Table 4.
Results of logistic regression and linear regression models for 12 months
| Variables | Model I: Logistic regression (adherent vs nonadherent) | Model II: Linear regression | ||
|---|---|---|---|---|
| OR (95% CI) | p-value | β estimate (SE) | p-value | |
| Intervention | ||||
| Yes vs no | 1.46 (1.05–2.04) | 0.02 | 0.06 (0.02) | 0.02 |
| Previous trajectories | ||||
| Gaps in adherence vs rapid decline | 2.65 (1.69–4.18) | <0.0001 | 0.24 (0.03) | <0.0001 |
| Gradual decline vs rapid decline | 2.19 (1.38–3.46) | 0.0008 | 0.15 (0.03) | <0.0001 |
| Gender | ||||
| Male vs female | 0.88 (0.63–1.22) | 0.46 | − 0.01 (0.02) | 0.67 |
| Age (y) | ||||
| ≥ 70 vs < 70 | 0.89 (0.65–1.24) | 0.52 | − 0.03 (0.02) | 0.19 |
| Health Plan | ||||
| No subsidy vs low-income subsidy | 0.89 (0.64–1.24) | 0.50 | − 0.01 (0.02) | 0.47 |
| Myocardial infarction | ||||
| Yes vs no | 1.03 (0.27–3.88) | 0.96 | 0.07 (0.10) | 0.45 |
| Depression | ||||
| Yes vs no | 0.78 (0.47–1.29) | 0.34 | − 0.07 (0.03) | 0.05 |
| Congestive heart failure | ||||
| Yes vs no | 1.26 (0.64–2.48) | 0.48 | 0.09 (0.05) | 0.07 |
| Stroke | ||||
| Yes vs no | 1.04 (0.44–2.41) | 0.92 | − 0.03 (0.06) | 0.55 |
| Coronary artery disease | ||||
| Yes vs no | 1.08 (0.70–1.65) | 0.71 | − 0.02 (0.03) | 0.44 |
| Refill type | ||||
| ≥ 90 days vs < 90 days | 1.13 (0.66–1.94) | 0.64 | 0.02 (0.04) | 0.59 |
| Number of other medications on index date | ||||
| >2 vs ≤ 2 | 1.24 (0.77–2.02) | 0.36 | 0.01 (0.03) | 0.68 |
| Previous hospitalizations | ||||
| ≥ 1 vs none | 0.70 (0.38–1.28) | 0.24 | − 0.12 (0.04) | 0.01 |
| Prevalent users | ||||
| Yes vs no | 1.69 (1.05–2.72) | 0.02 | 0.09 (0.03) | 0.006 |
| Prescriber specialty | ||||
| General vs specialty | 0.64 (0.38–1.10) | 0.11 | − 0.05 (0.04) | 0.18 |
| Regimen complexity | 0.98 (0.93–1.03) | 0.54 | − 0.003 (0.004) | 0.43 |
| CMS risk score | 0.90 (0.73–1.10) | 0.33 | − 0.02 (0.01) | 0.06 |
CI confidence interval, CMS Centers for Medicare and Medicaid Services, OR odds ratio, SE standard error