[Validity of the Keith Edwards scoring system for the diagnosis of childhood pulmonary tuberculosis]

The diagnosis of childhood pulmonary tuberculosis remains a challenge. The effectiveness of the Keith Edwards scoring system used to diagnose tuberculosis is controversial. We evaluated this scoring system in the present study.

A prospective randomized study was conducted between March 2008 and December 2011 at the Kimpese General Evangelical Hospital in the Democratic Republic of Congo. The results of the Keith Edwards score were considered for the 161 children (100 pulmonary tuberculosis and 61 case controls) who were enrolled in the study. The association between different parameters and the score and between these parameters and pulmonary tuberculosis were statistically analyzed using univariate and multivariate tests.

Eighty-five (85%) out of the 100 children diagnosed as having pulmonary tuberculosis and 20 (32.8%) of the case controls had a positive score. The age of the patient, duration of the disease, nutritional status, tuberculosis contact, positive tuberculin skin test, and lymph node enlargement showed a significant statistical association with the score and pulmonary tuberculosis (P<0.05). The score's sensitivity and specificity were 85% and 67.2%, respectively. The positive predictive value and negative predictive value were found to be 80.9% and 73.2%, respectively. The positive likelihood ratio was 2.57, the negative likelihood ratio was 0.22, and overall agreement was 76.1%.

The Keith Edwards score could be good tool for public health purposes, but it might be less effective for individual diagnosis of childhood pulmonary tuberculosis because of low specificity. Further studies are required to evaluate and validate the diagnostic value of clinical and radiological symptoms in childhood pulmonary tuberculosis.