Skip to main content

Table 2 Summaries of differential predictors and biomarkers used in the model for the development and validation datasets. (SD) [percentage of missing values]

From: Dirichlet process mixture models to impute missing predictor data in counterfactual prediction models: an application to predict optimal type 2 diabetes therapy

 

Development Dataset

Validation Dataset

Drug Taken

DPP4-inhibitor

SGLT2-inhibitor

DPP4-inhibitor

SGLT2-inhibitor

(n = 9,974)

(n = 6,152)

(n = 6,650)

(n = 4,101)

Age (years)

63.9 (10.8)

59.9 (9.1)

65.0 (10.7)

60.2 (9.3)

Number of Past Drugs

     2

3,884

1,167

2,556

731

     3

3,653

1,693

2,457

1,115

     4

972

1,569

683

1,045

     5+

186

945

117

672

Number of Current Drugs

     0

523

149

309

93

     1

5,078

2,191

3,424

1,418

     2

3,000

2,449

1,993

1,643

     3+

94

585

87

409

HbA1c (mmol/mol)

72.9 (13.5)

76.6 (14.2)

72.6 (13.2)

77.1 (14.1)

eGFR (mL/min/1.3m2)

83.1 (17.4) [0.2%]

88.8 (14.7) [0.3%]

84.9 (17.2) [0.2%]

88.6 (14.8) [0.4%]

ALT (IU/L) (logged)

3.3 (0.5) [10.0%]

3.4 (0.5) [10.4%]

3.3 (0.5) [10.1%]

3.4 (0.5) [10.8%]

BMI (kg/m2)

32.3 (6.4) [3.2%]

34.4 (6.5) [2.4%]

32.3 (6.4) [2.9%]

34.4 (6.6) [2.5%]

Outcome HbA1c (mmol/mol)

65.1 (16.0)

64.9 (14.2)

65.0 (16.2)

65.1 (14.6)

HbA1c_Month: Month of outcome HbA1c

9.2 (3.5)

9.0 (3.5)

9.2 (3.4)

9.0 (3.4)