Table 5

Outcome measures model estimates

Models were created by using the outcome variable as a dependent variable, and using the variables listed under the ‘model variables’ as covariates. ‘Activity’ quantifies segments of 30 min of daily activity, ‘time’ determines the no of days since each patient started recording their activity, and ‘dominant arm’ was used to determine whether or not the patient suffered a stroke on their dominant arm.
Outcome	Model adjusted R²	Model fit p value	Model variables	Estimates	Estimate p value
Left-Right ratio improvement* (Ref. category: No improvement)	0.169†	<0.001	Activity Time Dominant arm (Ref. category: No)	1.005‡ 0.995‡ 6.684‡	<0.001 0.116 <0.001
PAM	0.012	0.002	Activity Time	−0.653 −0.022	0.002 0.290
VAS	0.135	<0.001	Activity Time	−0.056 0.29	0.025 0.000
MAL—how much	0.452	<0.001	Activity Time	0.304 0.012	<0.001 <0.001
MAL—how well	0.433	<0.001	Activity Time	0.324 0.001	<0.001 <0.001
mRS	0.267	<0.001	Activity Time	−0.161 −0.007	<0.001 <0.001
MoCA	0.002	0.158	Activity Time	0.002 0.000	0.272 0.160
EQ-5D-5L index	<0.001	0.161	Activity Time	0.020 0.002	<0.001 <0.001
EQ-5D-5L VAS	0.226	<0.001	Activity Time	1.252 0.241	<0.001 <0.001

*Left-right ratio improvement was measured using a logistic regression model, as opposed to other variables which were modelled using a linear model.
†Pseudo R² (McFadden).
‡Estimated for the left-right ratio improvement are show as ORs.
EQ-5D-5L, EuroQol-5 Dimensions-5 Level; MAL, Motor Activity Log; MoCA, Montreal Cognitive Assessment; mRS, modified Rankin Scale; PAM, Patient Activation Measure; VAS, Visual Analogue Scale.