Table 4. Multinomial logistic regression predicting physical activity level as a categorical variable for all app users (n=4465).

Variablesa		Estimate (SE)		Z	P value	Odds ratio (95% CI)
Dependent variable contrast: inactive vs minimally active
	Show sensor info	0.331 (0.077)		4.308	<.001	1.393 (1.198-1.620)
	Goal setting	0.527 (0.095)		5.525	<.001	1.693 (1.405-2.041)
	Show goal progress	0.154 (0.102)		1.508	.13	1.166 (0.955-1.425)
	Energy analysis	0.192 (0.103)		1.861	.06	1.212 (0.990-1.485)
	Weight recording	0.139 (0.108)		1.284	.20	1.149 (0.930-1.419)
	Journaling	0.114 (0.104)		1.096	.27	1.120 (0.914-1.373)
	GPS/map	0.018 (0.107)		0.166	.87	1.018 (0.826-1.255)
	Show sleep information	0.072 (0.116)		0.621	.54	0.930 (0.741-1.168)
	Reward points	0.215 (0.124)		1.738	.08	1.240 (0.973-1.579)
	Blood pressure recording	0.246 (0.147)		1.671	.10	1.279 (0.958-1.707)
Dependent variable contrast: inactive vs HEPAb active
	Show sensor info	0.278 (0.081)		3.435	.001	1.320 (1.127-1.546)
	Goal setting	0.480 (0.099)		4.830	<.001	1.617 (1.330-1.964)
	Show goal progress	0.226 (0.106)		2.144	.03	1.254 (1.020-1.542)
	Energy analysis	0.265 (0.107)		2.491	.01	1.304 (1.058-1.607)
	Weight recording	0.104 (0.113)		0.918	.36	1.109 (0.889-1.384)
	Journaling	0.470 (0.103)		4.540	<.001	1.600 (1.306-1.960)
	GPS/map	0.306 (0.107)		2.847	.004	1.357 (1.100-1.675)
	Show sleep information	0.068 (0.119)		0.573	.57	1.070 (0.848-1.350)
	Reward points	0.106 (0.131)		0.807	.42	1.111 (0.860-1.437)
	Blood pressure recording	0.067 (0.156)		0.430	.67	1.069 (0.788-1.452)

^aThe independent variable was app function

^bHEPA: health-enhancing physical activity.