Table 2.
Estimated size of the injecting population in Bristol in 2011: assessment of sensitivity of results to incorporation of covariates and source dependencies.
| Model | Estimated number of PWID (95% Cr‐I) | Estimated prevalence (%) of PWID (95% Cr‐I) | Model fit | ||||||
|---|---|---|---|---|---|---|---|---|---|
| Male | Female | Total | Male | Female | Total |
|
pD | DIC | |
| Covariate capture–recapture models | |||||||||
| 1. No covariates | 1850 (1680, 2050) | 500 (450, 570) | 2350 (2140, 2600) | 1.2% (1.1, 1.4) | 0.3% (0.3, 0.4) | 0.8% (0.7, 0.9) | 217 | 13 | 230 |
| 2. Gender, age | 1830 (1650, 2060) | 510 (420, 660) | 2350 (2140, 2620) | 1.2% (1.1, 1.4) | 0.3% (0.3, 0.4) | 0.8% (0.7, 0.9) | 197 | 22 | 219 |
| 3. Gender, age, housing | 2100 (1830, 2470) | 580 (460, 780) | 2690 (2360, 3140) | 1.4% (1.2, 1.6) | 0.4% (0.3, 0.5) | 0.9% (0.8, 1.1) | 98 | 25 | 123 |
| Extensions of model 3 to allow for source dependencies | |||||||||
| 4. T×N | 2260 (1890, 2820) | 700 (500, 1040) | 2970 (2490, 3680) | 1.5% (1.2, 1.9) | 0.5% (0.3, 0.7) | 1.0% (0.8, 1.2) | 98 | 30 | 127 |
| 5. T×C | 2400 (1930, 3170) | 550 (400, 860) | 2960 (2430, 3890) | 1.6% (1.3, 2.1) | 0.4% (0.3, 0.6) | 1.0% (0.8, 1.3) | 88 | 30 | 117 |
| 6. N×C | 1910 (1670, 2250) | 610 (470, 860) | 2530 (2210, 2980) | 1.3% (1.1, 1.5) | 0.4% (0.3, 0.6) | 0.9% (0.7, 1.0) | 90 | 30 | 120 |
| 7. T×C + N×C | 2060 (1640, 2950) | 590 (400, 1190) | 2670 (2140, 3940) | 1.4% (1.1, 2.0) | 0.4% (0.3, 0.8) | 0.9% (0.7, 1.3) | 86 | 35 | 121 |
| Capture–recapture subject to constraints and accounting for mortality data a | |||||||||
| 3. * Gender, age, housingb | 2130 (1920, 2470) | 750 (680, 890) | 2890 (2650, 3260) | 1.4% (1.3, 1.6) | 0.5% (0.5, 0.6) | 1.0% (0.9, 1.1) | 102 | 23 | 125 |
| 4. * T×Nb | 2200 (1920, 2670) | 780 (690, 1010) | 3000 (2670, 3520) | 1.5% (1.3, 1.8) | 0.5% (0.5, 0.7) | 1.0% (0.9, 1.2) | 97 | 28 | 126 |
| 5. * T×Cb | 2330 (1990, 2940) | 820 (710, 1030) | 3170 (2760, 3810) | 1.5% (1.3, 1.9) | 0.6% (0.5, 0.7) | 1.1% (0.9, 1.3) | 95 | 28 | 123 |
| 6. * N×Cb | 2000 (1850, 2290) | 760 (680, 930) | 2770 (2570, 3110) | 1.3% (1.2, 1.5) | 0.5% (0.5, 0.6) | 0.9% (0.9, 1.0) | 91 | 27 | 119 |
| 7. * T×C + N×Cb | 2080 (1870, 2550) | 810 (690, 1090) | 2910 (2620, 3490) | 1.4% (1.2, 1.7) | 0.6% (0.5, 0.7) | 1.0% (0.9, 1.2) | 91 | 32 | 123 |
DIC values should not be compared between standard capture–recapture models and those subject to constraints. Estimates shown are posterior medians from Bayesian analyses with 95% credible intervals (Cr‐Is), rounded to the nearest 10. = residual deviance, pD = effective number of parameters, DIC = deviance information criterion (which is calculated as + pD). Models marked with a
are extensions of the corresponding CRC models to incorporate external data. Final estimates, from the best‐fitting model, are shaded. PWID = people who inject drugs. T = Treatment, N = needle exchange, C = Criminal Justice Intervention Team; e.g. ‘T×N’ refers to a treatment by needle exchange interaction term in the log‐linear model.