Estimating the Risk of Chronic Pain: Development and Validation of a Prognostic Model (PICKUP) for Patients with Acute Low Back Pain

doi:10.1371/journal.pmed.1002019

PMC full text:

PLoS Med. 2016 May; 13(5): e1002019.

Published online 2016 May 17. doi: 10.1371/journal.pmed.1002019

Copyright/License ►Request permission to reuse: Copyright © 2016 Traeger et al
This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Table 5

Summary performance measures in the external validation sample.

Aspect	Measure	PICKUP (n = 1,528)	Model 2a (n = 1,525)	Model 2b (n = 1,504)
Overall Performance	R² (Nagelkerke)	7.7	4.8	10.1
Discrimination	AUC	0.66 (0.63 to 0.69)	0.64 (0.60 to 0.68)	0.69 (0.64 to 0.72)
Calibration	Calibration intercept	−0.55	−0.81	−0.86
	Calibration slope	0.89	0.74	0.99
Decision curve analysis	Net benefit at incidence rate cutoff^a	0.04	0.06	0.04
	Net number of unnecessary interventions avoided at 30% risk cutoff^b	46	54	52

^aThe proportion of patients with poor outcomes who would correctly be recommended further intervention at the same rate of not recommending intervention for patients with good outcomes, when the threshold probability is set at the incidence rate in the external validation sample (i.e., 19% for PICKUP, 10% for Model 2a, 14% for Model 2b).

^bNet number of unnecessary interventions avoided per 100 acute LBP patients without missing any patients who developed chronic LBP, if only patients with predicted risks higher than the cutoff are recommended further intervention.