Derivation and validation of a clinical system for predicting pneumonia in acute stroke

We derived and validated a clinical prediction rule that can be used to predict post-stroke pneumonia.

We conducted a retrospective cohort study of patients admitted to hospital with a stroke. The cohort was subdivided into a derivation group and a validation group. Within the derivation group, a point scoring system was developed to predict pneumonia based on a logistic regression model. The point scoring system was then tested within the validation group.

Of the 1,363 patients with stroke, 10.5% of patients experienced new pneumonia. The most points were assigned for abnormal swallowing result and history of pneumonia (4 points), followed by greater NIHSS score (3 points), patient being 'found down' at symptom onset (3 points), and age >70 years (2 points). A 3-level classification system was created denoting low, medium and high risks of pneumonia, which accurately predicted pneumonia in the validation group. The discriminatory accuracy of the 3-level clinical prediction rule exceeded the acceptable range in both the derivation group (c statistic: 0.78) and validation group (c statistic: 0.76).

A simple scoring system was derived and validated. This clinical scoring system may better identify stroke patients who are at high risk of developing new pneumonia.