Skip to main content

Table 2 Multilevel logistic regression adjusted odds ratios (95% CIs) for patients receiving at least one potentially inappropriate indicator

From: Prescriber variation in potentially inappropriate prescribing in older populations in Ireland

Variable Model 1 random intercept patient level variables Model 2 random intercept patient and GP level variables Model 3 random slope patient and GP level variables
OR* (95% CIs) OR* (95% CIs) OR* (95% CIs)
Patient level fixed effects    
Gender    
Male 1 1 1
Female 0.92 (0.90, 0.93) 0.92 (0.90, 0.93) 0.92 (0.91, 0.93) 1
Age    
70-74 years 1 1 10,(0.95)
≥ 75 years 0.95 (0.93, 0.97) 0.95 (0.93, 0.97) 0.95 (0.93, 0.96)
No of different repeat drug classes    
0 1 1 1
1 2.00 (1.83, 2.18) 2.00 (1.83, 2.18) 1.43(1.43, 1.44) †
2 3.98 (3.69, 4.30) 3.98 (3.68, 4.30)  
3 6.31 (5.87, 6.79) 6.31 (5.86, 6.78)  
4 8.81 (8.20, 9.46) 8.81 (8.20, 9.45)  
5 12.79 (11.92, 13.72) 12.78 (11.91, 13.71)  
6 17.39 (16.21, 18.65) 17.38 (16.20, 18.64)  
7 23.22 (21.65, 24.92) 23.21 (21.63, 24.90)  
8 30.15 (28.09, 32.37) 30.13 (28.07, 32.35)  
9 38.89 (36.19, 41.79) 38.86 (36.17, 41.76)  
≥10 71.77 (67.00, 76.87) 71.71 (66.95, 76.81)  
GP level fixed effects    
Gender §    
Male - 1 1
Female - 0.94 (0 .91, 0.97) 0.94 (0.91, 0.97)
Area of residence    
Urban - 1 1
Rural - 0.98 (0.95, 1.01) 0.98 (0.94, 1.01)
Deprivation    
Deprivation score (centred) - 1.00 (0.99, 1.01) 1.00 (0.99, 1.01)
  1. *OR = odds ratio.
  2. The number of different repeat drug classes was treated as a continuous variable (created 10 dummy variables and the coefficients show an approximately linear increase).
  3. GP level data was unavailable for 108 (5%) GPs with 6,906 (2%) patients.
  4. §GP gender was missing for 22 (1%) GPs with 2,578 (0.76%) patients.