Table 3

Attributed variation within multilevel models of practices nested within CCGs

Outcome variable	Covariates	R²			ICC	AIC
Outcome variable	Covariates	Total	Fixed	Random	ICC	AIC
Multilevel linear regression
Standardised 2WWRR	None	0.209	0	0.209	0.209	58 954
Multilevel Poisson regression (primary analysis)
Detection rate	None	0.435	0	0.435	0.421	46 077
Detection rate	a	0.477	0.140	0.338	0.379	45 276
Detection rate	a+b	0.492	0.178	0.314	0.369	44 848
Multilevel poisson regression (secondary analysis)
Detection rate	c	0.496	0.122	0.375	0.413	45 437
Detection rate	c+d	0.492	0.129	0.36	0.403	45 277
Detection rate	c+d+e+f	0.477	0.154	0.323	0.368	44 972

Covariates: a=standardised 2WWRR; b=(standardised 2WWRR) 2; c=specificity; d=(specificity) 2 ;e=proportion patients aged 65+ years and index of multiple deprivation.
AIC, Aikake information criterion; CCG, Clinical Commissioning Group; ICC, intraclass correlation coefficient; 2WWRR, 2-week wait referral rate.