Table 3.
Estimated size of the injecting population in Bristol in 2011: assessment of sensitivity of results to incorporation of covariates and source dependencies.
| Log‐linear model | Men 15–24 | Men 25–34 | Men 35–64 | Women 15–24 | Women 25–34 | Women 35–64 |
|---|---|---|---|---|---|---|
| Observed | 66 | 638 | 1091 | 54 | 301 | 300 |
| Covariate capture–recapture models | ||||||
| 1. No covariates | 110 (90, 130) | 750 (680, 840) | 980 (890, 1100) | 70 (60, 80) | 250 (220, 290) | 180 (160, 220) |
| 2. Gender, age | 110 (80, 170) | 710 (620, 840) | 1010 (870, 1180) | 70 (50, 120) | 240 (200, 320) | 190 (150, 260) |
| 3. Gender, age, housing | 130 (90, 210) | 820 (690, 1010) | 1140 (960, 1390) | 80 (60, 150) | 280 (220, 380) | 220 (160, 300) |
| Extensions of model 3 to allow for source dependencies | ||||||
| 4. Treatment × beedle | 120 (80, 210) | 830 (670, 1080) | 1300 (1030, 1710) | 90 (50, 180) | 320 (230, 490) | 280 (190, 450) |
| 5. Treatment × CJIT | 250 (130, 670) | 970 (740, 1350) | 1150 (900, 1540) | 120 (60, 330) | 250 (180, 390) | 170 (120, 250) |
| 6. Needle exchange × CJIT | 90 (70, 150) | 750 (620, 930) | 1060 (890, 1300) | 70 (50, 130) | 300 (220, 430) | 230 (170, 350) |
| 7. Treatment × CJIT + needle x CJIT | 190 (90, 870) | 840 (620, 1250) | 970 (770, 1320) | 120 (60, 620) | 270 (180, 490) | 170 (120, 300) |
| Capture–recapture subject to constraints and accounting for mortality data | ||||||
| 3. a Gender, age, housing | 100 (80, 160) | 800 (680, 970) | 1220 (1100, 1460) | 80 (60, 130) | 340 (300, 440) | 320 (300, 380) |
| 4. a Treatment × needle exchange | 100 (70, 160) | 790 (670, 1000) | 1300 (1110, 1640) | 80 (60, 160) | 350 (300, 490) | 330 (300, 460) |
| 5. a Treatment × CJIT | 150 (90, 300) | 870 (700, 1170) | 1290 (1110, 1670) | 130 (70, 240) | 360 (300, 510) | 320 (300, 390) |
| 6. a Needle exchange × CJIT | 90 (70, 120) | 740 (650, 900) | 1160 (1090, 1370) | 70 (60, 120) | 350 (300, 470) | 320 (300, 410) |
| 7. a Treatment × CJIT + needle x CJIT | 100 (70, 250) | 780 (650, 1060) | 1180 (1090, 1480) | 100 (60, 270) | 370 (300, 550) | 330 (300, 430) |
| Prevalence estimates from final selected model | ||||||
| Population | 35100 | 39800 | 76200 | 35200 | 37700 | 73700 |
| Prevalence | 0.3% (0.2, 0.3) | 1.9% (1.6, 2.3) | 1.5% (1.4, 1.8) | 0.2% (0.2, 0.3) | 0.9% (0.8, 1.2) | 0.4% (0.4, 0.6) |
Estimates shown are posterior medians from Bayesian analyses with 95% credible intervals, rounded to the nearest 10. Models with an
a
are extensions of the corresponding capture–recapture (CRC) model to incorporate external data. Final estimates, from the best‐fitting model, are shaded. CJIT = Criminal Justice Intervention Team.