Table 5 Results of the modelling of practice detection rates with both specified between-practice variation and random case-mix variation

				‘Poorly performing practices’
Model parameters ^a	Observed data (detection rate) ^b			Correctly identified ^c		Total identified ^d
Cases	s.d. ^e	Min	Max	Median	IQR	Median	IQR
Low variation (s.d.=0.025)
25	0.1	0.16	0.8	18	14 – 19	87	80 – 93
50	0.07	0.24	0.72	19	16 – 21	77	70 – 82
75	0.06	0.27	0.69	24	21 – 27	87	79 – 92
100	0.06	0.29	0.66	28	24 – 29	87	78 – 94
150	0.05	0.32	0.64	30	27 – 34	89	84 – 95
200	0.04	0.34	0.63	35	33 – 38	91	86 – 96
Moderate variation (s.d.=0.05)
25	0.11	0.12	0.84	20	17 – 24	58	53 – 64
50	0.09	0.2	0.76	31	28 – 35	77	69 – 89
75	0.08	0.23	0.73	39	36 – 43	85	78 – 94
100	0.07	0.25	0.71	44	40 – 47	89	83 – 95
150	0.06	0.27	0.69	52	48 – 55	91	88 – 95
200	0.06	0.28	0.68	57	54 – 59	94	89 – 97
High variation (s.d.=0.075)
25	0.12	0.08	0.88	32	29 – 36	75	70 – 83
50	0.1	0.14	0.82	44	40 – 48	81	75 – 89
75	0.09	0.16	0.79	52	49 – 55	90	83 – 95
100	0.09	0.18	0.77	57	54 – 60	91	86 – 97
150	0.09	0.2	0.76	64	62 – 67	94	90 – 97
200	0.08	0.21	0.75	68	66 – 71	95	92 – 98

Abbreviations: IQR=interquartile range; Max=maximum; Min=minimum.
^aFor each set of parameters, 200 sets of 1000 practices were modelled.
^bValues for s.d., minimum and maximum represent median values for all runs of the simulation at each specification of practice variation and number of cancer cases.
^cCorrectly identified poorly performing practices represents the median number (with interquartile range) of practices, which were in the lowest decile of detection rate before adjusting for case-mix (N=100) and which were also in the lowest decile of detection rate after introducing random case-mix variation.
^dTotal identified poorly performing practices represents the median number (with interquartile range) of practices, which were in the lowest decile of detection rate after introducing random case-mix variation.
^eStandard deviation.

Search