Statistical Analyses

 In order to determine the relationship between comorbid depression or anxiety and

health care expenditures in pulmonary and cardiac patients, separate log-linear multiple

regressions were used for pulmonary patients and cardiac patients, with total health care

expenditures and medical expenditures as separate outcomes. Demographics,

socioeconomic factors, physical and mental health status, insurance status, and number of

comorbid conditions were control variables in each analysis. For significant results,

smearing estimation was used to determine differences between groups in dollars.

 Next, the relationship between mental health treatment and health care expenditures

in pulmonary or cardiac patients with depression or anxiety were determined with

separate log-linear multiple regressions. Demographics, socioeconomic factors, physical

and mental health status, insurance status, and number of comorbid conditions were

control variables in each analysis. Smearing estimation was employed for significant

results to obtain group differences in dollars.

 Finally, the relationship between mental health treatment and health care utilization

in pulmonary or cardiac patients with depression or anxiety were determined with

separate negative binomial regressions. The health care utilization variables were

number of office-based provider visits, number of outpatient hospital visits, number of

inpatient nights at discharge, and number of emergency room visits. Again,

demographics, socioeconomic factors, physical and mental health status, insurance status,

and number of comorbid conditions were control variables in each analysis.

 For each of the above analyses, Stata statistical software was used (StataCorp,

2002). Sample weights were employed to take into account the MEPS sampling

procedures and to produce nationally representative estimates.