Small-particle Inhaled Corticosteroid as First-line or Step-up Controller Therapy in Childhood Asthma Academic research paper on "Clinical medicine"

Original Article

Small-particle Inhaled Corticosteroid as First-line or Step-up Controller Therapy in Childhood Asthma

Willem M.C. van Aalderen, MD, PhDa, Jonathan Grigg, MDb, Theresa W. Guilbert, MD, MSc, Nicolas Roche, MD, PhDd, Elliot Israel, MDe, Richard J. Martin, MDf, Gene Colice, MDg, Dirkje S. Postma, MD, PhDh, Elizabeth V. Hillyer, DVMi, Anne Burden, MSc1, Victoria Thomas, MSc1, Julie von Ziegenweidt1, and David Price, FRCGPl,j Amsterdam, The Netherlands; London, United Kingdom; Cincinnati, Ohio; Paris, France; Boston, Mass; Denver, Colo; Washington, DC; Groningen, The Netherlands; Cambridge and Aberdeen, United Kingdom

What is already known about this topic? Evidence from randomized controlled trials regarding asthma controller therapies for children is limited, usually short-term, and often not generalizable to general practice, where most children with asthma are managed.

What does this article add to our knowledge? Over 1 outcome year, small-particle inhaled corticosteroid (ICS) was more effective than standard size—particle ICS for children initiating or stepping up ICS therapy and as effective as adding a long-acting b2-agonist in a fixed-dose combination inhaler.

How does this study impact current management guidelines? These findings challenge asthma guidelines that recommend adding a long-acting b2-agonist as the first-line alternative for stepping up therapy when asthma is not controlled by ICS monotherapy.

BACKGROUND: Because randomized controlled trials of established pediatric asthma therapies are expensive and difficult to perform, observational studies may fill gaps in the evidence base. OBJECTIVES: To compare the effectiveness of representative small-particle inhaled corticosteroid (ICS) with that of standard size—particle ICS for children initiating or stepping up ICS therapy for asthma (analysis 1) and to compare the effectiveness of ICS dose step-up using small-particle ICS with adding long-acting b2-agonist (LABA) to the ICS (analysis 2). METHODS: These historical matched cohort analyses drew on electronic medical records of children with asthma aged 5 to 11 years. Variables measured during 2 consecutive years (1 baseline year for confounder definition and 1 outcome year) included risk-domain asthma control (no hospital attendance for asthma, acute oral corticosteroids, or lower respiratory tract infection requiring antibiotics) and rate of severe exacerbations (asthma-related emergency, hospitalization, or oral corticosteroids).

RESULTS: In the initiation population (n = 797 in each cohort), children prescribed small-particle ICS versus standard size—particle ICS experienced greater odds of asthma control (adjusted odds ratio, 1.49; 95% CI, 1.10-2.02) and lower severe exacerbation rate (adjusted rate ratio, 0.56; 95% CI, 0.35-0.88).

Step-up outcomes (n = 206 in each cohort) were also significantly better for small-particle ICS, with asthma control adjusted odds ratio of 2.22 (95% CI, 1.23-4.03) and exacerbations adjusted rate ratio of 0.49 (95% CI, 0.27-0.89). The number needed to treat with small-particle ICS to achieve 1 additional child with asthma control was 17 (95% CI, 9-107) for the initiation population and 5 (95% CI, 3-78) for the step-up population. Outcomes were not significantly different for stepped-up small-particle ICS dose versus ICS/LABA combination (n = 185 in each cohort).

CONCLUSIONS: Initiating or stepping up the ICS dose with small-particle ICS rather than with standard size—particle ICS is more effective and shows similar effectiveness to add-on LABA in childhood asthma. © 2015 The Authors. Published by Elsevier Inc. on behalf of the American Academy of Allergy, Asthma & Immunology. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/ licenses/by-nc-nd/4.0/). (J Allergy Clin Immunol Pract 2015;b:B-B)

Key words: Asthma; Childhood; Small-particle beclomethasone; Fluticasone; Inhaled corticosteroid; Long-acting b2-agonist; Step-up therapy

^Department of Pediatric Respiratory Medicine and Allergy, Emma Children's

Hospital AMC, Amsterdam, The Netherlands bBlizard Institute, Queen Mary University London, London, UK cCincinnati Children's Hospital and Medical Center, Cincinnati, Ohio dCochin Hospital Group, AP-HP, University of Paris Descartes (EA2511), Paris, France eBrigham and Women's Hospital and Harvard Medical School, Boston, Mass fDepartment of Medicine, National Jewish Health and University of Colorado Denver, Denver, Colo

^Washington Hospital Center and George Washington University School of Medicine, Washington, DC hUniversity of Groningen, University Medical Center Groningen, Groningen, The

Netherlands Research in Real Life, Ltd, Cambridge, UK jAcademic Primary Care, University of Aberdeen, Aberdeen, UK Data acquisition and analyses were funded by Teva Pharmaceuticals; access to data from the Optimum Patient Care Research Database was cofunded by Research in

Abbreviations used adjOR- adjusted odds ratio adjRR- adjusted rate ratio GP- general practice ICS- inhaled corticosteroid LABA- long-acting b2-agonist NNT- number needed to treat pMDI- pressurized metered-dose inhaler

RCT- randomized controlled trial SABA-short-acting b2-agonist

Clinicians managing children with asthma in community settings, where most patients with asthma are seen, must rely on evidence for their therapeutic choices primarily from randomized controlled trials (RCTs). However, funding for large independent RCTs is limited; moreover, RCT results may not be widely

generalizable to unselected patient populations. They exclude many children with asthma who have intermittent, milder disease and who may be most prone to exacerbations. In addition, RCTs seldom include comparisons of step-up strategies, directly compare different types or formulations of inhaled corticosteroids (ICS), or provide the long-term comparisons (>12 weeks) needed for evaluating infrequent events, such as exacerbations.1-5 Head-to-head RCTs are challenging to institute because of difficulties in unravelling the roles of corticosteroid potency, initial dose from different inhaler devices, and delivered dose to the lower airways. Moreover, recruitment into RCTs of children with poor control sufficient to justify a change in therapy is notoriously very difficult.6 Finally, the enforced adherence to prescribed medication in RCTs is difficult to duplicate in clinical practice, further limiting the applicability of RCT results.

Information from observational studies of pediatric asthma therapies can complement the limited evidence currently

Real Life Ltd (RiRL, Cambridge, UK). Teva played no role in the design or conduct of the study; collection, management, analysis, or interpretation of the data; preparation, review, or approval of the manuscript; or the decision to submit the manuscript for publication. The research team at RiRL designed the study, conducted the analyses, and coordinated the writing and revision of the article in collaboration with the study steering committee.

Conflicts of interest: W. M. C. van Aalderen has received travel reimbursement from Research in Real Life (RiRL); is a member of medical advisory boards of Mun-dipharma BV, Astra-Zeneca, and Teva; and received a traveling grant from Teva and a speaker's fee from Forest. J. Grigg has received travel reimbursement from RiRL; is on the advisory boards for GlaxoSmithKline (GSK) and Novartis; and received honoraria from Novartis as a member of an advisory board for an asthma medication and received honoraria from GlaxoSmithKline for advice on an asthma medication study design. T. W. Guilbert has received travel reimbursement and medical writing support from RiRL; has received travel support to research meeting by Merck-Schering Plough; has received fees for developing questions for Pediatric Pulmonary Board exams from the American Board of Pediatrics, Pedi-atric Pulmonary Subboard; is on the Teva advisory board; has received research support from Teva; is a consultant for Teva, MAP Pharmaceuticals, GSK, and Merck; has received personal fees from GSK as a committee member, research advisor, advisory board member, and a subinvestigator; is on the Regeneron Pharmaceuticals research advisory board; was a subinvestigator for Altus Pharmaceuticals, Inspire Pharmaceuticals, Teva, and GSK; has received research support from the Centers for Disease Control and Prevention, Department of Health and Human Services, National Institutes of Health University of Wisconsin-Madison Medical and Education Research Committee, Abbott Laboratories, Array Biopharma, Mylan, Forest Research Institute, F. HoffmanLaRoche, MedImmune, KaloBios Pharmaceuticals, Vertex Pharmaceuticals, Roxane Laboratories and CompleWare Corporation, CF Foundation Therapeutics, and Roche-Genentech; and receives royalties from UpToDate. N. Roche has received travel reimbursement from RiRL; has received fees for speaking, organizing education and research, or consulting from Aerocrine, Nycomed, Astra-Zeneca, Boehringer Ingelheim, Chiesi, GSK, MEDA, MSD, Mundipharma, Novartis, Pfizer, and Teva; has received research funds from Boehringer Ingelheim, Pfizer, Novartis, and Nycomed; is on the advisory board for and has received lecture fees from Mundipharma; is on the advisory board for and has received consultancy and lecture fees from Chiesi; is on the advisory board for and has received manuscript preparation fees from Nycomed; is on the Almirall advisory board; is on the advisory board for and has received lecture fees and research support from Novartis; is on the advisory board for and has received research support from Boehringer Ingelheim; has received consultancy and lecture fees from AstraZeneca; has received lecture fees from GSK, Teva, Meda, Stallergenes, and Aerocrine; is on the advisory board for and has received lecture fees from Pfizer. E. Israel has received travel reimbursement from RiRL and Teva Specialty Pharmaceuticals; had grant support paid to his institution from Amgen, i3 Research (Biota); has received consultancy fees from Cowen & Co, Infinity Pharmaceuticals, Merck, Regeneron Pharmaceuticals, and Teva Specialty Pharmaceuticals; is on the speakers bureau and has received consultancy fees from Merck; has provided expert testimony for Campbell, Campbell, Edwards & Conroy, Ficksman & Conley, and Ryan Ryan Deluca LLP; received royalties from UpToDate; and is a Data Safety Monitoring Board member for Novartis. R. J. Martin has received

travel support from RiRL; has done consultancy work and/or received travel support and/or honoraria for attendance at advisory boards for Teva, AstraZeneca, MedImmune, Boehringer Ingelheimer, Theron, and Merck; has received research support from MedImmune and the National Heart, Lung, and Blood Institute; and received royalties from UpToDate. G. Colice has received travel expense reimbursement for meeting attendance and manuscript preparation from RiRL and has received speakers and consultancy fees from and is an advisory board member for Teva, MedImmune, Dey, Mylan, Novartis, and Alitair. In the past 3 years, the University of Groningen has received money for D. S. Postma regarding an unrestricted educational grant for research from AstraZeneca and Chiesi; travel to the European Respiratory Society and/or American Thoracic Society has been partially funded by AstraZeneca, Chiesi, GSK, and Nycomed; fees for consultancies were given to the University of Groningen by AstraZeneca, Boeringer Ingelheim, Chiesi, GSK, Nycomed, and Teva; and money for travel and lectures in China was paid by Chiesi. D. S. Postma has received travel reimbursement from RiRL. E. V. Hillyer has received travel reimbursement and consultancy fees from RiRL; has received payment for manuscript preparation from Merck and Teva-France; and is a consultant for RiRL. A. Burden, V. Thomas, and J. von Zie-genweidt have received research support from Teva and are employed by RiRL. D. Price has received research support from Teva; is a member of the following boards for which fees were paid to RiRL: Aerocrine, Almirall, Amgen, AstraZeneca, Boehringer Ingelheim, Chiesi, Meda, Mundipharma, Napp, Novartis, and Teva; has received consultancy fees paid to RiRL from Almirall, Amgen, AstraZeneca, Boehringer Ingelheim, Chiesi, GlaxoSmithKline, Meda, Mundipharma, Napp, Novartis, Pfizer, and Teva; has received research support for which fees were paid to RiRL from UK National Health Service, British Lung Foundation, Aerocrine, AstraZeneca, Boehringer Ingelheim, Chiesi, Eli Lilly, GlaxoSmithKline, Meda, Merck, Mundipharma, Novartis, Orion, Pfizer, Respiratory Effectiveness Group, Takeda, Teva, and Zentiva; has received fees (paid to RiRL) for manuscript preparation from Mundipharma and Teva; has received travel reimbursement from Aerocrine, Boehringer Ingelheim, Mundipharma, Napp, Novartis, and Teva; has received funding paid to RiRL for patient enrollment or completion of research from Almirall, Chiesi, Teva, and Zentiva; has received unrestricted funding, paid to RiRL, for investigator-initiated research from Aerocrine, AKL Ltd, Almirall, Boehringer Ingelheim, Chiesi, Meda, Mundipharma, Napp, Novartis, Orion, Takeda, Teva, and Zentiva; has a pending patent for Phytopharmaceuticals with AKL Ltd; has shares in AKL Ltd, which produces phytopharmaceuticals; and owns 80% of RiRL Ltd and its subsidiary social enterprise Optimum Patient Care; and is peer reviewer for grant committees: Medical Research Council (2012), Efficacy and Mechanism Evaluation programme (2012), and HTA (2012).

Received for publication September 17, 2014; revised April 9, 2015; accepted for publication April 23, 2015.

