Page 105 - ADULT-ONSET ASTHMA PREDICTORS OF CLINICAL COURSE AND SEVERITY

CLINICAL PREDICTORS OF REMISSION AND PERSISTENCE OF ADULT-ONSET ASTHMA
measured by body plethysmography. Bronchial challenge test was performed with inhaled methacholine to establish the concentra on causing a 20% fall in FEV1 (PC20-methacholine). In case pa ents did not reach a ≥20% fall in FEV1, a level of 32 mg/ml methacholine was taken as default value. Bronchial hyperresponsiveness can be divided into: mild 1-4mg/ml, moderate to severe <1.0 mg/ml.21
In ammatory parameters – Frac on of exhaled nitric oxide (FeNO) was measured at a ow rate of 50mL/s (NIOX System, Aerocrine, Sweden).22 Venous blood was collected and di eren al white blood cell count was performed. Total and speci c IgE to common aeroallergens were measured by ImmunoCAP; atopy was de ned as IgE >0.35 Ku/L for at least one allergen.
Sputum induc on was performed according to interna onal standards.23 Sputum processing was done according to full sample method and di eren al cell counts were stained and analyzed on cytospin prepara ons.
Sinonasal imaging – The presence of nasal polyps was evaluated based on sinus CT-scanning and nasal endoscopy.24
ASSESSMENT OF ASTHMA REMISSION AND PERSISTENCE
Clinical asthma remission8-10, 25 was the primary outcome of the study and de ned as: no asthma symptoms for ≥1 year and no asthma medica on use for ≥1 year at the 5-year follow- up visit. Asthma persistence was de ned as presence of asthma symptoms in the last year or use of any asthma medica on (beta-2-agonists or inhaled cor costeroids) in the last year. Secondary outcomes included pathophysiological con rma on of clinical remission by means of change in lung func on, bronchial hyperresponsiveness and in ammatory markers.9
STATISTICAL ANALYSIS
Comparisons of baseline variables between pa ents in remission and pa ents with persistent asthma were done by Student t-test, Mann Withney-U test or chi-square. Wilcoxon signed rank test and paired t-tests were used to analyze the changes of variables over me. Univariate logis c regression was used to select signi cant baseline variables for the mul variable logis c regression model. The signi cant predictor variables were used in a mul variable binary logis c regression model and selected by backward selec on of the model. Results are expressed as beta with standard error (SE). Sta s cal signi cance was set at a P value of less than .05.
Analyses were performed in SPSS version 23.0 (IBM SPSS, Chicago, Ill) and R-studio V0.99.467, Package logis (Integrated development environment for R, Boston, MA).
103