166 Garrett et al, J ALLERGY CLIN IMMUNOL AUGUST 1997 FEV 1 (ml) 500 - 400 300 200 100 Ii Ii '7 i ! i i i t I i 15 30 45 60 75 90 Time after start of nebulization (m<strong>in</strong>utes) * p
J ALLERGY CLIN IMMUNOL Garrett et alo 167 VOLUME 100, NUMBER 2 TABLE II. ANCOVA results Covariates Asthma medications <strong>in</strong> previous 6 hours ANCOVA IB + S Basel<strong>in</strong>e Duration Inhaled Inhaled Oral Inhaled Oral model no, vs S FEV 1 Age of attack 13-agonist IB <strong>the</strong>ophyll<strong>in</strong>e steroid steroid 1 0.02 0.0007 . . . . . . . 2 0.02 -- 0.05 . . . . . . 3 0.04 -- -- 0.02 . . . . . 4 0.08 -- -- -- 0.0001 . . . . 5 0.03 . . . . 0.01 -- -- -- 6 0.03 . . . . . 0.2 -- -- 7 0.02 . . . . . . 0.8 -- 8 0.03 . . . . . . . O. I3 9 0.21 0.01 0.08 0.02 0.0001 0.12 0.52 0.12 0.9 10 0.16 0.01 0.07 0.02 0.0001 . . . . Probability (p) values are adjusted for prior asthma medication <strong>and</strong> basel<strong>in</strong>e measurements on AFEv I 90 from add<strong>in</strong>g <strong>ipratropium</strong> to 2.5 mg of salbutamoI. The analysis identified basel<strong>in</strong>e FEV~, age, [3-agonist use <strong>with</strong><strong>in</strong> 6 hours, <strong>and</strong> duration of attack as hav<strong>in</strong>g a significant impact on response to <strong>the</strong>rapy. Of <strong>the</strong>se, only [3-agonist use <strong>with</strong><strong>in</strong> 6 hours had any effect on <strong>the</strong> benefit of add<strong>in</strong>g IB to S (reduced top = 0.08). IB, Ipratropium <strong>bromide</strong>; S, salbutamoI. on outcome (AFEV!), <strong>and</strong> on <strong>the</strong> differential effect of Combivent versus <strong>salbutamol</strong> on AFEV~ 90. Fur<strong>the</strong>r analyses were <strong>the</strong>n undertaken to assess <strong>the</strong> effects of: (1) serum <strong>the</strong>ophyll<strong>in</strong>e <strong>and</strong> serum <strong>salbutamol</strong> levels <strong>in</strong> patients tak<strong>in</strong>g <strong>the</strong>se medications; (2) an FEVf < 1 L <strong>and</strong> FEVI >_ 1 L on attendance; <strong>and</strong> (3) <strong>in</strong>haled [3-agonist use <strong>in</strong> <strong>the</strong> previous 6 hours def<strong>in</strong>ed as 0, 1 to 9 puffs, <strong>and</strong> 10 puffs or more (1 nebulized 13-agonist was considered equivalent to 20 puffs through a metered-dose <strong>in</strong>haler) on <strong>the</strong> differential effect of Combivent <strong>and</strong> <strong>salbutamol</strong> on AFEV~ 90. Paired Student's t tests were used when appropriate to compare differences between means. Data were analyzed by us<strong>in</strong>g SAS software (version 6.08, SAS Institute). RESULTS A total of 338 of 442 patmnts screened entered <strong>the</strong> study over a 542-day period. One center enrolled 231 patients, <strong>and</strong> <strong>the</strong> o<strong>the</strong>r enrolled 107 patients. The most common reasons for exclusion of <strong>the</strong> 104 patients were an FEV~ greater than 70% of predicted value (73%) <strong>and</strong> <strong>in</strong>ability to satisfactorily perform spirometry (19%). Fifty-n<strong>in</strong>e of 338 patients recruited <strong>in</strong>to <strong>the</strong> study were <strong>with</strong>drawn before <strong>the</strong> primary outcome measurement of FEV~ at 90 m<strong>in</strong>utes (AFEV~ 90) was obta<strong>in</strong>ed; 13 requested early <strong>with</strong>drawal, (9 receiv<strong>in</strong>g Combivent <strong>and</strong> 4 receiv<strong>in</strong>g <strong>salbutamol</strong>), 45 were <strong>with</strong>drawn early by <strong>the</strong> ED doctor because of a lack of satisfactory improvement (18 receiv<strong>in</strong>g Combivent <strong>and</strong> 27 receiv<strong>in</strong>g <strong>salbutamol</strong>), <strong>and</strong> one was <strong>with</strong>drawn before treatment was adm<strong>in</strong>istered because he was unable to provide blood samples. The mean age (2 SEM) of <strong>the</strong> 338 patients was 29.5 + 0.7 years; 60.7% were female, 55.9% European, 17.2% Maori, 24.3% Pacific Isl<strong>and</strong>, <strong>and</strong> 2.6% Asian. There was no significant difference between <strong>the</strong> treatment groups for any of <strong>the</strong> demographic or cl<strong>in</strong>ical variables measured (Table I). FEV~ at basel<strong>in</strong>e revealed a difference of 40 + 57 ml <strong>in</strong> favor of Combivent (p = 0.4485). Analysis of change <strong>in</strong> FEV t from basel<strong>in</strong>e, as well as <strong>in</strong>corporation of basel<strong>in</strong>e FEV l as a covariate <strong>in</strong> <strong>the</strong> statistical model, ensured that <strong>the</strong>re was no effect on <strong>the</strong> primary efficacy end po<strong>in</strong>t from this small difference. The mean absolute difference <strong>in</strong> change <strong>in</strong> FEV~ at 45 m<strong>in</strong>utes was 93 + 24 ml (p = 0.03) <strong>and</strong> at 90 m<strong>in</strong>utes 113 _+ 18 ml (p = 0.02) <strong>in</strong> favor of <strong>the</strong> Combivent group when compared <strong>with</strong> <strong>the</strong> <strong>salbutamol</strong> group (Fig. 1). Most of <strong>the</strong> improvement <strong>in</strong> ei<strong>the</strong>r treatment group occurred by 45 m<strong>in</strong>utes <strong>with</strong> m<strong>in</strong>imal additional bronchodilation observed at 90 m<strong>in</strong>utes (37 ml Combivent, 17 ml <strong>salbutamol</strong>). Eighty percent of patients reported us<strong>in</strong>g medications <strong>with</strong><strong>in</strong> 6 hours of presentation to <strong>the</strong> ED. Thirteen percent reported tak<strong>in</strong>g oral corticosteroids, 31.7% <strong>in</strong>haled corticosteroids (_+ cromolyn 2 ketotifen), 79.9% <strong>in</strong>haled 13-agonist, 10.1% oral <strong>the</strong>ophyll<strong>in</strong>e, <strong>and</strong> 7.4% <strong>in</strong>haled antichol<strong>in</strong>ergics. Independent of <strong>the</strong> study drug received, a poor response to treatment was predicted by frequent use of <strong>in</strong>haled [3-agonist <strong>in</strong> <strong>the</strong> 6 hours before presentation (p < 0.0001), severity of <strong>the</strong> attack (i.e., low basel<strong>in</strong>e FEV~) (p < 0.007), longer duration of attack (p < 0.05), <strong>and</strong> older age (p < 0.05) (Table II). Inclusion of covariates <strong>in</strong> <strong>the</strong> statistical model <strong>in</strong>corporat<strong>in</strong>g prior asthma medication reduced <strong>the</strong> importance of basel<strong>in</strong>e FEV~ as a covariate from ap value of 0.007 to ap value of 0.01 (model 9), even when oral <strong>the</strong>ophyll<strong>in</strong>e, <strong>in</strong>haled corticosteroids, oral corticosteroids, <strong>and</strong> <strong>in</strong>haled antichol<strong>in</strong>ergics were removed from analysis (model 10). Patients <strong>with</strong> more severe asthma on presentation, as def<strong>in</strong>ed by a lower FEV1, were less likely to show improvement after ei<strong>the</strong>r of <strong>the</strong> treatment regimens. Never<strong>the</strong>less, when adjustment was made for basel<strong>in</strong>e FEV~ <strong>in</strong> <strong>the</strong> ANCOVA model <strong>and</strong> when FEV~ was treated as a cont<strong>in</strong>uous variable, <strong>the</strong> advantage of Combivent over <strong>salbutamol</strong> rema<strong>in</strong>ed significant (p = 0.02) (model 1, Table I). Basel<strong>in</strong>e FEV~ was <strong>the</strong>n treated