TABLE 2—
2005 (n = 187) | 2011 (n = 196) | |
Characteristic | Generalized Ordered Logit Models,a OR (95% CI) | Generalized Ordered Logit Models,a OR (95% CI) |
Prevalence of injection drug users, % | ||
< 25 (Ref) | 1.00 | 1.00 |
25–74 | 1.18 (0.51, 2.72) | 1.72 (0.79, 3.76) |
≥ 75 | 0.75 (0.29, 1.97) | 0.84 (0.32, 2.19) |
African American clients, % | ||
< 10 (Ref) | 1.00 | 1.00 |
≥ 10 | 0.97 (0.51, 1.85) | |
≥ 10 (no vs off-site and on-site testing)b | 0.30* (0.11, 0.80) | |
≥ 10 (no and off-site vs on-site testing)b | 1.11 (0.54, 2.29) | |
Hispanic clients, % | ||
< 10 (Ref) | 1.00 | 1.00 |
≥ 10 | 1.50 (0.82, 2.72) | 1.01 (0.53, 1.92) |
Revenue from federal government | ||
None (Ref) | 1.00 | 1.00 |
≥ 1% | 2.03* (1.01, 4.07) | 1.88 (0.91, 3.89) |
Revenue from private insurance | ||
None (Ref) | 1.00 | 1.00 |
≥ 1% | 0.58* (0.30, 1.13) | 1.01 (0.48, 2.10) |
Staff-to-client ratio | ||
Log | 1.69* (1.16, 2.46) | 1.50 (0.86, 2.61) |
CARF accreditation | ||
No (Ref) | 1.00 | 1.00 |
Yes | 1.78 (0.85, 3.75) | 1.54 (0.75, 3.16) |
Ownership | ||
Private non-for-profit (Ref) | 1.00 | 1.00 |
Private for-profit | 1.84 (0.85, 3.98) | 1.16 (0.53, 2.56) |
Public | 1.40 (0.50, 3.92) | |
Public (no vs off-site and on-site testing)b | 0.51 (0.15, 1.78) | |
Public (no and off-site vs on-site testing)b | 2.33 (0.78, 6.99) | |
Hospital affiliation | ||
No (Ref) | 1.00 | 1.00 |
Yes (no vs off-site and on-site testing)b | 1.46 (0.42, 5.13) | . . .c |
Yes (no and off-site vs on-site testing)b | 7.71** (2.89, 20.53) | 3.92* (1.35, 11.38) |
Method of treatment | ||
Methadone only (Ref) | 1.00 | 1.00 |
Buprenorphine only | 0.48 (0.19, 1.19) | |
Buprenorphine only (no vs off-site and on-site testing)b | 1.94 (0.50, 7.45) | |
Buprenorphine only (no and off-site vs on-site testing)b | 0.32* (0.11, 0.90) | |
Both methadone + buprenorphine | 2.65 (0.85, 8.27) | 2.62* (1.26, 5.45) |
Note. CARF = CARF International (formerly known as the Commission on the Accreditation of Rehabilitation Facilities); CI = confidence interval; OR = odds ratio. The 95% CIs were estimated with robust SEs.
Dependent variable for generalized ordered logit model is coded so that higher values are equal to favorable assessment of type of HCV testing. Therefore, 1 = no HCV testing, 2 = off-site only HCV testing, 3 = on-site HCV testing.
For variables that violate the proportional odds assumption, we present 2 odds ratios: in the first, ORs are drawn from the comparison of no HCV testing (reference) vs off-site and on-site testing combined; in the second, the ORs are drawn from the comparison of no HCV testing and off-site testing (reference) vs on-site HCV testing.
The proportional odds assumption was violated, but there were too few cases to estimate this coefficient separately.
*P < .05; **P < .001.