Table 3:

Logistic regression models for potentially preventable unplanned admissions to hospital

CovariateUnadjusted OR (95% CI)p value*Adjusted OR (95% CI)p value*
Male sex0.96 (0.88–1.05)0.41.26 (1.15–1.38)< 0.001
Age, yr< 0.001< 0.001
 20–340.77 (0.63–0.95)1.34 (1.09–1.67)
 35–440.77 (0.62–0.95)1.07 (0.87–1.33)
 45–541.00 (ref)1.00 (ref)
 55–641.81 (1.51–2.16)1.22 (1.01–1.46)
 65–743.44 (2.90–4.08)1.56 (1.30–1.87)
 ≥ 757.28 (6.22–8.53)2.39 (2.01–2.84)
Deprivation quintile< 0.001< 0.001
 1 (least deprived)1.00 (ref)1.00 (ref)
 21.59 (1.32–1.90)1.33 (1.10–1.61)
 31.96 (1.66–2.32)1.62 (1.35–1.94)
 42.21 (1.87–2.63)1.63 (1.34–1.97)
 5 (most deprived)2.84 (2.40–3.37)1.98 (1.63–2.41)
No. of physical conditions< 0.001< 0.001
 01.00 (ref)1.00 (ref)
 12.68 (2.22–3.22)2.50 (2.07–3.03)
 25.94 (4.97–7.10)4.93 (4.06–5.99)
 39.48 (7.91–11.37)6.82 (5.55–8.37)
 ≥ 423.93 (20.49–27.93)14.38 (11.87–17.43)
Mental health condition2.65 (2.43–2.90)< 0.0011.80 (1.64–1.97)< 0.001
  • Note: CI = confidence inteval, OR = odds ratio, ref = reference.

  • * p values based on joint tests, which test the overall differences between the individual categories of a given variable.

  • Adjusted for other variables listed in the table. The adjusted model includes a random effect for practice. Intraclass correlation coefficient for practice variation, 0.019. OR for 95% midrange of practice variation (i.e., 2.5 and 97.5 percentiles, compared with average practice), 0.61–1.63.