Available online ■■

Corresponding author: David Price, FRCGP, Division of Applied Health Sciences, Academic Primary Care, University of Aberdeen, Polwarth Bldg, Foresterhill, Aberdeen AB25 2ZD, UK. E-mail: dprice@rirl.org.

2213-2198

© 2015 The Authors. Published by Elsevier Inc. on behalf of the American Academy of Allergy, Asthma & Immunology. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). http://dx.doi.org/10.1016/j.jaip.2015.04.012

available from RCTs.7 Here, we report the results of 2 observational analyses, using routinely collected health care data, designed to evaluate the effectiveness of 1-year controller therapy for managing several clinically common situations in children with asthma.

The objective of analysis 1 was to compare outcomes in children after a first prescription or a stepped-up dose of either a small-particle ICS or a standard size—particle ICS. We chose small-particle beclomethasone and fluticasone for comparison for several reasons. These 2 ICS are the most widely used in the United States today. In previous observational studies of adults, we found that small-particle beclomethasone (ICS particles with median mass aerodynamic diameter of 1.1 mm) is prescribed for asthma at significantly lower doses, but shows effectiveness over 1 year similar to or better than that of fluticasone (median mass aerodynamic diameter, 2.4-3.2 mm, depending on formulation).8-10 Formulations of small-particle ICS may be particularly relevant to the treatment of children with asthma because of children's physical size (smaller airways than those of adults), as well as the association of peripheral, small airways dysfunction with uncontrolled asthma.11,12 In adults, small-particle beclomethasone has greater and more uniform lung deposition, reaching both large and small airways, than does larger-particle ICS.13-15 In children with asthma, the lung deposition of small-particle ICS ranges from means of 37% to 55% of ex-actuator dose, depending on age and inhaler device.16,17 For young children (ages 3-7 years), 1 month's treatment with small-particle beclomethasone administered using a valved holding chamber significantly decreased bronchial hyperresponsiveness.18 In a small study, 20 children (ages 5-14 years) with stable, moderate asthma experienced improvements in lung function after being switched from standard ICS to small-particle beclomethasone.19 Two small randomized comparative trials failed to demonstrate significant differences in effectiveness between small-particle beclomethasone and fluticasone for

children.20,21

The objective of our analysis 2 was to compare the effectiveness of stepping up asthma therapy by increasing the ICS dose as a small-particle ICS versus adding a long-acting b2-agonist (LABA) to the ICS in fixed-dose combination or separate inhalers. A recent RCT and meta-analyses of previous RCTs comparing add-on LABA with increased ICS dose for

children have been inconclusive4,22 or report similar out-

comes23 with these 2 step-up strategies; however, the ICS studied were all larger-particle formulations (budesonide and fluticasone).

METHODS

Analyses and patients

These matched cohort analyses drew on anonymized clinical data (1997 through January 2011) contained in 2 UK primary care electronic databases: the General Practice Research Database, now part of the National Health Service Clinical Practice Research Datalink, and the Optimum Patient Care Research Database, previously well-described (see this article's Online Repository at www. jaci-inpractice.org).9,24-26 The study was done to standards suggested for observational studies, including an independent steering committee (not remunerated for participation), use of an a priori analysis plan, and well-maintained and monitored study databases.27 Children's characteristics were cross-referenced between the 2 data sets to avoid duplication.

TABLE I. Definitions of database-derived study outcome measures

Risk-domain asthma control, includes all of the following:

1. No asthma-related* hospital attendance or admission, ED

attendance, out-of-hours attendance, or outpatient hospital attendance, and

2. No GP consultation for lower respiratory tract infection requiring

antibiotics, and

3. No prescription for acute course of oral corticosteroids. Overall control,! includes all of the following:

1. No asthma-related* hospital attendance or admission, ED

attendance, out-of-hours attendance, or outpatient hospital attendance, and

2. No GP consultation for lower respiratory tract infection requiring

antibiotics, and

3. No prescription for acute course of oral corticosteroids, and

4. Average <2 puffs daily dose of SABA (salbutamol <200 mg/d or

terbutaline <500 mg/d) Number of severe exacerbations,! defined as any of the following:

1. Asthma-related* hospital attendance or admission or ED

attendance, or

2. Acute course of oral corticosteroids

Number of clinical exacerbations,! defined as any of the following:

1. Asthma-related* hospital attendance or admission or ED attendance,

2. GP consultation for lower respiratory tract infection requiring an-

tibiotics, or

3. Acute course of oral corticosteroids Treatment stability, includes all of the following:

1. Risk-domain asthma control (see above) and

2. No additional therapy after the index date as

a. increased ICS dose (by >50%), or

b. use of additional therapy as LABA, LTRA, or theophylline Respiratory hospitalizations, defined as

• hospitalizations with an asthma Read code + uncoded hospitalizations occurring within a 7-d window (either side of the hospitalization date) of a lower respiratory tract Read code

Mean daily ICS dose during the baseline and the outcome year, defined as

• number of days supply of ICS divided by 365

ED, Emergency department; LTRA, leukotriene receptor antagonist.

*Asthma-related events in the database included all events with a lower respiratory tract code, including all asthma codes and lower respiratory tract infection codes. fOverall control was not an outcome measure in the small-particle ICS step-up vs add-on LABA comparisons (analysis 2).

ZFor the exacerbation definitions, any criteria occurring within 2 wk of each other are counted as 1 exacerbation.

We studied 2 consecutive years of data for each eligible child, aged 5 to 11 years at the time of the index prescription: a baseline year preceding the index prescription date, included for defining potential baseline confounders, followed by an outcome year. We required a recorded asthma diagnosis in the database or evidence of active asthma, defined as 2 or more prescriptions for asthma therapy (controller or reliever) during the baseline year. In addition, during the outcome year, children had to have received at least 1 asthma prescription in addition to the index prescription. Children were excluded from the study for an ever-diagnosis of any chronic respiratory disease other than asthma or a baseline year prescription for maintenance oral corticosteroid therapy or an ICS/LABA combination inhaler.

For analysis 1 comparing small-particle and standard size—particle ICS, we included children prescribed their first ICS (initiation population) or increased ICS dose (step-up population) as beclomethasone

4 VAN AALDEREN ET AL j allergy clin Immunol pract

month 2015

TABLE II. Baseline characteristics of children in the small-particle ICS and standard size-particle ICS cohorts (analysis 1)

Initiation population Step-up population

Characteristic Small-particle ICS (n = 797) Standard SP ICS (n = 797) P value* Small-particle ICS (n = 206) Standard SP ICS (n = 206) P value*

Sex: malef 455 (57.1) 455 (57.1) NA 124 (60.2) 124 (60.2) NA

Age at index date (y),f mean ± SD Year of index prescription,f median (IQR) Recorded comorbidity! Possible atopy Rhinitis diagnosis/Rx Eczema diagnosis/Rx Preschool wheeze diagnosis 7.7 ± 2 2004 (2002-2006) 2003 7.7 ± 2 (2001-2005) NA <.001 7 2005 .5 ± 2.1 (2003-2007) 7 2005 .5 ± 2.1 (2002-2007) NA <.001

558 (70.0) 1 A H CKH 574 ono (72.0) .37 nm 159 /IO (77.2) nn -1 \ 161 (78.2) CT -1 T\ .81 1A

140 (1/.6) 557 (69.9) 74 (13.3) 202 554 68 (25.3) (69.5) (12.1) <.001 .87 .83 42 153 46 (20.4) (74.3) (30.1) 50 160 54 (24.3) (77.7) (35.1) .34 .43 .76

Preschool asthma diagnosis/Rx 296 (48.8) 293 (48.0) .95 120 (67.0) 128 (75.7) .14

Risk-domain asthma control Overall control 655 (82.2) 576 (72.3) 656 (82.3) 579 (72.6) .93 .79 160 (77.7) 87 (42.2) 160 (77.7) 87 (42.2) NA NA

Spacer device prescribed 232 (29.1) 233 (29.2) .95 88 (42.7) 78 (37.9) .29

Mean daily SABA dose (mg/d)f

0 325 (40.8) 325 (40.8) 3 (1.5) 3 (1.5)

1-100 ini onn 165 (20.7) OHO il A 1 \ 165 (20.7) OHO 1 \ 1 c 42 (20.4) 34 (16.5) 1 1

101-200 201-400 208 (26.1) 84 (10.5) 208 (26.1) 77 (9.7) .18 63 (30.6) 66 (32.0) 71 (34.5) 59 (28.6) .11

>400 Median (IQR) daily ICS dose (mg/d)§ Last ICS dose before the index date (mg/d)f •§ 15 (1.9) NA NA 22 (2.8) NA NA — 32 (15.5) 55 (27-82) 39 (18.9) 55 (25-82) .57

1-99 40 (19.4) 44 (21.4)

1 f\f\ 1 QQ 1 A1 /"71 A \ 1 A /H

100-199 >200 14 7 (71.4) 19 (9.2) 143 (69.4) 19 (9.2) .62

Severe exacerbationsf 0 743 (93.2) 1 49 (6.1) >2 5 (0.6) Asthma consultation and no oral corticosteroidsf 0 400 (50.2) 1 297 (37.3) 743 (93.2) 49 (6.1) 5 (0.6) NA 174 (84.5) 11 HQ 1 \ 174 (84.5) 27 (13.1) 5 (2.4) NA

27 (13.1) 5 (2.4)

400 (50.2) 297 (37.3) .15 64 (31.1) 75 (36.4) 64 (31.1) 75 (36.4) .11

2 82 (10.3) 75 (9.4) 45 (21.8) 36 (17.5)

>3 18 (2.3) 25 (3.1) 22 (10.7) 31 (15.0)

IQR, interquartile range; NA, not applicable; Rx, therapy; standard SP, standard size—particle.

Data are presented as n (%) except otherwise indicated.

*Matched cohorts were compared using conditional logistic regression.

fMatching variable (year of index prescription matched ± 4 y; for details, please see this article's Online Repository at www.jaci-inpractice.org).

^Possible atopy was defined as any 1 or more of the following: recorded rhinitis diagnosis, rhinitis therapy, eczema diagnosis, eczema therapy. Preschool wheeze was capturedthrough database coding; concomitant rhinitis, eczema, and preschool asthma were captured through database-coded diagnosis or therapy for same. Preschool was defined as age 1 to 3 y. §The doses of ICS were standardized to equivalence with small-particle beclomethasone and fluticasone; thus, doses of large-particle beclomethasone and budesonide were halved.

dipropionate hydrofluoroalkane (QVAR; Teva UK Ltd, Eastbourne, East Sussex, UK), of interest to the study sponsor and the only small-particle ICS available in the United Kingdom at the time, or flutica-sone propionate (hydrofluoroalkane or chlorofluorocarbon formulation; Flixotide; GlaxoSmithKline UK Ltd, Brentford, Middlesex, UK) by pressurized metered-dose inhaler (pMDI). Children in the step-up population were prescribed an ICS (any type) during the baseline year and on the index date were prescribed an increase of 50% or more in the ICS dose as either small-particle ICS or standard size—particle ICS.

For analysis 2 (step-up comparisons of small-particle ICS vs addon LABA), eligible children prescribed ICS during the baseline year by pMDI or breath-actuated metered-dose inhaler were stepped up to 1 of 3 options: 1) an increase of 50% or more in the ICS dose as small-particle beclomethasone by pMDI or breath-actuated inhaler (ICS step-up cohort); 2) addition of LABA, with no change in the

ICS dose, via fixed-dose ICS/LABA combination inhaler (ICS/LABA combination cohort) as either fluticasone propionate/salmeterol xinafoate (Seretide; GlaxoSmithKline UK Ltd, Brentford, Middlesex, UK) or budesonide/formoterol fumarate dihydrate (Symbi-cort; AstraZeneca Ltd, Luton, Bedfordshire, UK); or 3) addition of LABA by separate inhaler, with no change in the ICS drug, dose, or inhaler (separate ICS + LABA cohort).

