Table 3.
Univariate and multivariate Markov models of factors associated with transition between alcohol use and non-use states
| Unadjusted Models | Fully Adjusted Model | |||||||
|---|---|---|---|---|---|---|---|---|
| Direction of Change of Alcohol Consumption: | Abstinence to Use | Use to Abstinence | Abstinence to Use | Use to Abstinence | ||||
| RR | 95% CI | RR | 95% CI | RR | 95% CI | RR | 95% CI | |
| Age | 1.00 | 0.92, 1.08 | 0.99 | 0.94, 1.04 | - | - | ||
| White ethnicity | 1.76 | 0.89, 3.49 | 1.44 | 0.94, 2.19 | - | - | ||
| Male vs. Female | 0.63 | 0.30, 1.33 | 0.92 | 0.58, 1.45 | - | - | ||
| Intervention vs. control | 1.14 | 0.60, 2.16 | 1.06 | 0.71, 1.58 | - | - | ||
| Perceived homelessness | 1.36 | 0.72, 2.56 | 0.73 | 0.48, 1.10 | - | - | ||
| Actual homelessness | 0.74 | 0.38, 1.44 | 0.51 | 0.33, 0.79 | 0.68 | 0.31, 1.49 | 0.47 | 0.28, 0.78 |
| Incarcerated | 2.50 | 1.26, 4.94 | 0.92 | 0.61, 1.39 | 1.65 | 0.75, 3.60 | 0.90 | 0.55, 1.47 |
| Injection drug use last 30 days | 0.74 | 0.36, 1.51 | 1.10 | 0.64, 1.91 | - | - | ||
| Injection drug use last 3 months | 1.11 | 0.41, 3.01 | 1.05 | 0.46, 2.40 | - | - | ||
| Thought about taking own life | 1.72 | 0.84, 3.55 | 0.75 | 0.46, 1.21 | - | - | ||
| Smoked marijuana | 3.03 | 1.49, 6.17 | 0.94 | 0.60, 1.49 | 3.11 | 1.33, 7.23 | 1.05 | 0.62, 1.80 |
| Snorted/smoked cocaine | 1.22 | 0.62, 2.40 | 0.69 | 0.43, 1.09 | - | - | ||
| Snorted/swallowed/smoked heroin | 1.01 | 0.50, 2.05 | 0.73 | 0.44, 1.23 | - | - | ||
| Used any non-injection drugs | 0.87 | 0.36, 2.06 | 0.75 | 0.33, 1.68 | - | - | ||
| Told HCV caused cirrhosis at diagnosis | 1.16 | 0.47, 2.84 | 0.67 | 0.37, 1.19 | - | - | ||
| Diagnosed with HCV in research study | 1.03 | 0.54, 2.00 | 1.26 | 0.83, 1.91 | - | - | ||
| Told to stop drinking at diagnosis | 0.89 | 0.43, 1.88 | 0.53 | 0.32, 0.88 | 0.79 | 0.31, 2.00 | 0.60 | 0.32, 1.12 |
| Had a regular physician at diagnosis | 0.83 | 0.43, 1.58 | 0.79 | 0.52, 1.20 | - | - | ||
| Physician diagnosed liver disease | 1.98 | 0.87, 4.51 | 1.77 | 1.03, 3.03 | 1.60 | 0.58, 4.41 | 1.88 | 1.00, 3.54 |
| Recent alcohol treatment | 0.56 | 0.19, 1.63 | 0.57 | 0.32, 1.00 | 1.11 | 0.38, 3.28 | 0.69 | 0.33, 1.42 |
| In drug treatment at the time of interview | 0.91 | 0.43, 1.90 | 1.11 | 0.71, 1.75 | - | - | ||
Markov models estimate the likelihood of transition between alcohol abstinence and use from baseline to the first follow-up or the first to the second followup, as compared to remaining in a current state, allowing transition to occur in either direction; relative risks less than one indicate a lower likelihood of transition from one state to the next, conversely relative risks greater than one indicate a greater likelihood of change. Two relative risks are reported for every variable fitted because we have not assumed that the rate of changing from abstinence to alcohol use is the same as changing from alcohol use to abstinence.