Multilevel negative binomial regression coefficients of the total number of CP births on total live births by singletons and multiples
| Model parameters | Singletons | 95% CI | Multiples | 95% CI |
| Intercept | 0.002 | 0.001 to 0.003 | 0.010 | 0.001 to 0.013 |
| Yearly rate of change | 0.976 | 0.952 to 1.001 | 0.972 | 0.955 to 0.990 |
| Overdispersion: Ln(Alpha) | −5.691 | −7.332 to −4.049 | −3.008 | −4.039 to −1.977 |
| Between-centre variation | ||||
| Initial status | 0.239 | 0.092 to 0.621 | 0.042 | 0.006 to 0.288 |
| Rate of change | 0.002 | 0.001 to 0.004 | NA | |
| Covariance | −0.017 | −0.035 to 0.001 | NA | |
| Model fit | Wald χ2(1)=3.49, p=0.062 | Wald χ2(1)=8.94, p=0.003. | ||
Parameters in bold are significant at p<0.005. Comparisons of models using the likelihood ratio test indicated that estimating centre-level variance around the rate of change and the covariance between intercept and rate of change of Multiples did not improve model fit (at p<0.005). Therefore, these terms were not modelled.