Clinical predictors of hospitalization in an acute attack of bronchial asthma.

ABSTRACT

The present study was undertaken to identify the clinical predictors of hospitalization in an acute attack of bronchial asthma in subjects aged 2-12 years. Seventy five children with an acute attack of bronchial asthma were evaluated. A detailed clinical history and examination was recorded and baseline investigations sent before starting therapy. All subjects were treated with injection adrenaline (two doses) and those who did not respond were hospitalized. Twenty subjects who were hospitalized were compared with 32 cases who were discharged and did not have a relapse on follow up. After multivariate analysis, the factors independently predictive of hospitalization were, pulsus paradoxus (> 10 mm Hg; OR = 1.02), younger age (below 5 years, OR = 0.98) and severe accessory muscle use (OR = 89.6). Presence of any 2 of these 3 clinical predictors has a high sensitivity (90%) and specificity (96%). The investigative variables significant after multivariate analysis were low pH (OR = 0.00) and polymorphonuclear leucocytosis (> 70%) on peripheral smear (OR = 1.12). The sensitivity of this model was similar (90%) but specificity was lower (90.6%). The addition of investigative variables to clinical model did not improve the predictability. It is concluded that it is possible to identify at presentation, children with acute bronchial asthma who require hospitalization and clinical variables are sufficient for this purpose.