# Table 2 Multivariate model for non-adherence and risk score (N = 426)

Variables Coefficients Odds ratio (95% confidence interval) p value Risk score weights a
Susceptibility 0.58 1.78 (1.32–2.42) <0.01 6
Severity 0.81 2.35 (1.82–3.05) <0.01 9
Benefits 0.12 1.13 (0.96–1.33) 0.15 1
Barrier 0.09 1.10 (1.06–1.13) <0.01 1
Cues to action 0.13 1.14 (1.06–1.13) 0.02 1
Visible bleeding
Absence[presence] 0.90 2.54 (1.33–4.86) <0.01 10
Current concomitant therapy
Thiopurines
Absence [presence] 0.74 2.10 (0.88–5.05) <0.01 8
Number of tablets/day
8 tablets or less [9 tablets or more] 0.60 1.92 (1.12–3.30) 0.02 7
Hospital
Hospital B [A]   0.59 (0.29–1.19) 0.14
Hospital C [A]   1.81 (0.82–4.00) 0.14
1. Information in [brackets] are reference categories. Hosmer-Lemeshow statistics = 9.36 (degree of freedom = 8, p = 0.31).
2. aThe method described by Sullivan et al.  was used to calculate the risk score weight: Step 1: Divide each regression coefficient by the smallest coefficient in the final logistic regression model (in our model, this is barrier). Step 2: Round this quotient to the nearest whole number. For example, to calculate the score weight of susceptibility, its coefficient of 0.58 was divided by the 0.09, which was the less perceived barrier, resulting in a quotient of 6.44. Rounding this quotient to its nearest integer resulted in 6 for the score weight of this variable. Each subject’s overall screening instrument was then calculated by summing the points of all variables. 