Study end points

Composite measures to evaluate asthma-related outcomes, used in our previous studies,8,9,26 are defined in detail in Table I. In brief, risk-domain asthma control (asthma control) was defined as no hospital attendance for asthma, acute oral corticosteroid course, or general practice (GP) consultation for lower respiratory tract infection requiring antibiotics; the latter criterion was included because, in

TABLE III. Outcome-year results for matched cohorts prescribed small-particle ICS or standard SP ICS for first-line or step-up therapy (analysis 1)

Initiation population

Step-up population

Outcome

Small-particle ICS (n = 797) Standard SP ICS (n = 797) Small-particle ICS (n = 206) Standard SP ICS (n = 206)

Risk-domain asthma control 702 (88.1) 667 (83.7) 182 (88.3) 156 (75.7)

Overall control 488 (61.2) 441 (55.3) 88 (42.7) 83 (40.3)

Treatment stability 631 (79.2) 545 (68.4) 156 (75.7) 134 (65.0)

Severe exacerbation

0 759 (95.2) 735 (92.2) 189 (91.7) 170 (82.5)

1 29 (3.6) 54 (6.8) 14 (6.8) 29 (14.1)

>2 9 (1.1) 8 (1.0) 3 (1.5) 7 (3.4)

Clinical exacerbation

0 706 (88.6) 671 (84.2) 182 (88.3) 158 (76.7)

1 71 (8.9) 104 (13.0) 19 (9.2) 32 (15.5)

>2 20 (2.5) 22 (2.8) 5 (2.4) 16 (7.8)

Disaggregated results of composite measures

P value*

P value*

>1 asthma-related hospital attendance 10 (1.3) 10 (1.3) 1.0 2 (1.0) 7 (3.4) .12

>1 acute course of oral corticosteroids 36 (4.5) 59 (7.4) .057 17 (8.3) 35 (17.0) .013

> 1 GP consultation for LRTI requiring antibiotic 57 (7.2) 73 (9.2) .13 12 (5.8) 20 (9.7) .11

Mean >2 puffs daily SABA 261 (32.7) 297 (37.3) .041 110 (53.4) 109 (52.9) .92

Increase in ICS dose or additional therapy 89 (11.2) 157 (19.7) <.001 30 (14.6) 31 (15.0) .88

LRTI, lower respiratory tract infection; standard SP, standard size—particle. Data are presented as n (%). *Conditional logistic regression.

practice, asthma exacerbations can be confused with lower respiratory tract infection. , The definition of overall control included asthma control plus limited reliever use (daily average of <2 puffs of short-acting b2-agonist [SABA], defined as albuterol <200 mg/d or terbutaline <500 mg/d, calculated as the dispensed amount divided by 365).

Severe exacerbations were defined according to American Thoracic Society/European Respiratory Society criteria (asthma-related emergency or hospitalization or oral corticosteroids). A second extended definition of clinical exacerbation, developed with the guidance of respiratory clinicians, included the additional criterion of a GP consultation for lower respiratory tract infection as defining an exacerbation. Another composite measure, treatment stability, was defined as risk-domain asthma control plus no treatment change (Table I).

Statistical analysis

Composite outcome measures and analyses were prespecified according to standard operating procedures of the research group31 (see full details in this article's Online Repository at www.jaci-inpractice.org). The analyses were carried out using IBM SPSS Statistics version 19 (SPSS Statistics, IBM, Somers, NY), SAS versions 9.2 and 9.3 (SAS Institute, Marlow, Buckinghamshire, UK), and Microsoft Excel 2007 (Microsoft, Bellevue, Wash); statistically significant results were defined as P < .05.

To minimize potential confounding by baseline differences between treatment cohorts, we matched children sequentially on demographic characteristics (sex then age) and several clinically important indicators of baseline asthma severity (sequence described in this article's Online Repository at www.jaci-inpractice.org). Children in the step-up cohorts of analysis 1 and those in analysis 2 were also matched on the last ICS dose prescribed before the index prescription. We calculated the mean daily ICS dose during baseline as the dispensed amount divided by 365, standardizing the dose to

that of small-particle beclomethasone (Table I). Matching ratios were chosen to maximize patient numbers and thus statistical power for the comparisons (see this article's Online Repository at www.jaci-inpractice.org).

We evaluated all baseline and outcome variables using summary statistics. Conditional logistic regression was used to quantify baseline differences between matched cohorts. Potential confounding factors were examined for colinearity and clinical importance to select those used as potential confounders in the regression modeling of outcomes; a detailed list is provided in this article's Online Repository at www.jaci-inpractice.org. Variables that differed between treatment cohorts at P < .10 were considered as potential confounding factors, as well as baseline variables predictive of outcomes in multivariate analyses at P < .05.

The odds of achieving risk-domain asthma control during the outcome year were compared between matched treatment cohorts using conditional binary logistic regression models. Asthma control status was used as the dependent variable, with treatment and potential confounding factors as explanatory variables. Similar methods were used to calculate the adjusted odds ratio (adjOR) for overall control (analysis 1 only) and treatment stability. Where a significant difference was found between treatment cohorts in the primary outcome of risk-domain asthma control, we also calculated the difference in proportions achieving control between treatment cohorts (using the same conditional binary logistic regression model; with 95% CI) and the number needed to treat (NNT) for 1 additional child to achieve control (the inverse of the difference in proportions; with 95% CI) to better quantify the difference in treatment effect between cohorts. To account for multiple comparisons in the primary outcomes, a false-discovery rate controlling procedure was used (for description, see this article's Online Repository at www.jaci-inpractice.org).

The total number of severe exacerbations in the outcome year was compared between treatment cohorts using a conditional

FIGURE 1. AdjORs and adjRRs comparing treatment cohorts during 1 outcome year. A, Initiation asthma therapy comparing small-particle ICS vs standard size—particle ICS (standard size—particle ICS set at odds/rate = 1.0). LRTI, Lower respiratory tract infection. Adjusted for the following confounders: *Consultation for LRTI requiring antibiotic (yes/no). tAsthma control status, number of asthma/allergy prescriptions; ZLABA use (yes/no) and number of non—asthma-related consultations. §Rhinitis diagnosis, year of index date. #Consultation

Poisson regression model to obtain an estimate of relative exacerbation rates. The model used empirical standard errors (for more conservative CI estimations), and adjustments were made for potential baseline confounders. The adjusted rate ratios (adjRRs) for clinical exacerbations and hospitalizations were calculated in a similar fashion.

For children initiating ICS, we conducted several subgroup analyses to further examine factors that could be influencing the outcomes of therapy, including spacer prescriptions and age grouped as 5 to 6 and 7 to 11 years (described in this article's Online Repository at www.jaci-inpractice.org). For the step-up comparisons of small-particle ICS versus add-on LABA, we calculated the total b2-agonist load (SABA + LABA) in terms of total hours of coverage (see this article's Online Repository at www.jaci-inpractice.org).

RESULTS

Identification of children in the data sets and the results of cohort matching are depicted in Figures E1-E4 in this article's Online Repository at www.jaci-inpractice.org.

Analysis 1: Small-particle ICS versus standard size—particle ICS as first-line or step-up ICS therapy

Initiation population. The 2 initiation cohorts were well matched at baseline for composite measures of asthma severity (Table II; see Table E1 in this article's Online Repository at www.jaci-inpractice.org). A significant difference in median index date was minor and likely not clinically meaningful. Significantly fewer children in the small-particle ICS cohort had rhinitis or were receiving rhinitis therapy (18% vs 25% for standard size—particle ICS; P < .001); rates of possibly atopy were similar (70% vs 72%, respectively; Table II). There were significant mean ± SD differences in index date doses (167 ± 85 vs 221 ± 157 mg/d for small-particle vs standard size—particle ICS; P < .001), and fewer children in the small-particle ICS cohort were prescribed an ICS dose of 400 mg/d or more (7.4% vs 15.6% for standard size—particle ICS; see Figure E5, A, in this article's Online Repository at www.jaci-inpractice.org).

During the outcome year (Table III), the odds of achieving risk-domain asthma control were significantly greater for children initiating small-particle ICS than for those initiating standard size—particle ICS therapy (adjOR, 1.49; 95% CI, 1.10-2.02; Figure 1, A). The adjusted difference in proportions achieving asthma control (small-particle ICS vs standard size—particle ICS) was 0.06 (95% CI, 0.01-0.11; adjusted for baseline GP consultation for lower respiratory tract infection resulting in an antibiotic prescription). The NNT to achieve 1 additional child

with asthma control using small-particle ICS was 17.0 (95% CI, 9.3-107.2).

In addition, children in the small-particle ICS cohort had significantly greater odds of overall asthma control (adjOR, 1.30; 95% CI, 1.04-1.62) and significantly lower exacerbation rates (for both definitions; severe exacerbations adjRR, 0.56; 95% CI, 0.35-0.88) than did those in the standard size—particle ICS cohort, also with false-discovery rate correction. Children prescribed small-particle ICS had significantly greater odds of treatment stability, driven by significantly lower rates of therapy change (Table III; see Table E2 in this article's Online Repository at www.jaci-inpractice.org; Figure 1, A). Respiratory-related hospitalization rates were low and comparable between treatment cohorts.

Subgroup analyses: Initiation population. A total of 465 children in the standard size—particle ICS cohort were prescribed a spacer device in the baseline and/or outcome year, but of the 465 matched children prescribed small-particle ICS, only 69% were prescribed a spacer. Outcomes for this subgroup paralleled those for the full matched initiation cohorts, with adjusted odds of asthma control significantly higher (risk-domain asthma control adjOR, 1.81; 95% CI, 1.24-2.65) and relative exacerbation rates significantly lower (severe exacerbation adjRR, 0.53; 95% CI, 0.29-0.98) for the small-particle ICS subcohort (see Table E3 in this article's Online Repository at www.jaci-inpractice.org).

Results of unmatched subgroup analyses of children with no rhinitis or rhinitis therapy and those without exacerbations during the baseline year were consistent with the main findings (see Online Repository text and Table E3).

Results of matched cohort analyses by age group found that outcomes with small-particle ICS remained significantly better (compared with standard size—particle ICS) for 5- to 6-year-old children but were comparable with standard size—particle ICS for the 7- to 11-year-old children (Online Repository). Figure E6 (in this article's Online Repository at www.jaci-inpractice.org) suggests greater variability in ICS response in younger children and the standard size—particle ICS cohort.

Step-up population. The 2 step-up cohorts were well matched at baseline (Table II; Table E1). Statistically significant differences in mean index date dose (224 [95] vs 262 [170] mg/ d for small-particle vs standard size—particle ICS; P < .001) were likely not clinically significant (Figure E5, B).

for LRTI requiring antibiotic, rhinitis diagnosis, year of index date. -Rhinitis diagnosis, acetaminophen prescription (yes/no). B, Step-up asthma therapy comparing small-particle ICS vs standard size—particle ICS (standard size—particle ICS set at odds/rate =1.0). Adjusted for the following confounders: *Acetaminophen prescription, lower respiratory tract-related inpatient admissions (including vague). INumber of SABA prescriptions. zfNumber of asthma/allergy prescriptions (categorized). §Number of asthma consultations. #Number of asthma consultations. ANumber of ICS prescriptions and Charlson comorbidity index score. C, Adjusted outcome measures comparing step-up with increased dose of small-particle ICS vs add-on LABA in fixed-dose combination with ICS (ICS/LABA set at odds/ rate = 1.0). For details of confounding factors examined, see this article's Online Repository at www.jaci-inpractice.org. Adjusted for the following confounders: *No significant effects (unadjusted odds ratio). INumber of antibiotic prescriptions for LRTI, number of asthma consultations, emergency department attendance for asthma or lower respiratory tract reasons. |Year of index date, number of primary care consultations, average SABA daily dose, number of acute courses of oral corticosteroids, inpatient admissions for asthma or lower respiratory tract reasons; §Gastroesophageal reflux disease diagnosis, number of asthma consultations, number of antibiotic prescriptions for LRTI; #Definite and probable asthma-related inpatient admissions, year of index date.

During the outcome year (Table III), children stepping up their ICS therapy as small-particle ICS had significantly greater odds of achieving risk-domain asthma control. The adjusted difference in proportions achieving asthma control (small-particle ICS vs standard size—particle ICS) was 0.21 (95% CI, 0.100.40; adjusted for baseline acetaminophen prescriptions and baseline lower respiratory tract-related inpatient admissions). The NNT to achieve 1 additional child with asthma control using small-particle ICS was 4.8 (95% CI, 2.5-77.8).

Children prescribed small-particle ICS had significantly lower exacerbation rates (both definitions) than did children stepping up their ICS therapy as standard size—particle ICS, as well as significantly greater odds of treatment stability (at the 5% level; Figure 1, B). Results for overall control, incorporating SABA use, were comparable for the 2 cohorts, as were lower respiratory tract-related hospitalization rates during the outcome period (Figure 1, B).

Analysis 2: Small-particle ICS step-up versus add-on LABA in ICS/LABA combination inhaler

After 1:1 matching, statistically significant baseline differences between the ICS step-up and ICS/LABA combination cohorts were minor (Table IV; see Table E4 in this article's Online Repository at www.jaci-inpractice.org).

Median (interquartile range) ICS doses prescribed at the index date were 200 (200-400) mg/d for the ICS step-up and 100 (100200) mg/d for the ICS/LABA combination cohort (P < .001 for the comparison). There were no significant differences in the adjusted outcome measures between ICS step-up and ICS/LABA combination cohorts (Figure 1, C). However, for the ICS/LABA combination cohort, change in therapy was significantly more common, and the total b2-agonist coverage (LABA plus SABA) was significantly greater (Table V; see Table E5 and Figure E7, A, in this article's Online Repository at www.jaci-inpractice.org).

The ICS step-up and separate ICS + LABA cohorts were similar after 1:2 matching (see Tables E4 and E6 in this article's Online Repository at www.jaci-inpractice.org). The odds of treatment stability were higher for ICS step-up, but the odds of asthma control and exacerbation rates were comparable between cohorts (see Table E7 and Figure E8 in this article's Online Repository at www.jaci-inpractice.org). Total b2-agonist coverage was greater for the separate ICS + LABA cohort (Table E5; Figure E7, B).

