Table 3 Relationship between baseline characteristics and membership in clusters of responsiveness to behavioral factors

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

Referent: Group 3

Baseline Characteristic

Odds Ratio

(95% CI) for Group 1

Odds Ratio (95% CI) for Group 2

Age: ≥65 years vs. <65 years

0.80 (0.16–4.12)

0.33 (0.03–3.92)

Sex: Female vs. Male

1.33 (0.25–7.01)

1.67 (0.18–15.13)

Race/ethnicity: White vs. Non-White

0.23 (0.04–1.30)

1.36 (0.11–16.58)

Partner status: Married/partnered vs. other

7.70 (1.15–51.15)a

0.73 (0.06–8.92)

Diabetes control: baseline HbA1c < 9% vs. ≥9%

0.19 (0.03–1.14)

Too small to compute

Diabetes medication use: ≥4 years vs. <4 years

4.80 (0.48–48.46)

0.20 (0.02–2.39)

Number of medications: 1 vs. ≥2

2.10 (0.32–13.61)

1.80 (0.15–21.48)

Self-reported baseline adherence: >1 dose missed vs. ≤1 dose missed

0.64 (0.12–3.53)

1.29 (0.14–11.54)

  1. Among 29 intervention arm patients, we used exploratory multinomial logistic regression comparing baseline patient characteristics of those in Group 1 (responding best to observed feedback [history], n = 9, Orange in Fig. 6) and Group 2 (responding best to social reinforcement or observed feedback, n = 4, Yellow in Fig. 6) using Group 3 as the referent group (responding equally to each factor, n = 16, Blue in Fig. 6).
  2. ap < 0.05.
  3. CI Confidence Interval.
Back to article page