Background Over the last 10 years, EBUS-TBNA has become established as the first line nodal staging procedure of choice for lung cancer patients. However, the pathway for patients following a negative EBUS-TBNA has not been clearly defined.

Aims and objectives The primary aim of this study was to develop and validate a risk stratification model to categorise lymph nodes deemed negative by EBUS-TBNA into ‘low risk’ and ‘high risk’ groups, where ‘risk’ refers to the risk of false negative sampling.

Materials and methods A retrospective analysis of a prospectively maintained database at a UK tertiary EBUS-TBNA centre. Only patients with primary lung cancer and only negative lymph nodes by EBUS-TBNA were included in the analysis. A risk stratification model was built from a derivation set using independent predictors of malignancy and the validation set used to evaluate the constructed model. The study period was March 2010 to August 2013.

Results 329 lymph nodes were included in the analysis (derivation set n = 196, validation set n = 133). Lymph node SUV, the SUV ratio between the lymph node and primary tumour and heterogeneous echogenicity during sonographic assessment were the only independent predictors of malignancy. Using a simplified scoring system based on the natural logs of the odds ratios from the multivariable analysis on the derivation sample, lymph nodes can be stratified into ‘low risk’ (score ≤1) and ‘high risk’ (score ≥2). 141/142 and 94/96 lymph nodes classified as ‘low risk’ in the derivation and validation set respectively were ultimately proven to be benign and 35/54 and 24/37 lymph nodes classified as ‘high risk’ were proven malignant. The negative predictive value of the risk stratification model for the derivation set and validation set was 99.3% (95% CI 96.1–99.6) and 97.9% (95% CI 92–99.6%) respectively.

Discussion This risk stratification model may assist lung cancer MDTs in deciding which patients need further staging procedures and which may proceed directly to treatment after a negative EBUS.