Table 2.

Results from hierarchical linear models analyses of walking, health and fitness indices collected at baseline, week 12 (end of treatment), and week 24 (follow-up).

	β1 coefficient	Standard error	T ratio (df = 43)	p value
Steps walked
Control	7.0961	3.3132	2.14	.03
Reinforcement	21.8935	3.8831	5.64	.001
Group × time interaction	−14.7974	5.1045	−2.90	.005
Systolic blood pressure
Control	−0.0090	0.0292	−0.31	.76
Reinforcement	−0.0786	0.0148	−5.33	.001
Group × time interaction	0.0696	0.0327	2.13	.03
Diastolic blood pressure
Control	−0.0101	0.0110	−0.92	0.36
Reinforcement	−0.0468	0.0137	−3.41	.001
Group × time interaction	0.0367	0.0176	2.09	.04
Weight
Control	−0.0108	0.0095	−1.14	.26
Reinforcement	−0.0414	0.0131	−3.18	.002
Group × time interaction	0.0306	0.0161	1.90	.05
Waist circumference
Control	−0.0134	0.0065	−2.06	.04
Reinforcement	−0.0286	0.0089	−3.23	.002
Group × time interaction	0.0152	0.0110	1.38	.17
Six minute walk test
Control	−0.0453	0.0606	−0.75	.46
Reinforcement	0.1474	0.0638	2.31	.02
Group × time interaction	0.1927	0.0880	2.19	.03
Sit-to-stand test
Control	0.3253	0.2536	−1.28	.20
Reinforcement	−1.1846	0.3454	−3.43	.001
Group × time interaction	0.8593	0.4285	2.01	.05

Notes: Although 43 patients were randomized and included in these analyses, not all participants (n = 3) completed the 24-week follow-up, and extreme outliers were excluded at some time points, or in the case of anti-hypertensive medication changes for blood pressure data (see methods section). β1 coefficients represent the slopes, and slopes with p values < .05 are significantly different from 0 (i.e., reflect significant changes in slope over time). When group × time interaction effects are significant (p < .05), the slopes of the two conditions differ.