Background: Diagnosis of childhood tuberculosis (TB) remains a challenge, especially in high human immunodeficiency virus (HIV) prevalence areas.
Methods: Retrospective study of TB cases registered at a pediatric hospital over a 1-year period in Kinshasa, Democratic Republic of Congo. Data were used to calculate scores for eight diagnostic scoring systems. Correlations between scores, agreement among scoring systems on which children are in need of treatment, and clinical presentation by HIV infection status were assessed using Spearman rank correlations, kappa statistics and bivariate analysis.
Results: The 42 HIV-infected children were more likely to be older, exposed to TB, have a history of TB, and present with lymphadenopathy and malnutrition, compared to the 45 non-HIV-infected children. Correlations of scores between scales unrelated in their development and agreement among scales on decision to treat were moderate to poor. One in seven children would not have received treatment according to at least one scale.
Conclusion: The clinical presentation of TB in HIV-infected and non-infected children was quite similar, but HIV-infected children were more likely to have a prior history of TB. Correlation between clinical scoring systems was poor, with some disagreement on the decision of whom to treat, underscoring the need for improved childhood TB diagnostics.