Predicting Respiratory Tract Infection after Immune Thrombocytopenia Diagnosis

Researchers led by Jinhua Wei developed a nomogram model for individualized predictions of risk for upper respiratory tract infection (URI) within 6 months of a diagnosis of idiopathic thrombocytopenic purpura (ITP). Their process was described in Computational and Mathematical Methods in Medicine.

The researchers collected clinical data from patients diagnosed with ITP with 6 months of follow-up and, using uni- and multivariable logistic regression analyses, identified the following risk factors for URI: advanced age, glucocorticoid use, history of smoking, low platelet count, high serum C-reactive protein (CRP) level, low CD4+ cell count, and high CD8+ T cell count (P<.05).

These markers were used to construct the nomogram model for individualized prediction of the onset of URI symptoms in patients with ITP within 6 months of their diagnosis. The model was then tested in training and validation datasets of 242 and 50 patients with ITP, respectively. The authors noted that 52 (21.49%) patients in the training cohort developed URI, including 24 viral infections, 11 Mycoplasma pneumoniae infections, and 17 bacterial infections.

According to the article, the prediction of the calibration curve was highly consistent with real results based on the model’s internal verification via the bootstrap method. The researchers reported that external validation of the model returned a sensitivity of 0.949, a specificity of 0.727, and an area under the receiver operating characteristic curve of 0.890 (95% confidence interval, 0.757-0.975) for onset of URI symptoms within six months of an ITP diagnosis.

In short, the authors identified risk factors for URI within a half-year period of a diagnosis of ITP and constructed a predictive nomogram model using those factors, which they reported had “good discrimination and prediction accuracy.” They closed with the suggestion that that “[the model] can provide guidance for clinically accurate and personalized prediction of ITP patients.”

Related: Single-Center Perspective of ITP Presentations and Characteristics