Table 4.

Beta regressions estimating the effect of user characteristics on user engagement (user activity ratio) for nonautomated modules.

User characteristics			App module
			Medication (n=3933; average UARa=9 days)							Food (n=3821; average UAR=9 days)
			MEb		P value		Days		ME		P value		Days
Sociodemographic
	Gender (reference=male)
		Female		−0.019		<.001		−3.42		−0.014		<.001		−2.52
		Unknown		−0.020		<.001		−3.60		−0.015		<.001		−2.70
	Age (reference=30-40 years)
		<30 years		0.000		.99		0.00		0.000		.59		0.00
		40-50 years		0.003		.23		0.54		0.003		.06		0.54
		50-60 years		0.007		.004		1.26		0.007		<.001		1.26
		60-70 years		0.009		.003		1.62		0.010		<.001		1.80
		>70 years		0.008		.098		1.44		0.009		.03		1.62
		Unknown		−0.003		.69		−0.54		0.008		.42		1.44
BMI ( reference =normal)
	Underweight (BMI<18.5)			−0.007		.73		−1.26		−0.020		.007		−3.60
	Overweight (25≤BMI<30)			0.010		.12		1.80		0.009		.19		1.62
	Obese (30≤BMI)			0.003		.50		0.54		−0.001		.81		−0.18
	Unknown			0.000		.99		0.00		−0.003		.50		−0.54
Diagnosis
	Type (reference=type 1 diabetes)
		Type 2 diabetes		0.004		.095		0.72		0.000		.89		0.00
		Prediabetes		0.016		.097		2.88		0.007		.10		1.26
		Other		0.018		.05		3.24		0.003		.48		0.54
		Unknown		0.038		<.001		6.84		0.014		.01		2.52
	Recently diagnosed (reference=no)
		Yes		0.002		.16		0.36		0.003		.09		0.54
		Unknown		−0.012		.16		−2.16		−0.003		.46		−0.54
Disease status
	Treatment (reference=noninsulin)
		Insulin treatment		0.002		.17		0.36		−0.004		.02		−0.72
		Unknown		0.014		.12		2.52		0.007		.02		1.26
	Blood glucose in control (reference=no)
		Yes		−0.005		.20		−0.90		0.000		.97		0.00
		Unknown		−0.015		<.001		−2.70		−0.012		<.001		−2.16
Usage
	Number of modules used			0.015		<.001		2.70		0.009		<.001		1.62

^aUAR: user activity ratio.

^bME: marginal effect reflecting the change in the outcome as the continuous variable changes by one unit (from 0 to 1 for categorical variables).