Table 1.

Agreement between ethnic categorisation in CPRD and HES datasets

			Algorithm-generated higher-level ethnic categorisation using all HES datasets
			White	Mixed	Asian	Black	Other	Total
Algorithm-generated higher-level ethnic categorisation using CPRD GOLD or CPRD Aurum	White	N	10,641,573	96,569	59,201	51,769	99,475	10,948,587
		Row %	97.20	0.88	0.54	0.47	0.91	100.00
		Column %	97.77	36.38	5.81	7.46	47.88	83.77
	Mixed	N	84,799	88,143	33,154	53,501	16,333	275,930
		Row %	30.73	31.94	12.02	19.39	5.92	100.00
		Column %	0.78	33.20	3.26	7.71	7.86	2.11
	Asian	N	56,358	30,064	875,090	14,705	52,042	1,028,259
		Row %	5.48	2.92	85.10	1.43	5.06	100.00
		Column %	0.52	11.33	85.92	2.12	25.05	7.87
	Black	N	33,396	38,521	17,281	564,302	17,448	670,948
		Row %	4.98	5.74	2.58	84.11	2.60	100.00
		Column %	0.31	14.51	1.70	81.36	8.40	5.13
	Other	N	67,683	12,162	33,741	9324	22,451	145,361
		Row %	46.56	8.37	23.21	6.41	15.44	100.00
		Column %	0.62	4.58	3.31	1.34	10.81	1.11
	Total	N	10,883,809	265,459	1,018,467	693,601	207,749	13,069,085
		Row %	83.28	2.03	7.79	5.31	1.59
		Column %	100.00	100.00	100.00	100.00	100.00

Counts (N) and proportions (%) of currently registered acceptable English patients with ethnicity recorded in combined CPRD GOLD and CPRD Aurum and in any HES dataset showing the agreement between the algorithm-generated higher-level ethnic categorisation using CPRD data only with the algorithm-generated higher-level ethnicity categorisation using HES data only