Falls and fracture risk screening in primary care: update and validation of a postal screening tool for community dwelling older adults recruited to UK Prevention of Falls Injury Trial (PreFIT)

Table 2 Prediction of falls and fractures over 12 months by development and validation datasets

	N	AUC	95% CI	P value
Any fall
Model 1, control, screener items only	2,661	0.66	0.64, 0.68	< 0.001
Model 2, intervention, postal screener only	4,841	0.67	0.65, 0.68	< 0.001
Model 3, control, development, full model*	2,329	0.70	0.68, 0.72	< 0.001
Model 4, intervention, full model**	4,240	0.71	0.69, 0.72	< 0.001
Recurrent falls
Model 1, control, screener items only	2,574	0.70	0.68, 0.72	< 0.001
Model 2 intervention, postal screener only	4,712	0.71	0.69, 0.72	< 0.001
Model 3, control, development, full model	2,259	0.75	0.73, 0.78	< 0.001
Model 4, intervention, validation, full model	4,133	0.76	0.75, 0.78	< 0.001
Any fracture
Model 1, control, screener items only	3,221	0.60	0.54, 0.65	< 0.001
Model 2, intervention, postal screener only	5,779	0.60	0.56, 0.64	< 0.001
Model 3, control, development, full model	3,078	0.73	0.67, 0.79	< 0.001
Model 4, intervention, validation, full model	5,503	0.71	0.67, 0.74	< 0.001

^*Models 3 = development of expanded variable set, using control arm participant dataset, screener items from baseline questionnaire
^**Models 4 = validation of expanded variable set using treatment arm dataset receiving postal screener

ISSN: 1471-2318