status was not predictive in this model.
Conclusion and Implications: A substantial minority of these state hospital inpatients have serious medical comorbidities, and the majority have ≥1 metabolic symptoms. Female gender and advancing age was predictive of more severe medical comorbidity. There were contradictory findings regarding ethnic minorities; further research is recommended.
