Table 1.

Selection criteria for clustering of diseases.

Clinical inclusion criteria		Eligible hospital episodes
Prior disease	Diseases were the first primary hospital diagnosis in each ICD-10 chapter	425,383 male
		502,771 female
+ Clinical considerations	Clinically distinct, age-related disease, or R-coded diseases of unknown aetiology	400,006 male
		468,398 female
Statistical inclusion criteria		Eligible diseases
Successful fit	At least 50 cases, and a covariance matrix with no unusually large or small eigen values that exceed the mean by 2.5 standard deviations when outliers were included	343 male
		346 female
+ Statistically significant	Statistically significant risk factors at the 0.05 level,after a multiple-testing Bonferroni adjusted multivariate ${\chi }^2$ test	150 male
		140 female
+ Proportional hazards	Test of proportional hazards assumption—no statistically significant deviation at the 0.05 level, after a multiple-testing FDR adjustment	138 male
		127 female
+ Unisex	The set must include the same disease in both men and women	86 male and female