Prognostic factors in childhood asthma: a logistic regression analysis

Thirty-two factors related to childhood asthma were studied in 200 asthmatic children. Remission of asthma was defined as a period of at least 2 years free of asthma while receiving no treatment. Univariate analysis showed significant associations between persistence of asthma and perennial symptoms with an odds ratio (OR) of 2.5 (95% confidence intervals 1.2 to 5.5); sensitization to house dust mites OR 3.5 (1.2 to 9.6); sensitization to molds, OR 7.9 (2.9 to 21.6); sensitization to pollen, OR 4.8 (1.4 to 16.3); and sensitization to milk protein, OR 5.4 (1.8 to 15.9). There was a positive association of remission of asthma with good treatment compliance, OR 12.1 (1.6 to 91.6). A stepwise logistic regression analysis selected the variables: perennial symptoms (S: 0 = no, 1 = yes), length of follow-up time (T: number of months), treatment compliance (C: 0 = poor, 1 = good), sensitization to fungi (F: 0 to 4), and sensitization to milk protein (M: 0 to 4). The analysis yielded the following formula for calculation of probability of asthma remission P(R): [formula: see text] where S indicates perennial symptoms; T, follow-up time in months; C, compliance; F, allergy to fungi; and M, allergy to milk.