Table 3. Regression analysis of association between composite harm score and health outcome measuresa.
| Health outcome measure | n | Adjusted estimates of CHS effect (95% CI)b | p–value |
|---|---|---|---|
| Physical health | |||
| Mortality | 288 | 1.47 (1.07–2.01) | 0.016 |
| Hepatitis C virus exposure | 279 | 1.56 (1.28–1.92) | <0.001 |
| Hepatitis C virus persistent infection | 185 | 1.29 (1.02–1.67) | 0.043 |
| Psychological health | |||
| Psychotic illness | 286 | ||
| None (reference) | 1.00 | ||
| Functional psychosis | 0.73 (0.56–0.93) | 0.014 | |
| Psychosis not otherwise specified | 1.11 (0.89–1.38) | 0.348 | |
| Substance–induced psychosis | 1.39 (1.13–1.67) | 0.001 | |
| Depressive illness | 288 | 1.11 (0.93–1.32) | 0.251 |
| Substance dependence diagnoses | 287 | 2.69 (2.29–3.19) | <0.001 |
| Social health | |||
| Role functioning scale | 284 | -0.02 (-0.27–0.23) | 0.875 |
| SOFAS | 287 | -0.44 (-1.22–0.34) | 0.270 |
| Committed a crime in past month | 283 | 1.74 (1.46–2.10) | <0.001 |
| Drug trafficking | 283 | 1.97 (1.61–2.45) | <0.001 |
| Theft | 283 | 1.16 (0.93–1.44) | 0.177 |
| Any employment in past month | 281 | 0.92 (0.73–1.13) | 0.415 |
| Drug spending in past month | 283 | 1.51 (1.40–1.62) | <0.001 |
| Multimorbidity score (0-12) | 288 | 1.43 (1.26-1.63) | <0.001 |
a Binary logistic regression was used to model the relationship between CHS and mortality, hepatitis C virus exposure, persistent hepatitis C Infection, depression, employment and committing any crime, drug trafficking or theft. Ordinal logistic regression was used to model the relationship between CHS and number of multimorbid illnesses and dependence diagnoses. Multinomial logistic regression was used to model the relationship between CHS and psychotic illness diagnosis. Linear regression was used to model the relationship between CHS and Role Functioning Score, and SOFAS. Quasi-Poisson regression was used to model the relationship between CHS and drug spending.
b For binary, ordinal, and multinomial logistic regression models, adjusted odds ratios (95% CI) were reported for a 1000-unit increase in CHS, adjusting for age and sex. For linear regression models, adjusted effect coefficients (95% CI) for a 1000-unit increase in CHS, adjusting for age and sex. For quasi-Poisson regression models, the adjusted risk ratios (95% CI) were reported for a 1000-unit increase in CHS, adjusting for age and sex.