Skip to main content

Table 4 Cost modeling approaches for prediction of pharmaceutical expenditure.

From: Variability in prescription drug expenditures explained by adjusted clinical groups (ACG) case-mix: A cross-sectional study of patient electronic records in primary care

 

Total population

Two-part model

    

Expenditure incurrence

Level of expenditure

 

Linear mixed models

Logistic regression

Linear mixed models

Adult population

       

Fixed effects

V

SE

%V

%EV

V

SE

%EV

Age*

  

1.4%

0.6%

10.5%

 

10.5%

Random factors

       

ACG

5.01

0.89

44.7%

28.8%

2.78

0.49

35.4%

Physician (center)

0.27

0.06

2.4%

1.5%

0.14

0.03

1.8%

Residual

5.77

0.04

51.5%

69.1%

4.11

0.03

52.3%

Total

11.21

 

100%

100%

7.85

 

100%

Pediatric population

       

Fixed effects

V

SE

%V

%EV

V

SE

%EV

Age*

  

0.1%

0.4%

  

0.0%

Random factors

       

ACG

1.22

0.28

22.2%

20.6%

0.66

0.17

22.4%

Physician (center)

1.62

0.69

29.5%

38.8%

0.32

0.14

10.9%

Residual

2.66

0.04

48.3%

40.2%

1.98

0.04

66.7%

Total

5.51

 

100%

100%

2.96

 

100%

  1. Statistical tests applied to the total population and two-part model. Linear mixed models include: estimation of variance (V), standard error (SE) and percentage of explained variance (%EV).
  2. * Age fixed effects include linear and quadratic terms.
Back to article page

BMC Health Services Research

ISSN: 1472-6963

Contact us