DISCUSSION

The observational design of these 2 analyses provides direct comparative evidence, not currently available from RCTs, about clinically relevant outcomes using different treatment approaches to common clinical situations in children with asthma. We found that the rate of severe exacerbations among children with asthma aged 5 to 11 years prescribed small-particle ICS was significantly lower than that for children prescribed standard size—particle ICS. both for those initiating ICS (adjRR, 0.56) and for those stepping up the ICS dose (adjRR, 0.49). Moreover, the adjusted odds of the risk-domain composite measure of asthma control were significantly greater for children initiating small-particle ICS, as well as for children stepping up the dose of small-particle ICS, when compared with those for children initiating or stepping up the dose of standard size—particle ICS. The NNTs to achieve 1 additional child with asthma control using small-particle ICS were 17 and 5 for the initiation and step-up populations, respectively.

TABLE IV. Baseline characteristics of children in the small-particle ICS step-up and combination ICS/LABA cohorts (analysis 2)

ICS step-up vs ICS/LABA combination

ICS ICS/LABA

step-up combination Characteristic (n = 185) (n = 185) P value*

Sex: malef 114(61.6) 114(61.6) NA

Age at index date (y),f mean 8.3 ± 1.9 8.3 ± 1.9 NA ± SD

Recorded comorbidity!

Possible atopy 148 (80.0) 143 (//.3) .54

Rhinitis diagnosis/Rx 48 (25.9) 46 (24.9) .82

Eczema diagnosis/Rx 135 (/3.0) 131 (/0.8) .66

Preschool wheeze diagnosis /0 (3/.8) 62 (33.5) .39

Preschool asthma 94 (50.8) 98 (53.0) .66

diagnosis/Rx

129 (69.7) 129 (69.7) NA

70 (37.8) 85 (45.9) .12

Mean daily SABA dose (mg/d)f

0-100 33 (17.8) 101-200 58 (31.4) 201-400 66 (35.7) 401-800 20 (10.8) >800 8 (4.3)

Median (IQR) daily ICS dose 55 (27-110)

(mg/d)§

Last ICS dose before the index date (mg/d)f'§

1-50 0 (0.0) 51-100 107 (57.8) 101-200 78 (42.2)

Severe exacerbations

0 147 (79.5)

1 27 (14.6)

2 6 (3.2) >3 5 (2.7)

Asthma consultation/no oral corticosteroidst

0 61 (33.0) 61 (33.0)

1 61 (33.0) 61 (33.0) .09

2 3/ (20.0) 28 (15.1)

>3 26 (14.1) 35 (18.9)

IQR, interquartile range; NA, not applicable; Rx, therapy; standard SP, standard size—particle.

Data are presented as n (%) except otherwise indicated. *Matched cohorts were compared using conditional logistic regression. fMatching variable (for details, please see this article's Online Repository at www. jaci-inpractice.org).

^Possible atopy was defined as any 1 or more of the following: recorded rhinitis diagnosis, rhinitis therapy, eczema diagnosis, eczema therapy. Preschool wheeze was captured through database coding; concomitant rhinitis, eczema, and preschool asthma were captured through database-coded diagnosis or therapy for same. Preschool was defined as age 1 to 3 y.

§The doses of ICS were standardized to equivalence with small-particle beclomethasone and fluticasone; thus, doses of large-particle beclomethasone and budesonide were halved.

In the comparison of step-up strategies in analysis 2, increasing small-particle ICS dose was comparable to adding LABA in a combination ICS/LABA inhaler with regard to both asthma control and exacerbation measures.

Risk-domain asthma controlt Spacer device prescribed

24 (13.0) 6/ (36.2)

66 (35./) .29

22 (11.9) 6 (3.2) 55 (2/-110) .26

1 (0.5)

TABLE V. Outcome-year results for matched cohorts prescribed increased ICS dose of small-particle ICS or combination ICS/LABA (analysis 2)

ICS step-up vs ICS/LABA combination

Outcome ICS dose step-up (n = 185) ICS/LABA combination (n = 185)

Risk-domain asthma control 148 (80.0) 146 (78.9)

Treatment stability 132 (71.4) 118 (63.8)

Severe exacerbation

0 162 (87.6) 161 (87.0)

1 17 (9.2) 14 (7.6)

>2 6 (3.2) 10 (5.4)

Clinical exacerbation

0 151 (81.6) 148 (80.0)

1 24 (13.0) 25 (13.5)

>2 10 (5.4) 12 (6.5)

Disaggregated results of composite measures P value*

>1 asthma-related hospital 1 (0.5) 2(1.1) .57

attendance

>1 acute course of oral 22 (11.9) 23 (12.4) .52

corticosteroids

>1 GP consultation for LRTI 19 (10.3) 19 (10.3) 1.0

requiring antibiotic

Mean >2 puffs daily SABA 110 (59.5) 98 (53.0) .11

Increase in ICS dose or additional 29 (15.7) 43 (23.2) .002

therapy

LRTI, lower respiratory tract infection; standard SP, standard size—particle. Data are n (%) unless otherwise stated. *Conditional logistic regression.

In analysis 1, as compared with findings in previous adult studies,8,9 greater differences were evident in outcomes favoring small-particle ICS relative to standard size—particle ICS, for both initiation and step-up cohorts. There are few other studies comparing small- and standard size—particle ICS for children with asthma. A Cochrane analysis compared ciclesonide with other ICS for children but was inconclusive.3 The Cochrane analysis that compared small-particle hydrofluoroalkane-beclo-methasone and fluticasone for children with asthma was limited because only 2 studies were found.2 Robroeks et al20 reported no difference between 12-week therapy with small-particle beclomethasone or fluticasone in either anti-inflammatory effects or clinical outcomes for 33 children with asthma in a small crossover study, although children at study start had normal lung function and generally well-controlled asthma. van Aal-deren et al12,21 showed that small-particle beclomethasone and fluticasone had similar effects on lung function and overall asthma control in an 18-week ICS step-down study in 280 children aged 5 to 12 years with mild-to-moderate asthma. Neither study examined comparative effects of these 2 ICS on exacerbations in children over a 1-year period. The present analyses add to these shorter-term, smaller studies and demonstrate the value of pragmatic research for understanding asthma therapies in children.

The pattern of results according to the ICS dose during the outcome year suggests that the odds of risk-domain asthma control were better for children at lower doses of both

small-particle and standard size—particle ICS (depicted in Figure E6). We found a similar pattern of results in a previous US retrospective study,8 whereby adults with lower medication possession ratio had better risk-domain asthma control and fewer exacerbations. We speculate that this is because adherence with ICS increases after exacerbations, as supported by previous observational studies,33,34 and that patients whose asthma is well-controlled may be less likely to take their ICS regularly.

In our subanalyses, small-particle ICS offered an advantage especially in younger children (5-6 years old vs 7-11 years old). The reason for this additional advantage is unknown, but Amirav and Newhouse35 have speculated that better outcomes with ICS in young children will be achieved using formulations with smaller particle size. We also found no difference from the main findings in comparative results when standard size—particle ICS was given under ideal circumstances with a spacer (the small-particle ICS subanalysis cohort was not required to have a spacer, and 31% did not). This differs from the accepted view that spacer use is necessary to achieve improved lung deposition in children prescribed a pMDI.36 Deposition studies in children as young as 5 to 7 years have confirmed that even without a spacer, 38% or more of a single inhalation of small-particle beclomethasone reaches the lungs. Furthermore, good lung deposition (>30%) occurs with small-particle beclomethasone even in patients who are unable to precisely coordinate inhalation and actuation of a pMDI.13

In analysis 2 comparing the stepped-up dose of small-particle ICS versus adding LABA by fixed-dose combination inhaler, we found that the percentages of children meeting outcome measures improved substantially in both cohorts, without significant differences between cohorts in effectiveness. Although both UK asthma management guidelines and US drug labeling recommend the prescribing of ICS/LABA as fixed-dose combination inhalers for children and adolescent patients to ensure their adherence to concomitant therapy,37,38 we found a substantial number of children receiving ICS and LABA by separate inhalers. Of potential interest, a novel finding in this analysis was that the b2-agonist load over the 1-year outcome period was significantly lower, and shorter-acting, in the small-particle ICS than in the ICS/LABA combination cohort.

Our findings that increasing the ICS dose for children with uncontrolled asthma can provide clinically meaningful improvement is supported by the work of Lemanske et al. In their rigorously performed, National Institutes of Health— sponsored, randomized crossover trial assessing differential responses to 3 step-up strategies for 165 children with uncontrolled asthma on low-dose fluticasone, the best response for most of the children was obtained with LABA step-up therapy; however, many children also demonstrated best response to ICS step-up.23 Response was measured over only 16 weeks in their study, and fluticasone was administered as the ICS. We can speculate, on the basis of the results of the present study, that more children may have demonstrated a best response to ICS step-up if a small-particle ICS had been used as the ICS. Moreover, our longer-term findings provide further information that calls into question the recommendations of UK and international asthma guidelines, which identify add-on LABA as the first-line alternative for stepping up therapy when asthma is not controlled by ICS mono-therapy.37,39 Properly designed RCTs might be of help to further explore the findings reported here.

A limitation of this study is the absence of data on adverse effects of therapy, particularly growth. Information on height is not routinely available from database sources, and height measurements in daily practice are not performed using the rigorous criteria of an RCT. Although linear growth could not be measured in our study, recent RCTs found that daily use of beclomethasone is associated with a decreased linear growth of approximately 1 cm compared with the use of intermittent ICS or leukotriene receptor antagonist; a similar pattern has been found with regular use of other ICS.40-43

An important limitation of observational studies is the possibility of unrecognized confounders. Although the matching process is designed to reduce this possibility by comparing cohorts of similar baseline asthma severity, we note that for the initiation population in analysis 1, there were fewer children with rhinitis, a factor that can influence asthma control, in the small-particle cohort than in the standard size—particle ICS cohort. However, similar proportions of children in the 2 cohorts had possible atopy, and the results of the no rhinitis subanalysis supported the main findings.

As small-particle beclomethasone is not approved for use in the United Kingdom for children younger than 12 years, it might be argued that GPs prescribing this product would be more knowledgeable about asthma. However, during the study period, similar numbers of children in the data sets were initiated on small-particle beclomethasone (~4%) and fluticasone (~6%). Two other observations argue against large differences between the 2 physician groups in asthma management knowledge: (1) similar (and low) proportions of children had recorded peak expiratory flow readings at baseline, and (2) socioeconomic scores were similar for practices in which small-particle beclo-methasone or fluticasone was prescribed.

Strengths of the present analyses include the large numbers of children, the capture of clinically relevant outcomes over a full year for each child, and the consistency of findings across main analyses and subanalyses. We matched, in addition to sex and age, on all clinically reasonable indicators of baseline asthma severity available in the database. Because the criteria were applied sequentially, for example, first on sex and then on age, children were matched on age within the female and male groupings. We believe that this matching approach is particularly relevant to the study of children, for whom age is an important factor; in addition, it enabled us to examine subgroups of patients who might interact with outcome, for example, the youngest children with the smallest airways. This matching approach differs from the use of propensity score matching, which instead produces 2 cohorts with similar distributions of multiple covariates used to construct the propensity score but not paired patients precisely matched by age and sex.44

We limited the study to a comparison of small-particle beclomethasone (the only small-particle ICS available in the United Kingdom at the time) and fluticasone, which is a larger-particle ICS that is recommended for administration at the same doses as small-particle beclomethasone.37,45,46 Thus, our conclusions are limited to the studied comparisons and are not generalizable to other ICS. Additional studies, including prospective pragmatic clinical trials, are needed to further investigate the effectiveness of other small-particle ICS, such as ciclesonide, for children with asthma.

In conclusion, we found that initiating or stepping up the dose of ICS with a small-particle ICS is significantly more effective

than doing so with a standard size—particle ICS, and has similar effectiveness as add-on LABA for children with asthma treated in primary care practice. The differential effects of small-particle ICS versus standard size—particle ICS were more pronounced in children than in our previous adult studies8,9 and more pronounced in the younger children than in the older children. We cannot identify the factors associated with the beneficial effect but speculate that this is, in part, due to smaller particle size and better ICS deposition in children with smaller airways.

Acknowledgments

We thank Jeremy Brockman for contributions to the data extraction and analysis and Francesca Barion, PhD, for assistance with subanalyses (the latter a former employee of Research in Real Life Ltd [RiRL]). We thank Professor Neil Barnes, MBBS, FRCP, for his contributions to the early discussions when planning the analyses (formerly at London Chest Hospital, Barts, and The London National Health Service Trust, London, UK; no compensation received).

REFERENCES

1. Adams N, Lasserson TJ, Cates CJ, Jones PW. Fluticasone versus beclomethasone or budesonide for chronic asthma in adults and children. Cochrane Database Syst Rev 2007:CD002310.

2. Lasserson TJ, Cates CK, Jones AB, Steele EH, White J. Fluticasone versus HFA-beclomethasone dipropionate for chronic asthma in adults and children. Cochrane Database Syst Rev 2006:CD005309.

