Abstract
Background:
Identification of nerve root involvement (NRI) in patients with low back-related leg pain (LBLP) can be challenging. Diagnostic models have mainly been developed in secondary care with conflicting reference standards and predictor selection. This study aims to ascertain which cluster of items from clinical assessment best identify NRI in primary care consulters with LBLP
Methods:
Cross-sectional data on 395 LBLP consulters were analysed. Potential NRI indicators were seven clinical assessment items. Two definitions of NRI formed the reference standards: (i) high confidence (≥80%) NRI clinical diagnosis (ii) high confidence (≥80%) NRI clinical diagnosis with confirmatory magnetic resonance imaging (MRI) findings. Multivariable logistic regression models were constructed and compared for both reference standards. Model performances were summarised using the Hosmer-Lemeshow statistic and area under the curve (AUC). Bootstrapping assessed internal validity.
Results:
NRI clinical diagnosis model retained five items. The model with MRI in the reference standard retained six items. Four items remained in both models: below knee pain, leg pain worse than back pain, positive neural tension tests, neurological deficit (myotome, reflex or sensory). NRI clinical diagnosis model was well calibrated (p=0.17) and discrimination was AUC 0.96 (95%CI: 0.93, 0.98). Performance measures for clinical diagnosis plus confirmatory MRI model showed good discrimination (AUC 0.83, 95% CI: 0.78, 0.86) but poor calibration (p=0.01). Bootstrapping revealed minimal overfitting in both models.
Conclusion:
A cluster of items identified NRI in LBLP consulters. These criteria could be used clinically and in research to improve accuracy of identification and homogeneity of this subgroup of low back pain patients.
