Table 2.
Arm means and differences between arm means for all outcomesa
| Outcomes | Intervention (n=319) |
Control (n=322) |
Mean arm difference (95% CI) |
P-value of arm effect: covariate- adjusted (unadjusted) |
Holm’s Procedureb Alpha Level |
||
|---|---|---|---|---|---|---|---|
| Baseline Mean (se) |
Change Mean (se) |
Baseline Mean (se) |
Change Mean (se) |
||||
| Primary Outcome | |||||||
| SF-Physical Function | 75.9 (1.1) | −2.15 (0.9) | 75.6 (1.1) | −4.84 (0.9) | 2.69 | 0.03 (0.03) | |
| Range 10–100 | (0.17, 5.21) | ||||||
| Secondary Outcomes | |||||||
| LLF- Basic Lower Extremity | 78.4 (0.8) | +0.34 (0.6) | 78.1 (0.9) | −1.89 (0.6) | 2.24 (0.56, 3.91) | 0.005 (0.005) | 0.0050 |
| Range 45.6–100 | |||||||
| LLF-Advanced Lower Extremity | 52.8 (0.8) | −0.37 (0.5) | 52.9 (0.8) | −2.30 (0.6) | 1.92 (0.45, 3.39) | 0.02 (0.01) | 0.0063 |
| Range 0–100 | |||||||
| Behavioral Targets | |||||||
| Duration of Strength Exercise (minutes/wk) | 7.0 (2.1) | +18.7 (2.4) | 11.5 (2.5) | +0.5 (2.7) | 18.2 (11.2, 25.2) | <0.0001 (<0.0001) | 0.0029 |
| Range 0–600c | |||||||
| Duration of Endurance Exercise (minutes/wk) | 24.6 (2.1) | +36.3 (4.9) | 28.7 (2.3) | +23.4 (5.6) | 12.9 (1.89, 27.6) | 0.004 (0.003) | 0.0042 |
| Range 0–149 | |||||||
| Strength Exercise Frequency (sessions/wk) | 0.5 (0.1) | +1.4 (0.2) | 0.5 (0.1) | +0.2 (0.1) | 1.12 (0.70, 1.54) | <0.0001 (<0.0001) | 0.0031 |
| Range 0–7 | |||||||
| Endurance Exercise Frequency (sessions/wk) | 1.6 (0.1) | +1.6 (0.2) | 1.8 (0.2) | +0.5 (0.2) | 1.05 (0.39, 1.72) | 0.005 | 0.0045 |
| Range 0–15 | |||||||
| Fruits & Vegetables (daily servings) | 3.72 (0.1) | +1.24 (0.14) | 3.54 (0.1) | +0.13 (0.11) | 1.11 (0.76, 1.47) | <0.0001 (<0.0001) | 0.0033 |
| Range 0–15.80 | |||||||
| Saturated Fat Intake (g/day) | 19.6 (0.5) | −3.06 (0.51) | 19.32 (0.5) | −1.07 (0.49) | −1.99 (−0.58, −3.40) | 0.002 (<0.0001) | 0.0038 |
| Range 2 −57 | |||||||
| Weight (kg) | 85.7 (0.7) | −2.06 (0.19) | 84.7 (0.7) | −0.92 (0.2) | −1.14 | <0.0001 | 0.0036 |
| Range 59.1 – 125.5 | (−0.59, −1.69) | ||||||
| Body mass index | 29.1 (0.2) | −0.69 (0.07) | 29.2 (0.2) | −.031 (0.08) | −0.38 | <0.0001 | 0.0038 |
| Range 25.0 – 47.0 | (−0.19, −0.57) | ||||||
| Health Quality of Life | |||||||
| SF-General Health | 71.8 (0.9) | +0.77 (0.72) | 72.6 (0.9) | −1.94 (0.80) | 2.71 | 0.03 (0.02) | 0.0071 |
| Range 15 −100 | (0.58, 4.84) | ||||||
| SF-Pain | 72.2 (1.2) | −0.78 (1.07) | 72.6 (1.2) | −3.19 (1.22) | 2.40 | 0.16 (0.13) | 0.0167 |
| Range 10–100 | (−0.79, 5.59) | ||||||
| SF-Vitality | 61.9 (0.9) | −0.47 (0.89) | 61.5 (1.0) | −2.42 (0.98) | 1.95 | 0.10 (0.09) | 0.0125 |
| Range 0 – 100 | (−0.64, 4.55) | ||||||
| SF-Social Functioning | 90.2 (1.0) | −1.29 (1.05) | 90.8 (0.9) | −5.05 (1.22) | 3.75 | 0.03 (0.02) | 0.0083 |
| Range 12.5 – 100 | (0.58, 6.92) | ||||||
| SF- Mental Health | 85.6 (0.7) | +0.50 (0.53) | 86.3 (0.7) | −2.04 (0.74) | 2.54 | 0.01 (0.08) | 0.0056 |
| Range 32 – 100 | (0.75, 4.33) | ||||||
| SF- Role Physical | 75.7 (1.9.) | −2.43 (2.02) | 78.6 (1.9) | −4.68 (2.14) | 2.25 | 0.32 (0.30) | 0.0250 |
| Range 0 – 100 | (−3.54, 8.05) | ||||||
| SF-Role Emotional | 92.1 (1.2) | −0.73 (1.32) | 92.0 (1.2) | −0.62 (1.38) | −0.11 | 0.93 (0.98) | 0.0500 |
| Range 0 – 100 | (−3.86, 3.64) | ||||||
Abbreviations: se, standard error; CI, confidence interval
Since the observed means and covariate-adjusted means were almost identical, only the observed means are shown. Both covariate-adjusted and unadjusted p-values for the test of the arm effects are given. Missing year-1 outcomes were imputed to baseline value.
Holm’s procedure first ranks the 17 p-values from lowest to highest. The first (lowest) p-value has to be less than 0.0029 (0.05/17) to be statistically significant and to permit continuation to the other t-tests. The Holm’s procedure continues sequentially in this fashion using alpha levels of 0.0.0031 (0.05/16), 0.0033 (0.05/15), … , and 0.05 (0.05/1) for the remaining 16 tests, respectively.
600 minutes of light strength training is an outlier (next highest value is 210). Because we were unable to validate the accuracy of this data point it was included in the analysis. Inclusion or exclusion of this value made no difference on the estimates of the p-value.