3. Kramer S, Rottier BL, Scholten RJ, Boluyt N. Ciclesonide versus other inhaled corticosteroids for chronic asthma in children. Cochrane Database Syst Rev 2013;2:CD010352.

4. Ducharme FM, Ni Chroinin M, Greenstone I, Lasserson TJ. Addition of long-acting beta2-agonists to inhaled steroids versus higher dose inhaled steroids in adults and children with persistent asthma. Cochrane Database Syst Rev 2010: CD005533.

5. Cates CJ, Oleszczuk M, Stovold E, Wieland LS. Safety of regular formoterol or salmeterol in children with asthma: an overview of Cochrane reviews. Cochrane Database Syst Rev 2012;10:CD010005.

6. Lenney W, McKay AJ, Tudur Smith C, Williamson PR, James M, Price D. Management of Asthma in School age Children On Therapy (MASCOT): a randomised, double-blind, placebo-controlled, parallel study of efficacy and safety. Health Technol Assess 2013;17:1-218.

7. Price D, Bateman ED, Chisholm A, Papadopoulos NG, Bosnic-Anticevich S, Pizzichini E, et al. Complementing the randomized controlled trial evidence base: evolution not revolution. Ann Am Thorac Soc 2014;11:S92-8.

8. Colice G, Martin RJ, Israel E, Roche N, Barnes N, Burden A, et al. Asthma outcomes and costs of therapy with extrafine beclomethasone and fluticasone. J Allergy Clin Immunol 2013;132:45-54.e10.

9. Price D, Martin RJ, Barnes N, Dorinsky P, Israel E, Roche N, et al. Prescribing practices and asthma control with hydrofluoroalkane-beclomethasone and flu-ticasone: a real-world observational study. J Allergy Clin Immunol 2010;126: 511-8.e1-10.

10. Cripps A, Riebe M, Schulze M, Woodhouse R. Pharmaceutical transition to non-CFC pressurized metered dose inhalers. Respir Med 2000;94:S3-9.

11. Shi Y, Aledia AS, Tatavoosian AV, Vijayalakshmi S, Galant SP, George SC. Relating small airways to asthma control by using impulse oscillometry in children. J Allergy Clin Immunol 2012;129:671-8.

12. van Schayck CP, Donnell D. The efficacy and safety of QVAR (hydro-fluoroalkane-beclometasone diproprionate extrafine aerosol) in asthma (part 2): clinical experience in children. Int J Clin Pract 2004;58:786-94.

13. Leach CL, Davidson PJ, Hasselquist BE, Boudreau RJ. Influence of particle size and patient dosing technique on lung deposition of HFA-beclomethasone from a metered dose inhaler. J Aerosol Med 2005;18:379-85.

14. Leach CL, Davidson PJ, Hasselquist BE, Boudreau RJ. Lung deposition of hydrofluoroalkane-134a beclomethasone is greater than that of chlorofluoro-carbon fluticasone and chlorofluorocarbon beclomethasone: a cross-over study in healthy volunteers. Chest 2002;122:510-6.

15. Leach CL, Davidson PJ, Boudreau RJ. Improved airway targeting with the CFC-free HFA-beclomethasone metered-dose inhaler compared with CFC-beclomethasone. Eur Respir J 1998;12:1346-53.

16. Devadason SG, Huang T, Walker S, Troedson R, Le Souef PN. Distribution of technetium-99m-labelled QVAR delivered using an Autohaler device in children. Eur Respir J 2003;21:1007-11.

17. Roller CM, Zhang G, Troedson RG, Leach CL, Le Souef PN, Devadason SG. Spacer inhalation technique and deposition of extrafine aerosol in asthmatic children. Eur Respir J 2007;29:299-306.

18. Costa-Katz CL, Livnat G, Hakim F, Vilozni D, Bentur Y, Bentur L. The effect of beclomethasone dipropionate in ultrafine particles on bronchial hyper-reactivity in young children. Acta Paediatr 2012;101:e219-24.

19. Eid N, Morton R. Lung function changes in asthmatic children treated with HFA-BDP. Pediatr Pulmonol 2011;46:837-41.

20. Robroeks CM, van de Kant KD, van Vliet D, Kester AD, Hendriks HJ, Damoiseaux JG, et al. Comparison of the anti-inflammatory effects of extra-fine hydrofluoroalkane-beclomethasone vs fluticasone dry powder inhaler on exhaled inflammatory markers in childhood asthma. Ann Allergy Asthma Immunol 2008;100:601-7.

21. van Aalderen WM, Price D, De Baets FM, Price J. Beclometasone dipropionate extrafine aerosol versus fluticasone propionate in children with asthma. Respir Med 2007;101:1585-93.

22. Castro-Rodriguez JA, Rodrigo GJ. A systematic review of long-acting beta2-agonists versus higher doses of inhaled corticosteroids in asthma. Pediatrics 2012;130:e650-7.

23. Lemanske RF Jr, Mauger DT, Sorkness CA, Jackson DJ, Boehmer SJ, Martinez FD, et al. Step-up therapy for children with uncontrolled asthma receiving inhaled corticosteroids. N Engl J Med 2010;362:975-85.

24. Clinical Practice Research Datalink. Available from: http://www.cprd.com/ home/. Accessed April 10, 2014.

25. Optimum Patient Care Research Database (OPCRD). Available from: http://www. optimumpatientcare.org/Html_Docs/OPCRD.html. Accessed April 10, 2014.

26. Barnes N, Price D, Colice G, Chisholm A, Dorinsky P, Hillyer EV, et al. Asthma control with extrafine-particle hydrofluoroalkane-beclometasone vs. large-particle chlorofluorocarbon-beclometasone: a real-world observational study. Clin Exp Allergy 2011;41:1521-32.

27. Price D, Hillyer EV, van der Molen T. Efficacy versus effectiveness trials: informing guidelines for asthma management. Curr Opin Allergy Clin Immunol 2013;13:50-7.

28. Akinbami LJ, Sullivan SD, Campbell JD, Grundmeier RW, Hartert TV, Lee TA, et al. Asthma outcomes: healthcare utilization and costs. J Allergy Clin Immunol 2012;129:S49-64.

29. Kozyrskyj AL, Dahl ME, Ungar WJ, Becker AB, Law BJ. Antibiotic treatment of wheezing in children with asthma: what is the practice? Pediatrics 2006;117: e1104-10.

30. Reddel HK, Taylor DR, Bateman ED, Boulet LP, Boushey HA, Busse WW, et al. An official American Thoracic Society/European Respiratory Society statement: asthma control and exacerbations: standardizing endpoints for clinical asthma trials and clinical practice. Am J Respir Crit Care Med 2009;180: 59-99.

31. Research in Real Life: standard operating procedures. Available from: http://www. optimumpatientcare.org/Docs/SOP%20Observational%20Database%20Studies. pdf. Accessed April 10, 2014.

32. Benjamini Y, Liu W. A distribution-free multiple test procedure that controls the false discovery rate. RP-SOR-99-3. Tel Aviv, Israel: Department of Statistics and O.R., Tel Aviv University; 1999.

33. Williams LK, Peterson EL, Wells K, Ahmedani BK, Kumar R, Burchard EG, et al. Quantifying the proportion of severe asthma exacerbations attributable to inhaled corticosteroid nonadherence. J Allergy Clin Immunol 2011;128: 1185-1191.e2.

34. Ivanova JI, Bergman R, Birnbaum HG, Colice GL, Silverman RA, McLaurin K. Effect of asthma exacerbations on health care costs among asthmatic patients with moderate and severe persistent asthma. J Allergy Clin Immunol 2012;129: 1229-35.

35. Amirav I, Newhouse MT. Deposition of small particles in the developing lung. Paediatr Respir Rev 2012;13:73-8.

36. Bacharier LB, Boner A, Carlsen KH, Eigenmann PA, Frischer T, Gotz M, et al. Diagnosis and treatment of asthma in childhood: a PRACTALL consensus report. Allergy 2008;63:5-34.

37. British Thoracic Society (BTS), Scottish Intercollegiate Guidelines Network (SIGN). British Guideline on the Management of Asthma, May 2008, revised Jan 2012. Available from: http://www.sign.ac.uk/guidelines/fulltext/101/index. html. Accessed April 10, 2014.

38. Chowdhury BA, Dal Pan G. The FDA and safe use of long-acting beta-agonists in the treatment of asthma. N Engl J Med 2010;362:1169-71.

39. Global Initiative for Asthma. Global Strategy for Asthma Management and Prevention, revised 2014. Available from: www.ginasthma.org. Accessed June 10, 2014.

40. Martinez FD, Chinchilli VM, Morgan WJ, Boehmer SJ, Lemanske RF Jr, Mauger DT, et al. Use of beclomethasone dipropionate as rescue treatment for children with mild persistent asthma (TREXA): a randomised, double-blind, placebo-controlled trial. Lancet 2011;377:650-7.

41. Allen DB. Effects of inhaled steroids on growth, bone metabolism and adrenal function. Expert Rev Respir Med 2007;1:65-74.

42. Pruteanu AI, Chauhan BF, Zhang L, Prietsch SO, Ducharme FM. Inhaled corticosteroids in children with persistent asthma: dose-response effects on growth. Cochrane Database Syst Rev 2014;7:CD009878.

43. Kelly HW, Sternberg AL, Lescher R, Fuhlbrigge AL, Williams P, Zeiger RS, et al. Effect of inhaled glucocorticoids in childhood on adult height. N Engl J Med 2012;367:904-12.

44. Joffe MM, Rosenbaum PR. Invited commentary: propensity score. Am J Epidemiol 1999;150:327-33.

45. National Asthma Education and Prevention Program. Expert Panel Report 3: Guidelines for the Diagnosis and Management of Asthma. Available from: http:// www.nhlbi.nih.gov/guidelines/asthma/asthgdln.pdf. Accessed July 31, 2014.

46. Switching to CFC-free beclometasone for asthma. Drug Ther Bull 2008;46: 46-8.

METHODS

Our hypothesis for analysis 1 was that asthma-related outcomes for children, as for adults, would be similar or better with small-particle ICS, even when prescribed at lower doses, than with standard size—particle ICS. Our hypothesis for analysis 2 was that increasing the dose of small-particle ICS would be an effective alternative for children with persistent asthma who require more treatment than their initial ICS therapy.

The study was done to standards suggested for observational studies, including an independent steering committee, use of an a priori analysis plan, and well-maintained and monitored study databases.E1-E3 The members of the steering committee are practicing physicians working in different areas of respiratory medicine who provide varied perspectives and guidance to Research in Real Life (RiRL), functioning in complete intellectual independence. The group comprises adult practitioners and pediatricians because multiple studies are being conducted in various populations. The steering committee members are not paid for their participation but are reimbursed by RiRL for travel to annual or biannual meetings to discuss multiple ongoing studies that are being conducted by RiRL. The design of the study, analyses and subanalyses, and manuscript represent consensus work of this committee, all of whom are authors on the article together with the research team at RiRL.

Although clinical trial registration was not required because this was a noninterventional study, each of the analyses was registered, as a separate study, with registered clinical trials registry numbers NCT01141439 and NCT01697722.

Data sources

The data sets we used contain anonymized primary care medical records, data entered by health care providers during patient consultations in the form of Read codes. The General Practice Research Database is a large, well-regarded, and well-validated primary care database containing anonymized medical record data from subscribing practices throughout the United Kingdom and used frequently for pharmacoepidemiological researchE4-E7; it now forms part of the National Health Service Clinical Practice Research Datalink, the English National Health Service observational data and interventional research serv-ice.E8,E9 The Optimum Patient Care Research Database is a newer primary care database, established to rigorous standards, that contains anonymous patient data, including both medical records and patient-completed respiratory questionnaire results, from more than 300 primary care practices subscribing to the Optimum Patient Care respiratory review service.E10 The value of this database information specifically for studying respiratory diseases has been documented in numerous high-quality pub-lications.E4-E12

The available data for this study spanned 1997 through January 2011 and was approved for use by the General Practice Research Database Independent Scientific Advisory Committee and, for the Optimum Patient Care Research Database, by the Trent Multi Centre Research Ethics Committee. The study protocol was approved by the Anonymised Data Ethics Protocols and Transparency Committee, the independent scientific advisory committee for Optimum Patient Care.

Patients and outcomes

Allowed ICS therapies during the baseline year for step-up cohorts included beclomethasone, budesonide, or fluticasone

administered by pMDI or breath-actuated inhaler. Small-particle beclomethasone dipropionate hydrofluoroalkane (HFA) (QVAR; Teva Pharmaceuticals) is labeled for administration to both adults and children at half the daily dose of chlorofluorocarbon (CFC) or larger-particle HFA-beclomethasone (Clenil Modulite; Chiesi Ltd, Highfield, Cheadle, UK) and at the same dose as HFA- or CFC-fluticasone propionate.E13,E14 Thus, we standardized ICS doses to small-particle beclomethasone for the analyses, using a 1:1 ratio for small-particle beclomethasone and fluticasone, and for both the latter a 1:2 dose ratio relative to budesonide and larger-particle CFC- or HFA-beclomethasone. Heights and weights were not routinely available in the data sets for all children.

