Skip to main content

Table 2 Covariate balance across subgroups identified by glmtree algorithm

From: Identifying treatment effects of an informal caregiver education intervention to increase days in the community and decrease caregiver distress: a machine-learning secondary analysis of subgroup effects in the HI-FIVES randomized clinical trial

 

Outcome 1: days not in the community at follow-up

Outcome 2: caregiver depression scores at follow-up

Baseline Zarit Burden > 28

Baseline Zarit Burden ≤ 28

Baseline CESD-10 > 8

Baseline CESD-10 ≤ 8

Hi-Fives

n = 23

Control

n = 19

SMD

Hi-Fives

n = 95

Control

n = 104

SMD

Hi-Fives

n = 54

Control

n = 60

SMD

Hi-Fives

n = 64

Control

n = 63

SMD

Patient characteristics

 Baseline CESD-10 score, mean (SD)

15.04 (6.70)

15.47 (5.42)

0.07

7.73 (5.65)

7.56 (4.37)

0.03

–

–

–

–

–

–

 Patient age, mean (SD)

71.74 (13.14)

72.05 (9.96)

0.03

74.16 (10.76)

73.08 (12.51)

0.09

72.07 (11.87)

72.18 (11.94)

0.01

75.05 (10.59)

73.62 (12.35)

0.12

 Caregiver age, mean (SD)

59.65 (9.36)

59.05 (13.93)

0.05

59.93 (12.33)

62.30 (12.35)

0.19

58.72 (10.77)

62.05 (12.78)

0.28

60.84 (12.57)

61.56 (12.53)

0.06

 White race, n (%)

14 (60.9)

7 (36.8)

0.50

42 (44.2)

41 (39.4)

0.10

28 (51.9)

24 (40.0)

0.24

28 (43.8)

24 (38.1)

0.12

 Relationship satisfaction score, mean (SD)

41.83 (7.04)

40.42 (5.54)

0.22

46.85 (5.01)

46.78 (5.14)

0.01

43.48 (5.88)

44.17 (5.81)

0.12

47.89 (4.91)

47.35 (5.11)

0.11

 Zarit Burden score, mean (SD)

–

–

–

–

–

–

26.15 (9.00)

21.82 (9.38)

0.47

14.09 (7.19)

14.88 (7.73)

0.11

 Financial difficulty, n (%)

15 (65.2)

11 (57.9)

0.15

46 (48.4)

38 (36.5)

0.24

34 (63.0)

31 (51.7)

0.23

27 (42.2)

18 (28.6)

0.29

 Nosos comorbidity index, mean (SD)

3.84 (4.42)

3.12 (2.57)

0.20

3.35 (3.18)

3.81 (3.35)

0.14

3.63 (3.93)

4.14 (3.74)

0.13

3.28 (2.98)

3.28 (2.65)

0.00

  1. SMD Standardized mean differences; bolded SMD are greater than 0.2 and indicate poor balance; Statistics from imputed dataset; Super-user status, patient cognitive status, and caregiver Hispanic ethnicity were removed from the table because all cell sizes were smaller than 10