We repeated this analysis to assess the association between reported harmful alcohol use in the past year using the same demographic (gender, ethnicity, age) and psychiatric case-mix variables but sub- stituting ‘presence of drug use’ for ‘presence of harmful alcohol use’. The univariate odds ratio of harmful alcohol use for London CMHT patients was 1.18 compared with Nottingham and Sheffield (95% CI 0.68–2.04). However, as indicated by the confidence interval, the difference in odds is not statistically significant. When the above variables are included in the multiple regression model the adjusted odds ratio of alcohol misuse in London centres over Nottingham and Sheffield is reduced to a marginal level (AOR1⁄41.05, 95% CI 0.52–2.11)