Statistical analysis

For the small-particle ICS versus standard size—particle ICS comparison, the coprimary outcomes were risk-domain asthma control and the total number of severe exacerbations; for the small-particle ICS versus add-on LABA comparisons, the primary outcome was risk-domain asthma control (see Table I in the main article). We used the false-discovery rate controlling procedure of Benjamini and LiuE15 to account for multiple comparisons in the primary outcomes. This procedure modifies the threshold (0.05) for each comparison; with 2 coprimary outcomes, the "most significant" result is required to achieve significance at the 2.5% level (P = .025) while the second result is required to achieve significance at the 5% level (P = .05).

For the patient population initiating ICS, we conducted several subanalyses to further examine factors that could be influencing the outcomes of therapy. We compared results with small-particle ICS versus standard size—particle ICS by age (5-6 years and 7-11 years); with use of a spacer for all children receiving standard size—particle ICS; for children with no rhinitis diagnosis or therapy; and according to baseline asthma severity (0 or >1 exacerbation during the baseline year); the latter 2 subanalyses were conducted with unmatched patient data to increase sample sizes. Patient numbers in the small-particle ICS and standard size—particle ICS step-up populations were insufficient to warrant subanalysis.

For the spacer subanalysis, we included all children in the standard size—particle ICS cohort with a spacer prescribed in baseline and/or outcome year (total of 465 children). We then compared these 465 children with their matches in the small-particle ICS cohort; of that subgroup of matched children in the small-particle ICS cohort, 69% were prescribed a spacer. (Overall, however, approximately equal numbers of the 797 children in each of the initiation cohorts were prescribed a spacer.)

For the step-up comparisons of small-particle ICS versus addon LABA, we calculated the total b2-agonist hours of coverage (SABA + LABA). We defined albuterol 200 mg and terbutaline 500 mg as 2 puffs of SABA lasting 4 hours, whereas we defined b2-agonist coverage with LABA 2 puffs via pMDI or 1 puff via dry powder inhaler as lasting 12 hours.

Different matching ratios were evaluated, and the 1:1 ratio for the small-particle ICS versus standard size—particle ICS cohorts and the ICS dose step-up versus ICS/LABA combination cohorts and the 1:2 ratio for the ICS dose step-up versus separate ICS + LABA cohorts were chosen to maximize patient numbers and thus statistical power for the comparisons.

Matching criteria were applied sequentially as ordered below:

1. For small-particle ICS versus standard size—particle ICS initiation cohorts (analysis 1):

a. Sex

b. Age

c. Mean SABA daily dose during the baseline year (0, 1-100,

101-200, >200 mg/d)

d. Number of severe exacerbations during the baseline year (0, 1, >2)

e. Number of asthma consultations without an oral cortico-steroid prescription during the baseline year (0, 1, >2)

f. Year of index prescription (±4 years)

2. For small-particle ICS versus standard size—particle ICS step-up cohorts (analysis 1):

a. Sex

b. Age

c. Last ICS dose prescribed before the index prescription (1100,101-300, >300 mg/d)

d. Number of severe exacerbations during the baseline year (0, 1, 2, >3)

e. Risk-domain asthma control status during the baseline year (controlled/uncontrolled)

f. Mean SABA daily dose during the baseline year (0, 1-200, > 200 mg/d)

g. Number of asthma consultations without an oral cortico-steroid prescription during the baseline year (0, 1, >2)

h. Year of index prescription (±4 years)

3. For small-particle ICS versus add-on LABA cohorts (analysis 2):

a. Sex

b. Age

c. Last ICS dose prescribed before the index prescription (150, 51-100, 101-200, 201-300, 301-400, >400 mg/d)

d. Mean SABA daily dose during the baseline year (0, 1-200,

201-400, >400 mg/d)

e. Number of asthma consultations without an oral cortico-steroid prescription during the baseline year (0, 1, >2)

f. Risk-domain asthma control status during the baseline year (controlled/uncontrolled)

Potential confounding factors considered: Previous research in respiratory disease has identified a range of potential confounders that can influence study outcomes. These include a range of demographic, disease severity, treatment, and comorbid factors.

Variables that differed between treatment cohorts at P < .10 were considered potential confounding factors, as well as baseline variables predictive of outcomes in multivariate analyses at P < .05. Spearman correlation coefficients were calculated between all potential confounders to determine strengths of linear relationships between variables. The correlation coefficients were considered, in conjunction with clinical interpretation, to identify pairings of variables that could present collinearity issues at the modeling stage. Scatter plots and error bars were used, if necessary, to further investigate relationships.

We fit the regression models separately for each subgroup. However, because baseline differences across treatment arms may have been different between full cohorts and subgroups, the entire modeling process was repeated separately for each subgroup to identify the unique adjustments needed for that subgroup.

Potential confounders examined, where available, at (or closest

to) the relevant index date for all children:

• Age

• Sex

• Height

• Weight

• Body mass index

• Lung function as percent predicted peak flow readings before the index date

Potential confounders examined regardless of when they

occurred relative to the index date:

• Date of first asthma, other respiratory and allergy-related, diagnosis (where known)

• Presence/absence of comorbid rhinitis (diagnosis ever and/or prescriptions for rhinitis therapy in the baseline/outcome year)

• Presence/absence of comorbid eczema (diagnosis ever and/or prescriptions for eczema therapy in the baseline/outcome year)

• Presence/absence of preschool wheeze (diagnosis ever)

• Presence/absence of preschool asthma (diagnosis ever and/or prescriptions for asthma therapy in the baseline/outcome year)

• Presence of gastroesophageal reflux disease (diagnosis ever and/ or prescriptions for gastroesophageal reflux disease therapy in the baseline/outcome year)

• Presence of cardiac disease (diagnosis ever and/or prescriptions for cardiac drugs in the baseline/outcome year)

Potential confounders examined in the baseline year before the

index date:

• Where rhinitis is present, use of nasal corticosteroids for treatment

• Other important unrelated comorbidities were expressed using the Charlson comorbidity index

• Number of asthma consultations that did not result in a prescription for an oral corticosteroid

• Number of hospital outpatient attendances where asthma was recorded as the reason for referral

• Number of hospitalizations for asthma or possibly respiratory related (a nonspecific hospitalization code and an asthma/ respiratory code within a 1-week window)

• Number of asthma-related emergency department visits

• Number of prescriptions for any antibiotic where the reason for the prescription was lower respiratory tract infection

• Other medications, number of prescriptions for the following in the year before the index date:

o Acetaminophen

o Nonsteroidal anti-inflammatory drugs

• Number of prescriptions for any respiratory therapy (split by number of prescriptions for each)

• Number of asthma exacerbations

• Number of acute courses of oral corticosteroids

• Number of SABA prescriptions and average daily SABA dose received (calculated on the basis of the total combined dose of refilled prescriptions and averaged over 365 days)

• Average ICS daily dose during the baseline year (calculated on the basis of total combined dose of refilled prescriptions and averaged over 365 days) (step-up patients only)

• Last ICS dose prescribed before the index date (step-up patients only)

• Adherence to ICS therapy (step-up patients only)

• Spacer use/prescription

• Medication possession ratio (step-up patients only)

• Controller-to-reliever therapy ratio (step-up-population only)

• Oral candidiasis

RESULTS

Analysis 1: Small-particle ICS versus standard size—particle ICS as first-line or step-up ICS therapy Unmatched versus matched baselines and results. For

analysis 1, we found that unmatched patients prescribed small-particle ICS, as compared with standard size—particle ICS, tended to be slightly older and to have later index prescription dates; moreover, in the step-up population, those prescribed small-particle ICS had less severe asthma, with a smaller percentage experiencing multiple exacerbations during baseline (eg, 6% had >2 exacerbations vs 12% in the unmatched standard size—particle ICS cohort; other data not shown). After matching, both treatment cohorts tended to have less severe asthma than did the unmatched initiation population (eg, 82% with risk-domain asthma control vs 77% of the unmatched initiation population at baseline; other data not shown).

The pattern of unmatched results during the outcome year followed patterns for the matched cohorts, namely, better outcomes for small-particle ICS (data not shown).

Initiation population. Coprimary outcomes remained significant after having controlled the false-discovery rate at the 0.05

level.

Initiation population subanalyses. Matched cohort subanalyses by age group found that outcomes with small-particle ICS remained significantly better (compared with standard size—particle ICS) for 5- to 6-year-old children initiating ICS (n = 286 per treatment cohort) but were comparable with standard size—particle ICS for 7- to 11-year-olds (n = 511 per cohort). For 5- to 6-year-olds in the small-particle ICS cohort, the adjOR for achieving risk-domain asthma control was 2.00 (95% CI, 1.21-3.29) and for achieving overall control was 1.66 (95% CI, 1.11-2.46) compared with standard size—particle ICS; the corresponding odds ratios for 7- to 11-year-olds were 1.25 (95% CI, 0.85-1.83) and 1.19 (95% CI, 0.91-1.57). The severe exacerbation adjRR for the small-particle ICS cohort relative to the standard size—particle ICS cohort was 0.51 (0.23-1.10) and the clinical exacerbation rate ratio was 0.48 (0.30-0.77) for 5- to 6-year-olds; the corresponding adjRRs for 7- to 11-year-olds were 0.76 (0.45-1.28) and 0.87 (0.60-1.25), respectively.

Baseline asthma severity was greater among 5- to 6-year-olds initiating ICS (data not shown); therefore, we hypothesized that better outcomes with small-particle ICS in this cohort were related to baseline severity rather than age. However, the subanalysis of outcomes split by asthma severity (0 or > 1 baseline exacerbations) indicated better results for children with less severe asthma (Table E3). Further exploration of results for risk-domain asthma control by outcome-year ICS dose indicated greater variation in results for 5- to 6-year-olds (compared with 7-11-year-olds) and greater variation in the standard size—particle ICS cohort (Figure E6).

Results of the subanalysis including children with no rhinitis diagnosis or prescription for nasal spray (unmatched cohorts) were reflective of the main analysis results: the odds ratios remained consistent, although with smaller patient numbers the

odds ratio for risk-domain asthma control became statistically nonsignificant (Table E3).

Analysis 2: Unmatched versus matched baselines and results

For analysis 2, the unmatched ICS step-up cohort, as compared with both add-on LABA cohorts, received a lower baseline ICS dose and included more children with asthma control and fewer children with severe exacerbations during baseline (data not shown). After matching, the matched add-on LABA cohorts had (similar to the ICS step-up cohort) less severe asthma than did the unmatched cohorts (67%-70% with baseline asthma control vs 58% of the unmatched add-on LABA cohorts).

During the outcome year, a significantly greater percentage of unmatched patients in the ICS step-up cohort experienced asthma control and treatment stability as compared with both add-on LABA cohorts (data not shown). Unadjusted exacerbation results were similar for unmatched ICS step-up and ICS/ LABA combination cohorts but better for the unmatched ICS step-up than for the separate ICS + LABA cohort.

Analysis 2: ICS dose step-up versus add-on LABA as separate inhaler

Several minor differences remained at baseline after 1:2 matching of the small-particle ICS dose step-up (n = 276) and separate ICS + LABA (n = 552) cohorts (Tables E4 and E6). Children in the ICS dose step-up cohort recorded fewer exacerbations during the baseline year than did children in the separate ICS + LABA cohort (Table E6).

Median (interquartile range) ICS doses prescribed at the index date were 200 mg/d (200-400 mg/d) for the ICS dose step-up and 100 mg/d (100-200 mg/d) for the separate ICS + LABA cohort (P < .001 for the comparison). Change in therapy was significantly more common, and the total b2-agonist coverage (LABA plus SABA) was significantly greater in the separate ICS + LABA cohort (Table E5). Figure E7, B, depicts coverage by time period for each cohort.

During the outcome year, children stepping up their ICS therapy with small-particle ICS had significantly higher odds of treatment stability, after adjustments for residual confounding factors, than did children remaining on the same dose but adding a separate LABA; odds of asthma control and exacerbation rates were comparable between cohorts (Figure E8; Table E7).

REFERENCES

E1. von Elm E, Altman DG, Egger M, Pocock SJ, Gotzsche PC, Vandenbroucke JP. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. Lancet 2007;370:1453-7.

E2. Vandenbroucke JP, von Elm E, Altman DG, Gotzsche PC, Mulrow CD, Pocock SJ, et al. Strengthening the Reporting of Observational Studies in Epidemiology (STROBE): explanation and elaboration. PLoS Med 2007;4: e297.

E3. Price D, Hillyer EV, van der Molen T. Efficacy versus effectiveness trials: informing guidelines for asthma management. Curr Opin Allergy Clin Immunol 2013;13:50-7.

E4. Khan NF, Harrison SE, Rose PW. Validity of diagnostic coding within the General Practice Research Database: a systematic review. Br J Gen Pract 2010; 60:e128-36.

