Table 1 Baseline characteristics of trial patients

From: The impact of using reinforcement learning to personalize communication on medication adherence: findings from the REINFORCE trial

 

Reinforcement learning (n = 29)

Control (n = 31)

Absolute standardized differences

Age, mean (SD)

60.3 (11.4)

56.7 (12.9)

0.02

Female sex, n (%)

12 (41.4%)

14 (45.2%)

0.08

Race/ethnicity, n (%)

White

17 (58.6%)

18 (58.1%)

0.01

Black or African American

6 (20.7%)

8 (25.8%)

0.12

Other (Hispanic/Latino, Asian, Other)

6 (20.7%)

7 (22.6%)

0.05

Education level, n (%)

High school graduate or below

7 (24.1%)

5 (16.1%)

0.20

Some college/college graduate

22 (51.7%)

19 (61.3%)

0.19

Post-graduate

7 (24.1%)

7 (22.6%)

0.04

Married or partnered, n (%)

13 (44.8%)

17 (54.8%)

0.20

Baseline HbA1c, mean (SD)

8.99 (1.39)

9.10 (1.47)

0.06

<9, n (%)

17 (58.6%)

17 (54.8%)

0.08

≥9, n (%)

12 (41.4%)

14 (45.2%)

0.08

Diabetes medication use (self-reported), n (%)

<4 years

10 (34.5%)

13 (41.9%)

0.15

≥4 years

19 (65.5%)

18 (58.1%)

0.15

3 or more unique physicians, n (%)

19 (65.5%)

16 (51.6%)

0.29

Number of medications, n (%)

1

20 (69.0%)

25 (80.6%)

0.27

≥2

9 (31.0%)

6 (19.4%)

0.27

Non-adherence (number of self-reported doses missed in prior 30 days), %

≤1

17 (58.6%)

17 (54.8%)

0.08

>1

12 (41.4%)

14 (45.2%)

0.08

Automaticitya with medication-taking, n (%)

5 (17.2%)

10 (32.3%)

0.36

Full patient health activationb

16 (55.2%)

15 (48.4%)

0.13

  1. This table shows the baseline characteristics of the 60 randomized patients to the trial.
  2. SD Standard deviation, HbA1 glycated hemoglobin A1c.
  3. aHighest possible score based on the automaticity subscale of the self-report behavioral automaticity index.
  4. bFull patient activation based on the consumer health activation index.