Multivariable Cox regression analysis for associations with HIV seroconversion in a two-year cohort of injection drug users in Karachi, Pakistan (n = 474).

a Government organization, non-governmental organization, community, and friends. b Non-Muslim included 54 Christians and 2 Hindus vs. 418 Muslims. ≈US$ 56.c Approximately 90 rupees per US dollar in this time period, such that 5000 rupees d Pharmacy, friends, and hospital garbage. e Drop-in center (DIC) and mobile service unit (MSU). = 0.046).f If source of registration, physical disability, and monthly income, are excluded from the above Cox regression to achieve a more parsimonious model, then the effect sizes are similar and source of syringes becomes a significant predictor (p