E5. Wood L, Martinez C. The general practice research database: role in phar-macovigilance. Drug Saf 2004;27:871-81.

E6. Wong IC, Murray ML. The potential of UK clinical databases in enhancing paediatric medication research. Br J Clin Pharmacol 2005;59:750-5.

E7. Price D, Martin RJ, Barnes N, Dorinsky P, Israel E, Roche N, et al. Prescribing practices and asthma control with hydrofluoroalkane-beclomethasone and fluticasone: a real-world observational study. J Allergy Clin Immunol 2010; 126:511-8.e1-10.

E8. Bhaskaran K, Forbes HJ, Douglas I, Leon DA, Smeeth L. Representativeness and optimal use of body mass index (BMI) in the UK Clinical Practice Research Datalink (CPRD). BMJ Open 2013;3:e003389.

E9. Clinical Practice Research Datalink. Available from: http://www.cprd.com/ home/. Accessed April 10, 2014.

E10. Optimum Patient Care Research Database (OPCRD). Available from: http:// www.optimumpatientcare.org/Html_Docs/OPCRD.html. Accessed April 10, 2014.

E11. Hansell A, Hollowell J, Nichols T, McNiece R, Strachan D. Use of the General Practice Research Database (GPRD) for respiratory epidemiology: a

comparison with the 4th Morbidity Survey in General Practice (MSGP4). Thorax 1999;54:413-9.

E12. Jones RCM, Price D, Ryan D, Sims EJ, von Ziegenweidt J, Mascarenhas L, et al. Opportunities to diagnose chronic obstructive pulmonary disease in routine care in the UK: a retrospective study of a clinical cohort. Lancet Respir Med 2014;2:267-76.

E13. Switching to CFC-free beclometasone for asthma. Drug Ther Bull 2008;46: 46-8.

E14. British Thoracic Society (BTS), Scottish Intercollegiate Guidelines Network (SIGN). British Guideline on the Management of Asthma, May 2008, revised Jan 2012. Available from: http://www.sign.ac.uk/guidelines/fulltext/101/ index.html. Accessed April 10, 2014.

E15. Benjamini Y, Liu W. A distribution-free multiple test procedure that controls the false discovery rate. RP-SOR-99-3. Tel Aviv, Israel: Department of Statistics and O.R., Tel Aviv University; 1999.

11.e5 VAN AALDEREN ET AL j allergy clin Immunol pract

month 2015

FIGURE E1. Initiation population: patient selection and matching (analysis 1). Patients in the 2 treatment cohorts were matched on clinically and demographically significant characteristics. FDC, Fixed-dose combination ICS/LABA; FP, fluticasone propionate; GPRD, General Practice Research Database; OPCRD, Optimum Patient Care Research Database; SP-BDP, small-particle beclomethasone dipropionate.

Respiratory patients in the GPRD & OPCRD

n=1,540,877

Inclusion Criteria:

Patients prescribed a >50% increase in ICS dose as SP-BDP or fluticasone by pMDI

Patients prescribed any ICS inhalers n=706,913

Excluded:

-no prior prescription (n=29,341) or no current prescription (n=548,082) for SP-BDP or fluticasone by pMDI -prescribed lower, same, or <50% increase in

ICS dose (n=92,962) -multiple prescriptions on the index date (n=5973)

Patients prescribed >50% dose increase of small-particle beclomethasone and/or fluticasone as single script: n=30,555

Matching criteria applied:

-same sex -same age

-last ICS dose before index date (1-100, 101-300, >300 pg/day) -baseline year no. severe

exacerbations (0, 1, 2, >3) -risk-domain asthma control status

(controlled/uncontrolled) -baseline mean SABA dose (0,

1-200, >200 pg/day) -baseline no. asthma

consultations without oral steroid prescription (0, 1, >2) -year of index date ±4 years

Eligible patients aged 5-11 stepping up ICS therapy as:

Small-particle beclomethasone n=294 Fluticasone n=1647

Excluded:

-not aged 5-11 years at index date

(n=27,223) -not registered at practice for study

period (n=435) -prescribed FDC therapy during

baseline or on index date (n=14) -no evidence of asthma (n=69) -maintenance oral steroids during

baseline (n=89) -prescribed ICS on multiple

prescriptions (n=575) -duplicated in GPRD or OPCRD (n=209)

Double matches lost on randomization 1:1:

Small-particle beclomethasone n=11 Fluticasone n=335

Total matched patients included

1:1 uniquely matched pairs: Small-particle beclomethasone n=206 Fluticasone n=206

FIGURE E2. Step-up population: patient selection and matching (analysis 1). Patients in the 2 treatment cohorts were matched on clinically and demographically significant characteristics. FDC, Fixed-dose combination ICS/LABA; FP, fluticasone propionate; GPRD, General Practice Research Database; OPCRD, Optimum Patient Care Research Database; Script, prescription; SP-BDP, small-particle beclomethasone dipropionate.

11.e7 VAN AALDEREN ET AL j allergy clin Immunol pract

month 2015

FIGURE E3. Small-particle ICS dose step-up versus add-on LABA in ICS/LABA combination inhaler: patient selection and matching (analysis 2). Patients in the 2 treatment cohorts were matched on clinically and demographically significant characteristics. BAI, Breath-actuated inhaler; COPD, chronic obstructive pulmonary disease; DPI, dry powder inhaler; FDC, fixed-dose combination ICS/LABA; GPRD, General Practice Research Database; IPD, index prescription date; OCS, oral corticosteroid; OPCRD, Optimum Patient Care Research Database; Rx, therapy; Script, prescription; SP-BDP, small-particle beclomethasone dipropionate.

FIGURE E4. Small-particle ICS dose step-up versus add-on LABA to ICS in separate inhalers: patient selection and matching (analysis 2). Patients in the 2 treatment cohorts were matched on clinically and demographically significant characteristics. BAI, Breath-actuated inhaler; COPD, chronic obstructive pulmonary disease; DPI, dry powder inhaler; FDC, fixed-dose combination ICS/LABA; GPRD, General Practice Research Database; IPD, index prescription date; OCS, oral corticosteroid; OPCRD, Optimum Patient Care Research Database; Rx, therapy; Script, prescription; SP-BDP, small-particle beclomethasone dipropionate.

FIGURE E5. ICS dose for matched cohorts as prescribed on the index date for (A) the initiation population and (B) the step-up population. Differences in prescribed doses between small-particle ICS and standard size—particle ICS cohorts were statistically significant (P < .001).

FIGURE E6. Probability of achieving risk-domain asthma control during the outcome year for children in the initiation population. Probability categorized by (A) age group and (B) treatment cohort, categorized by ICS daily dose exposure (ie, prescribed ICS dose divided by 365). Standard SP, Standard size—particle.

FIGURE E7. Total b2-agonist load comparing small-particle ICS dose step-up versus add-on LABA to ICS in (A) fixed-dose combination inhaler with ICS and (B) separate inhalers. Hours covered per 24 hours. FDC, fixed-dose combination.

FIGURE E8. AdjORs and adjRRs comparing step-up with increased dose of small-particle ICS versus add-on LABA as separate inhaler with ICS. (For details of confounding factors examined, please see above in Online Repository.) LRTI, Lower respiratory tract infection; NSAIDs, nonsteroidal anti-inflammatory drugs. Adjusted for the following confounders: *Number of asthma consultations, number of acute courses of oral corticosteroids, number of antibiotic prescriptions for LRTI. t Asthma diagnosis, prescriptions for NSAIDs, number of acute courses of oral corticosteroids, number of antibiotic prescriptions for LRTI, controller-to-reliever ratio. zfNumber of asthma consultations, number of acute courses of oral corticosteroids, controller-to-reliever ratio. §Asthma diagnosis, number of primary care consultations, number of acute courses of oral corticosteroids, number of antibiotic prescriptions for LRTI, inpatient admission for lower respiratory tract-related reasons. #Controller-to-reliever ratio, definite plus probable asthma-related hospitalizations.

11.e13 VAN AALDEREN ETAL j allergy clin Immunol pract

month 2015

TABLE E1. More baseline characteristics of children in the small-particle ICS and standard SP ICS cohorts (analysis 1)

Initiation population Step-up population

Characteristic Small-particle ICS (n = 797) Standard SP ICS (n = 797) P value* Small-particle ICS (n = 206) Standard SP ICS (n = 206) P value*

Socioeconomic scoref available Mean ± SD Median (IQR) BMI (kg/m2), mean ± SD| Years since first asthma code, median (IQR) Database Read code for asthma Patients with PEF reading at baseline 778 (97.6) 20.8 ± 14.9 16.2 (10.2-27.0) 17.4 ± 3.4 0.6 (0.0-3.8) 786 (98.6) 114 (14.3) 786 (98.6) 21.7 ± 16.2 16.2 (10.5-27.2) 17.4 ± 3.7 0.6 (0.0-4.0) 785 (98.5) 143 (17.9) 203 (98.5) AI. 11 O _l_ 1 (Crt 205 (99.5) 21.0 ± 15.2 16.3 (10.5-25.0) 16.8 ± 2.9 3.7 (1.7-5.3) 203 (98.5) 67 (32.5) .49 1.0 .075 .71

.43 22.2 ± 16.0 17.1 (10.4-29.3)

.OJ .46 1 / .\j ~ / 3.0 (1.4-5.3)

.OJ JLKJJL ^O.l) 79 (38.3)

Mean ± SD %predicted PEF >1 prescription in the previous 12 mo Nonsteroidal anti-inflammatory drug 101 ± 26.4 44 (5.5) 101 ± 29.2 49 (6.1) .82 .60 115 ± 27.8 14 (6.8) 102 ± 28.1 13 (6.3) .061 .85

Acetaminophen Clinical exacerbations 0 1 >2 Asthma prescriptions, median (IQR) All primary care consultations, median (IQR) GP consultation for LRTI requiring antibiotic 0 99 (12.4) 132 (16.6) .017 38 (18.4) 39 (18.9) .90

659 (82.7) 105 (13.2) 33 (4.1) 2 (0-3) 4 (2-6) 661 (82.9) 106 (13.3) 30 (3.8) 2 (0-4) 4 (2-6) .75 161 (78.2) 163 (79.1) 32 (15.5) 11 (5.3) 9 (5-15) 5 (3-9) .80

JJ ! ) 10 (4.9) <.001 .03

.28 U y-i- LL) 4 (3-8)

698 (87.6) 706 (88.6) .53 187 (90.8) 188 (91.3) .76

>1 99 (12.4) 91 (11.4) 19 (9.2) 18 (8.7)

BMI, Body mass index; IQR, interquartile range; LRTI, lower respiratory tract infection; PEF, peak expiratory flow; standard SP, standard size—particle. Data are n (%) unless otherwise noted.

*Matched cohorts were compared using conditional logistic regression.

fThe socioeconomic status score for practices used in the analyses was that assigned by local postcodes using the Index of Multiple Deprivation as a proxy measure (https:// www.gov.uk/government/collections/english-indices-of-deprivation). A high score indicates a high level of deprivation, whereas the least deprived areas have a low score. ^Recorded BMI data were available for 287 (36%) and 286 (36%) children in small-particle ICS and standard SP ICS initiation cohorts, respectively, and for 91 (44%) and 103 (50%) children in small-particle ICS and standard SP ICS step-up cohorts, respectively.

TABLE E2. Asthma therapy during the outcome year for matched cohorts prescribed small-particle ICS or standard SP ICS for first-line or step-up therapy (analysis 1)

Outcome Initiation population Step-up population

Small-particle ICS (n = 797) Standard SP ICS (n = 797) P value* Small-particle ICS (n = 206) Standard SP ICS (n = 206) P value*

Spacer device prescribed 397 (49.8) 340 (42.7) .004 95 (46.1) 73 (35.4) .034

Two or more ICS inhalers 2 (1-4) 531 (66.6) 3 (2-6) 621 (77.9) < 0.001 <0.001 4 (3-7) 190 (92.2) 6 (4-9) 201 (98.6) <.001 <.001

Increase in ICS dose 57 (7.2) 157 (19.7) 91 (11.4) <.001 .004 30 (14.6) 12 (5.8) 31 (15.0) 8 (3.9) .88 .35

Fluticasone-salmeterol 46 (5.8) 13 (1.6) 87 (10.9) 35 (4.4) <.001 .002 22 (10.7) 5 (2.4) 16 (7.8) .34 .023

LABA 2 (U.3) 19 (2.4) 1 (0.1) 39 (4.9) .57 .010 3 (1.5) 10 (4.9) 1 (U.iJ 9 (4.4) .34 .81

Leukotriene receptor antagonist§ 20 (2.5) 26 (3.3) .37 8 (3.9) 5 (2.4) .37

Standard SP, Standard size—particle. Data are n (%) unless otherwise noted. *Conditional logistic regression.

fNumber of ICS inhalers prescribed during the outcome year in addition to the index prescription. ICS inhaler durations ranged from 30 d (eg, for Flixotide) to 50 d (eg, for QVAR).

^Change in therapy and additional therapy could be at any time during the outcome year, and children could have had more than 1. §Montelukast is the only leukotriene receptor antagonist licensed in the United Kingdom.

