. 2018 Mar 12;178(4):542–553. doi: 10.1001/jamainternmed.2018.0150

Table 3. Effect of the LINK LA Intervention on Linkage to and Retention in HIV Primary Care After Jail Release^a.

Measurement Time	Intervention (95% CI) (n = 180)	Control (95% CI) (n = 176)	Difference (95% CI)^b^,^c	P Value^d
Linkage to HIV Care^e
0-3 Months (n = 312)	0.64 (0.57 to 0.72)	0.63 (0.56 to 0.71)	0.01 (−0.09 to 0.12)	.81
0-6 Months (n = 260)	0.89 (0.84 to 0.95)	0.77 (0.69 to 0.84)	0.12 (0.04 to 0.22)	.01
0-12 Months (n = 220)	0.92 (0.87 to 0.97)	0.88 (0.82 to 0.94)	0.04 (−0.04 to 0.12)	.32
Retention in HIV Care^f
Baseline (n = 350)	1.64 (1.29 to 2.00)	2.26 (1.74 to 2.77)	−0.61 (−1.24 to 0.01)^g	.054
3 Months (n = 312)	3.08 (2.56 to 3.61)	3.04 (2.52 to 3.55)	0.04 (−0.69 to 0.77)	.90
Change at 3 months (n = 307)^a^,^c	1.44 (0.89 to 1.98)^h	0.78 (0.18 to 1.38)^g	0.66 (−0.15 to 1.47)	.11
6 Months (n = 261)	2.15 (1.79 to 2.50)	2.15 (1.75 to 2.55)	−0.001 (−0.54 to 0.53)	>.99
Change at 6 months (n = 256)^a^,^c	0.50 (0.10 to 0.90)^g	−0.11 (−0.66 to 0.44)	0.61 (−0.07 to 1.30)	.08
12 Months (n = 235)	2.25 (1.87 to 2.64)	2.16 (1.79 to 2.53)	0.09 (−0.44 to 0.62)	.73
Change at 12 months (n = 232)^a^,^c	0.61 (0.17 to 1.06)^g	−0.10 (−0.63 to 0.44)	0.71 (0.01 to 1.40)^g	.047

Abbreviations: GLMM, generalized linear mixed model; HIV, human immunodeficiency virus; LINK LA, Linking Inmates to Care in Los Angeles; ZIP, zero-inflated Poisson.

^{^a}

Follow-up interview value minus baseline value.

^{^b}

Intervention arm estimate minus control arm estimate.

^{^c}

Difference and change values are not always precise totals owing to rounding.

^{^d}

P value for test between intervention and control arm values at each measurement time and difference-in-difference tested for change from baseline by study arm, based on a single regression model.

^{^e}

Probability of having HIV primary care visits from baseline (0 months) to each measurement time at 3, 6, and 12 months after release from incarceration by arm, and difference (with 95% CIs) estimated using unadjusted binomial model.

^{^f}

Estimated number of HIV primary care visits over the preceding 3 months, given linkage to care (at least 1 visit), using a repeated measures, random-intercept, ZIP model (a GLMM) with predictors of intervention arm, categorical time, and intervention × time interaction. The model accommodates data missing at random, and loss to follow-up was not different between study arms. ZIP models had 2 uncorrelated random effects, one for the zero-inflation component and the other for the Poisson component. All values are estimated per-12-month rate, given at least 1 visit (or hospital night) in the previous 12 months.

^{^g}

P < .05 for change from baseline.

^{^h}

P < .001 for change from baseline

Table 3. Effect of the LINK LA Intervention on Linkage to and Retention in HIV Primary Care After Jail Releasea.

Table 3. Effect of the LINK LA Intervention on Linkage to and Retention in HIV Primary Care After Jail Release^a.