. 2018 Jun 22;20(6):e207. doi: 10.2196/jmir.9225

Table 4.

Analysis of health care utilization.

Utilization			Cohort^a													Difference in slopes		P value^b
			BCBH-D intervention (N=558)							Matched controls (N=1669)								Difference in slopes		Cohorts time 1		Cohorts time 2
			Pre		Post		Slope		Pre			Post		Slope				Difference in slopes		Cohorts time 1		Cohorts time 2
All-cause medical utilization (visits per 1000 members per year)
	Inpatient	110		80		−30		100			110		10		−40		.10		.55		.15
	Emergency department	190		150		−40		190			260		70		−110		.004		.80		<.001
	Outpatient	22,880		17,100		–5780		22,880			22,590		−290		–5490		<.001		.99		<.001
Diabetes-related medical utilization (visits per 1000 members per year)
	Inpatient	10		0		−10		10			0		−10		0		.57		.75		.71
	Emergency department	10		0		−10		10			10		0		−10		.27		.91		.30
	Outpatient	6630		5200		−1430		6760			6530		−230		−1200		<.001		.68		<.001

^aAll subjects have 1 year of pretreatment time and 1 year of follow-up time.

^bAll analysis were conducted using generalized estimating equation repeated measures model with a logit link and negative binomial distribution for inpatient and emergency department visits and zero-inflated negative binomial distribution for outpatient services. The matched control cohort is the reference category in all analyses.