TABLE E3. Summary of matched spacer subanalysis and unmatched full analysis and subanalyses for the initiation population: small-particle ICS or standard SP ICS comparison

AdjOR/RR (95% CI) for small-particle ICS relative to standard SP ICS (1.00) Matched cohort analysis Unmatched analyses

Initiation population Spacer subanalysis* Full unmatched No rhinitis diagnosis or therapy 0 Exacerbations Exacerbations

Sample size

Small-particle ICS 465 1058 858 911 147

Standard SP ICS 465* 1706 1281 1487 219

Risk-domain asthma control 1.81 (1.24-2.65)f 1.32 (1.05-1.65)f 1.31 (0.99-1.72)f 1.35 (1.04-1.75)f 1.03 (0.64-1.64)f

Overall asthma control 1.49 (1.12-2.00)! 1.28 (1.08-1.51)! 1.25 (1.02-1.53)! 1.27 (1.06-1.53)! 1.22 (0.79-1.89)!

Treatment stability 1.80 (1.32-2.46)§ 1.80 (1.50-2.16)§ 1.68 (1.36-2.06)§ 1.97 (1.61-2.41)§ 1.21 (0.79-1.86)§

Severe exacerbations 0.53 (0.29-0.98)# 0.63 (0.47-0.84)# 0.67 (0.48-0.94)# 0.52 (0.34-0.77)# 0.88 (0.53-1.45)#

Clinical exacerbations 0.55 (0.38-0.79)- 0.77 (0.62-0.95)- 0.71 (0.55-0.91)- 0.71 (0.55-0.93)- 0.85 (0.54-1.32)-

LRTI hospitalizations 0.71 (0.23-2.25)ff 0.90 (0.43-1.88)ff 1.06 (0.49-2.27)ff — —

IQR, Interquartile range; LRTI, lower respiratory tract infection; standard SP, standard size—particle.

*All children in the standard SP ICS cohort and 319 (68.6%) children in the small-particle ICS cohort were prescribed a spacer device during baseline and/or outcome year. Spacer subanalysis adjusted for:

fGP consultation for LRTI requiring antibiotics. ^Number of prescriptions for asthma or allergies. §Baseline LABA use and number of non—asthma-related consultations. #Mean SABA daily dose, outpatient department attendance for lower respiratory reasons, and year of index date. -GP consultation for LRTI requiring antibiotics and year of index date. ttNo significant effects. Full unmatched analysis adjusted for:

tMean SABA daily dose, age, number of severe exacerbations. zMean SABA daily dose, sex, asthma control status, rhinitis diagnosis and/or therapy. xAsthma control status, mean SABA daily dose, use of LABA. #Mean SABA daily dose, number of severe exacerbations. Number of severe exacerbations, oral candidiasis, age, rhinitis diagnosis and/ or therapy. ttInpatient admissions for lower respiratory reasons (including vague). No rhinitis subanalysis adjusted for:

tAge, year of index date, mean SABA daily dose, number of acute oral corticosteroid courses. zMean SABA daily dose, year of index date, number of acute oral corticosteroid courses. §Mean SABA daily dose, LABA use, number of acute oral corticosteroid courses, number of non—asthma-related consultations. #Year of index date, LABA use, mean SABA daily dose, number of acute oral corticosteroid courses. Age, year of index date, mean SABA daily dose, number of acute oral corticosteroid courses, number of consultations for LRTI. ttYear of first asthma coding, mean SABA daily dose, baseline inpatient admissions for lower respiratory events. 0 exacerbations subanalysis adjusted for:

tMean SABA daily dose, age, rhinitis diagnosis and/or therapy. zAsthma control status, mean SABA daily dose, sex, rhinitis diagnosis and/or therapy, asthma diagnosis. §Asthma control status, mean SABA daily dose. #Mean SABA daily dose, rhinitis diagnosis and/or therapy. -Oral candidiasis, rhinitis diagnosis and/or therapy, number of consultations for LRTI requiring antibiotic. >1 exacerbations subanalysis adjusted for:

tAge. zMean SABA daily dose. xNo significant effects. #Sex, number of severe exacerbations. Number of severe exacerbations.

TABLE E4. More baseline characteristics of children in the small-particle ICS step-up and add-on LABA cohorts (analysis 2)

ICS step-up vs ICS/LABA combination

ICS step-up vs separate ICS d LABA

Characteristic

ICS dose step-up ICS/LABA combination ICS dose step-up Separate ICS D LABA

(n = 185) (n = 185) P value* (n = 276) (n = 552) P value*

BMI (kg/m2), mean ± SDf 18.0 ± 4.2 17.0 ± 3.4

Year of study index prescription, median (IQR) 2005 (2002-2007) 2006 (2004-2007)

Years since first asthma code, median (IQR) 3.4 (1.6-6.0) 3.3 (1.2-5.5)

Database code for asthma 177 (95.7) 177 (95.7)

.79 17.5 ± 3.3

<.001 2005 (2002-2007)

.36 3.1 (1.3-5.4)

1.0 265 (96.0)

17.3 ± 3.8 2004 (2002-2006) 3.6 (1.8-5.5) 543 (98.4)

.57 <.001 .038 .047

>1 prescription in the previous 12 mo

Nonsteroidal anti-inflammatory drug 12 (6.5) 12 (6.5) 1.0 26 (9.4) 29 (5.3) .019

Acetaminophen 33 (17.8) 34 (18.4) .90 56 (20.3) 106 (19.2) .71

Recorded %predicted PEF 83 (44.9) 82 (44.3) — 111 (40.2) 249 (45.1) —

%predicted PEF, mean ± SD 106 ± 29 104 ± 31 .25 108 ± 30 110 ± 28 .96

Clinical exacerbations

0 131 (70.8) 136 (73.5) .42 188 (68.1) 376 (68.1) .26

1 39 (21.1) 33 (17.8) 60 (21.7) 108 (19.6)

2 9 (4.9) 12 (6.5) 20 (7.2) 45 (8.2)

>3 6 (3.2) 4 (2.2) 8 (2.9) 23 (4.2)

Asthma prescriptions, median (IQR) 4 (3-6) 5 (3-7) .29 4 (2-7) 4 (3-7) .090

All primary care consultations, median (IQR) 5 (3-8) 5 (3-8) .10 5 (3-8) 5 (3-9) .049

GP consultation for LRTI requiring antibiotic

0 159 (85.9) 167 (90.3) .12 236 (85.5) 496 (89.9) .17

>1 26 (14.1) 18 (9.7) 40 (14.5) 56 (10.1)

BMI, Body mass index; IQR, interquartile range; LRTI, lower respiratory tract infection; PEF, peak expiratory flow. Data are n (%) except otherwise indicated.

*Matched cohorts were compared using conditional logistic regression.

fRecorded data were available for BMI for 89 (48.1%) and 86 (46.5%) in ICS step-up and ICS/LABA combination cohorts, respectively, and for 127 (46.0%) and 265 (48.0%) in ICS step-up and separate ICS + LABA cohorts, respectively.

TABLE E5. Asthma therapy during the outcome year for matched cohorts in the small-particle ICS step-up and add-on LABA cohorts (analysis 2)

ICS step-up vs ICS/LABA combination ICS step-up vs separate ICS d LABA

ICS dose ICS/LABA ICS dose Separate

step-up combination step-up ICS + LABA

Outcome (n = 185) (n = 185) P value* (n = 276) (n = 552) P value*

Spacer device prescribed 66 (35.7) 54 (29.2) .18 112 (40.6) 229 (41.5) .81

No. ICS or ICS/LABA inhalers, median (IQR)f 4 (2-6) 6 (4-8) <.001 4 (2-6) 4 (2-7) .65

Three or more ICS inhalers 130 (70.3) 160 (86.5) <.001 205 (74.3) 388 (74.2) .44

Median (IQR) b2-agonist coverage (h){ 2.2 (1.1-4.4) 15.7 (10.1-23.6) <.001 2.2 (1.1-4.4) 9.1 (5.3-16.4) <.001

Change in b2-agonist coverage (h) 0 (-0.6 to 1.6) 13.8 (7.9-21.4) <.001 0.5 (-0.5 to 1.6) 6.6 (3.1-13.4) <.001

Any prescribed change in therapyx 29 (15.7) 43 (23.2) .002 44 (15.9) 166 (30.1) <.001

Increase in ICS dose 5 (2.7) 30 (16.2) <.001 9 (3.3) 125 (22.6) <.001

Any additional therapy 28 (15.1) 18 (9.7) .10 39 (14.1) 111 (20.1) .036

Fluticasone-salmeterol 11 (5.9) 0(0) NA 13 (4.7) 74 (13.4) <.001

Budesonide-formoterol 2(1.1) 0 (0) NA 2 (0.7) 13 (2.4) .12

Long-acting b2-agonist 12 (6.5) 1 (0.5) .017 20 (7.2) 0(0) .012

Leukotriene receptor antagonist|| 8 (4.3) 17 (9.2) .058 11 (4.0) 40 (7.2) .073

DPI, Dry powder inhaler; IQR, interquartile range; NA, not applicable. Data are n (%) unless otherwise stated. *Conditional logistic regression.

fNumber of ICS or ICS/LABA inhalers prescribed during the outcome year in addition to the index prescription.

{Total b2-agonist coverage (SABA + LABA) was calculated as SABA 2 puffs lasting 4 h plus LABA 2 puffs via pMDI or 1 puff via DPI lasting 12 h. Change was the difference between the baseline and outcome years.

§Change in therapy and additional therapy could be at any time during the outcome year; therefore, children could have had more than 1 change/additional therapy. ||Montelukast is the only leukotriene receptor antagonist licensed in the United Kingdom.

TABLE E6. Baseline characteristics of children comparing ICS step-up and separate ICS + LABA cohorts (analysis 2)

ICS step-up vs separate ICS D LABA

Characteristic

ICS dose Separate

step-up ICS + LABA (n = 276) (n = 552) P value*

TABLE E7. Outcome-year results for matched cohorts comparing small-particle ICS step-up and separate ICS + LABA cohorts (analysis 2)

Sex: malef 170 (61.6)

Age at index date (y), 7.7 ± 2.0

mean ± SDf Recorded comorbidityz

Possible atopy 214 (77.5)

Rhinitis diagnosis/Rx 61 (22.1)

Eczema diagnosis/Rx 199 (72.1)

Preschool wheeze 113 (40.9)

diagnosis Preschool asthma diagnosis/Rx Risk-domain asthma controlf 185 (67.0)

Spacer device prescribed 110 (39.9) Mean daily SABA dose (mg/d)f

340 (61.6) 7.7 2.0

442 (80.1) 142 (25.7) 416 (75.4) 249 (45.1)

147 (53.3) 326 (59.1)

370 (67.0) 230 (41.7)

.39 .26 .32 .25

NA .61

ICS step-up vs separate ICS d LABA

Outcome ICS dose step-up (n = 276) Separate ICS d LABA (n = 552)

Risk-domain asthma control 215 (77.9) 404 (73.2)

Treatment stability 191 (69.2) 303 (54.9)

Severe exacerbation

0 232 (84.1) 447 (81.0)

1 33 (12.0) 76 (13.8)

>2 11 (4.0) 29 (5.3)

Clinical exacerbation

0 218 (79.0) 412 (74.6)

1 42 (15.2) 97 (17.6)

>2 16 (5.8) 43 (7.8)

Disaggregated results of composite measures P value*

0-100 50 (18.1) 86 (15.6) >1 asthma-related hospitalization 2 (0.7) 2 (0.4) .49

101-200 100 (36.2) 214 (38.8) .35 >1 acute course of oral 42 (15.2) 100 (18.1) .39

201-400 83 (30.1) 166 (30.1) corticosteroids

401-800 35 (12.7) 70 (12.7) >1 GP consultation for LRTI 26 (9.4) 55 (10.0) .65

>800 8 (2.9) 16 (2.9) requiring antibiotic

Median (IQR) daily ICS dose 55 (27-110) 55 (27-110) .32 Mean >2 puffs daily SABA 165 (59.8) 303 (54.9) .37

(mg/d)x Increase in ICS dose or additional 44 (15.9) 166 (30.1) <.001

Last ICS dose before the index therapy

date (mg/d)t,§

LRTI, lower respiratory tract infection. Data are n (%) unless otherwise stated. *Conditional logistic regression.

IQR, Interquartile range; NA, not applicable; Rx, therapy. *Matched cohorts were compared using conditional logistic regression. fMatching variable (please see above in Online Repository text for details). ^Possible atopy was defined as any 1 or more of the following: recorded rhinitis diagnosis, rhinitis therapy, eczema diagnosis, eczema therapy. Preschool wheeze was captured through database coding; concomitant rhinitis, eczema, and preschool asthma were captured through database-coded diagnosis or therapy for same. Preschool was defined as age 1 to 3 y.

§The doses of ICS were standardized to equivalence with small-particle beclome-thasone and fluticasone; thus, doses of large-particle beclomethasone and budesonide were halved.