Table 3.

Logistic regression model^* to identify determinants of prevalence of antibodies to hepatitis C virus in injecting drug users

	Injecting drug users (n=509)†	No (%) positive for antibodies to hepatitis C virus (n=414)	Odds ratio (95% CI)	P value
Months spent in prison over past 10 years:
<3	40	19 (47.5)	1
3-11	49	36 (73.5)	2.3 (0. 8 to 7.1)	0.14
12-36	120	97 (80.8)	2.4 (0.9 to 6.6)	0.08
>36	296	260 (87.8)	2.9 (1.1 to 7.6)	0.03
Years since first injecting:
<3	85	55 (64.7)	1
⩾3	383	327 (85.40	2.9 (1.5 to 5.4)	0.001
Sharing needles in prison:
No	145	90 (62.1)	1
Yes	347	314 (90.5)	2.9 (1.5 to 5.7)	0.002
No of times injected in month before survey:
0	221	177 (80.1)	1
1-19	139	119 (85.6)	1.1 (0.5 to 2.1)	0.89
⩾20	74	67 (90.5)	3.0 (1.1 to 10.0)	0.05

Whole model χ²=53, P<0.0001. 

^*Initial model included variables age, sex, time spent in prison in preceding 10 years, smoking heroin, length of time since first injection, started injecting in prison, sharing practices in prison and outside prison, injecting frequency in prison, taking methadone before committal, ever had sex with a man inside or outside prison, ever treated for sexually transmitted infection, and use of condoms during heterosexual intercourse. Significant factors were retained in final model. Interaction between length of time spent in prison in past 10 years and number of years since first injecting drugs was not significant. 

^†Numbers do not always add up to total because not all respondents answered all questions.