Table 3.
Mean and Standard Deviation on Adherence Measures by Group
| Self-Report Metric | Time | Usual Care | Reminder Calls | Motivational Interviewing | Omnibus Test Between Groups |
|---|---|---|---|---|---|
| MEMS | Pre | 78.7% (19.5%) | 75.7% (21.4%) | 79.7% (18.2%) | F(2, 162) = 4.14, |
| (% of days) | Post | 76.0% (20.4%) | 83.0% (23.5%) | 79.3% (22.1%) | p = .02, η2 = .05 |
| 3-Day Recall | Pre | 92.4% (19.1%) | 91.4% (21.9%) | 92.0% (19.3%) | F(2, 176) = 3.66, |
| (% of days) | Post | 94.2% (10.4%) | 98.2% ( 8.2%) | 97.3% ( 7.7%) | p = .03, η2 = .04 |
| Morisky Scale | Pre | 93.0% ( 9.4%) | 93.0% (10.7%) | 91.0% (13.5%) | F(2, 176) = 4.89, |
| (% calculated from 1–7 scale) | Post | 94.9% (13.7%) | 97.4% ( 9.9%) | 96.9% (12.4%) | p = .009, η2 = .05 |
Note. Average adherence was calculated for MEMS as the overall percentage of doses taken as prescribed during the baseline phase (2 months) and during the intervention phase (3 months); for recall as the percentage out of 3 days that the participant reported taking their medication correctly; and for the Morisky scale as a transformation of the raw average scale score (on a 1–7 Likert-type scale) into a percentage by this formula: (average - 1)/6. The transformed Morisky scale score was computed to aid interpretation in descriptive analyses by converting all variables to the same 0% – 100% range, with no effect on the variable’s underlying distribution. Adjusted post-intervention means are presented. Each analysis (multilevel modeling for MEMS or analysis of covariance for the other outcomes) controlled for the following covariates: pretest scores on the same measure; recruitment site; the urn randomization stratification variables of age, visual field, and MEMS-based baseline adherence; plus marital status which differed by experimental group at baseline, and number of doses per day, number of comorbidities, number of other medications, treatment duration, and intra-ocular pressure, all of which differed by site at baseline.