Incidence rate ratio of hospital admissions between individuals with type I diabetes compared to normal population controls at different levels of exposures of interest while adjusting for significant variables explored in table 4
| Parameter estimates | Variances | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Category | IRR | (SE) | (95% CI) | p≤ | ESS | Level | Variance (SE) | (95% CI) | ESS | ||
| DIC (22299) | |||||||||||
| Reference (SES 1, small, <1990, <5) | 5.888 | (1.395) | (4.118 to 8.568) | 0.0001 | 371 | 1 | – | – | – | – | |
| SES | 2 | 1.132 | (1.017) | (0.870 to 1.473) | 0.1749 | 1441 | 2 | 1.272 | (0.047) | (1.183 to 1.366) | 3347 |
| 3 | 1.266 | (1.032) | (0.973 to 1.643) | 0.0404 | 1525 | 3 | 0.048 | (0.022) | (0.008 to 0.095) | 335 | |
| 4 | 1.096 | (1.012) | (0.854 to 1.417) | 0.2372 | 1375 | 4 | 0.013 | (2.836) | (0.003 to 10.798) | 147 | |
| 5 | 1.391 | (1.043) | (1.082 to 1.774) | 0.0045 | 1344 | ||||||
| Centre size | Large | 0.839 | (0.985) | (0.709 to 0.990) | 0.0189 | 3974 | |||||
| Date of birth | 90 to 94 | 0.926 | (0.991) | (0.727 to 1.166) | 0.2639 | 951 | |||||
| 95 to 99 | 1.208 | (1.027) | (0.917 to 1.577) | 0.0905 | 651 | ||||||
| 00 to 04 | 1.288 | (1.047) | (0.899 to 1.821) | 0.0825 | 651 | ||||||
| >05 | 1.135 | (1.035) | (0.663 to 1.931) | 0.3215 | 1651 | ||||||
| Age at diagnosis (years) | Age | 0.846 | (0.989) | (0.744 to 0.965) | 0.0061 | 664 | |||||
Results are from a multivariable multilevel models estimated using MCMC. Results are reported using the mean of the posterior distribution to indicate IRR, the SD of the posterior chain is used to indicate the parameter SE, 95% posterior probability intervals (95% CI) represent the 2.5th and 97.5th centiles of the posterior distribution, and directional posterior probabilities (p≤). ESS indicates the effectiveness of MCMC chain mixing. Bayesian DIC is used to indicate model fit. The reference category comparing the IRR between cases and controls in the highest social class, small centres, born prior to 1990 and with an age at diagnosis less than 5 years is indicated, and the IRR of the exposure of interest is represented by an interaction with case–control status. The level 2 variance indicates the overdispersion parameter, the level 3 variance indicates the matched set, and the level 4 variance indicates the county at diagnosis. Age is modelled as a linear change per 5 year increase in the age at diagnosis.
DIC, Deviance Information Criterion; IRR, incidence rate ratio; SES, socioeconomic status.