Table 6

Impact of Bleeding Complication Status on Total Hospital Costs: Comparison of GEE and OLS Models

Surgical Cohort

GEE

OLS

Model Estimate (CI)*

P-value

Model Estimate (CI)*

P-value

Cardiac

1.31 (1.30, 1.32)

< 0.001

1.36 (1.35, 1.36)

< 0.001

Vascular

1.93 (1.93, 1.95)

< 0.001

1.93 (1.92, 1.95)

< 0.001

Non-cardiac thoracic

1.60 (1.58, 1.60)

< 0.001

1.57 (1.55, 1.58)

< 0.001

Solid organ

1.72 (1.68, 1.73)

< 0.001

1.70 (1.68, 1.72)

< 0.001

General

1.46 (1.46, 1.48)

< 0.001

1.34 (1.34, 1.35)

< 0.001

Knee/hip

1.22 (1.21, 1.22)

< 0.001

1.20 (1.20, 1.21)

< 0.001

Reproductive organ

1.49 (1.48, 1.51)

< 0.001

1.43 (1.42, 1.43)

< 0.001

Spinal

1.79 (1.77, 1.80)

< 0.001

1.82 (1.80, 1.84)

< 0.001

CI = confidence interval, GEE = generalized estimating equations, OLS = ordinary least squares

*Ratio of average total costs for patients with bleeding-related complications versus those without complications

Note: 15 patients were excluded from the economic analysis because their hospital costs exceeded $1,000,000.

GEE model estimated by gamma log link function with total costs as the dependent variable, adjusted for baseline clinical, demographic, hospital characteristics and clustering on hospital.

OLS model with log total costs as the dependent variable, adjusted for baseline clinical, demographic, and hospital characteristics.

Stokes et al. BMC Health Services Research 2011 11:135   doi:10.1186/1472-6963-11-135