Table 3

Mixed-effects  models for change in glycaemic control (HbA1c) during the first 6 months postdiagnosis adjusted for pH levels at diagnosis, in children with type 1 diabetes

CovariatesModel
1: adjusted for sex, age at diagnosis,
pH at diagnosis and ethnicity
 (Mean difference in HbA1c, mmol/mol, (95% CI))
Model 2: additionally adjusted for SES
and clinic attended
 (Mean difference in HbA1c, mmol/mol, (95% CI))
Fixed effects
Constant/intercept202.1 (134.2, 270.9)203.7 (136.4, 271.1)
Duration
  Linear−20.6 (−22.5,–18.8)−20.8 (−22.6,–18.9)
  Quadratic2.5 (2.3, 2.8)2.6 (2.3, 2.8)
Sex 
 MaleRefRef
  Female−0.5 (−3.5, 2.5)−0.4 (−3.3, 2.6)
Age at diagnosis0.1 (−0.2, 0.5)0.1 (−0.3, 0.4)
pH at diagnosis−15.2 (−24.7, −5.7)−15.7 (−25.1, −6.3)
Ethnicity
  WhiteRefRef
  Mixed ethnicity7.1 (1.7, 12.6)6.5 (1.1, 11.9)
  Black4.2 (0.2, 8.8)4.6 (0.9, 9.1)
  African-Somali4.7 (−0.9, 10.3)5.2 (−0.4, 10.9)
  Bangladeshi10.1 (4.9, 15.3)10.6 (5.4, 15.7)
  Asian-Other7.5 (3.4, 11.5)5.3 (1.8, 9.4)
Socioeconomic status
 Quartile 1 (poorest)
 Quartile 22.2 (−1.7, 6.1)
 Quartile 31.9 (−2.1, 5.9)
 Quartile 4 (richest)2.7 (−1.9, 7.3)
Pediatric diabetes clinic
 Clinic 1Ref
 Clinic 21.9 (−2.5, 6.3)
 Clinic 36.1 (2.4, 9.6)
Goodness of fit
 Aikake information criterion (AIC)6434.6446430.969
 Bayesian information criterion (BIC)6504.1846523.688