Journal of Asthma, 45:293–299, 2008Copyright C© 2008 Informa Healthcare USA, Inc.ISSN: 0277-0903 print / 1532-4303 onlineDOI: 10.1080/02770900801911178

ORIGINAL ARTICLE

Costs and Resource Use of Mild Persistent Asthma Patients Initiated

on Controller Therapy

GENE L. COLICE, M.D.,1 ANDREW P. YU, PH.D.,2 JASMINA I. IVANOVA, M.A.,3,∗ MATTHEW HSIEH, B.A.,2HOWARD G. BIRNBAUM, PH.D.,2 MAUREEN J. LAGE, PH.D.,4 AND CORRINE BREWSTER, R.PH., M.B.A.5

1The George Washington University School of Medicine and Washington Hospital Center, Washington, DC2Analysis Group, Inc., Boston, Massachusetts, USA

3Analysis Group, Inc., New York, New York, USA4HealthMetrics Outcomes Research, LLC, Groton, Connecticut, USA

5Teva Specialty Pharmaceuticals, Horsham, Pennsylvania, USA

Background. The treatment of mild persistent asthma is controversial. Objectives. A retrospective database approach was used to evaluate differentalternatives to treating mild persistent asthma. We hypothesized that treatment with inhaled corticosteroids (ICS) would result in lowest costs thantreatment with leukotriene modifiers (LM) and combination therapy with ICS long-acting inhaled β2-agonists (LABA) because it would be associatedwith fewer acute care visits and hospitalizations than LM and it would have lower drug acquisition costs than both ICS+LABA and LM. Methods.Costs and resource utilization were compared in 1,283 mild persistent asthma patients initiating regular use of either ICS, ICS+LABA, or LM. Mildpersistent asthma patients were identified from a privately insured claims database (1999–2005) using an established algorithm. Wilcoxon rank-sumtests and generalized linear models were used to compare costs. Results. Of the total patients who met study criteria, 319 patients (24.9%) initiatedregular ICS use, 414 (32.3%) ICS+LABA use, and 550 (42.9%) LM use. Over the 1 year after controller therapy initiation, asthma-related directcosts were significantly lower with ICS compared with ICS+LABA or LM ($819 for ICS, $1,094 for ICS+LABA, and $869 for LM, p < 0.001 forall comparisons). There were no significant differences in resource use. Conclusion. In this analysis, physicians, despite guideline recommendations,chose to treat patients with mild persistent asthma more often with LM and ICS+LABA than with ICS. However, therapy with ICS was less costlythan treatment with either LM or ICS+LABA, primarily due to differences in drug costs, and provided similar outcomes.

Keywords mild persistent asthma, inhaled corticosteroids, leukotriene modifiers, asthma-related costs, asthma-related resource use

INTRODUCTION

The treatment of mild persistent asthma is controversial.Asthma management guidelines recommend regular treat-ment with low doses of inhaled corticosteroids (ICS) asthe preferred approach but accept use of leukotriene mod-ifiers (LM) as an alternative (1, 2). Controlled clinical trialshave consistently shown that ICS treatment of mild persistentasthma more effectively improves symptom control and lungfunction than LM (3, 4). However, an LM such as montelukasthas an advantageous safety profile (1, 2). Combination ther-apy, an ICS with a long-acting inhaled β2-agonist (LABA),is not recommended by guidelines for the treatment of mildpersistent asthma (1, 2). Despite these guideline recommen-dations, it is generally recognized that clinicians commonlyrely on ICS+LABA to treat all forms of asthma. Newer ap-proaches to treating mild persistent asthma include intermit-tent use of ICS when symptoms become problematic (5) oronce-daily use of ICS+LABA (6). However, intermittent useof ICS does not provide similar symptom benefits and long-term control of the underlying airway inflammation as dailyuse of ICS (5, 7). Once-daily use of ICS+LABA is probably

Research funding from Teva Specialty Pharmaceuticals.∗Corresponding author: Jasmina Ivanova, M.A., Manager, Analysis

Group, Inc., 10 Rockefeller Plaza 15th Floor, New York, NY 10020; E-mail:jivanova@analysisgroup.com

not reasonable because controlled clinical trials have shownthat the ICS component of the ICS+LABA tested, fluticasonepropionate, requires dosing twice-daily to be effective (8).

In this analysis we used a retrospective database approachto evaluate different alternatives to treating mild persistentasthma. Database analyses have a disadvantage for study-ing asthma because information on symptom control andlung function is not available. However, the advantage ofthe database approach is that it reflects treatment decisionsthat clinicians have made on a daily basis in their practiceand expresses outcomes in terms of costs, which are mean-ingful to payers (9). We relied on an established method toidentify patients with mild persistent asthma who were notregularly using controller medications (10, 11) and calculatedhealth care costs over a 1-year period after these patients hadbeen started on either ICS, LM, or ICS+LABA. We hypoth-esized that ICS treatment would result in the lowest costs ofthe three options because it would be associated with feweracute care visits and hospitalizations than LM and it wouldhave lower drug acquisition costs than both ICS+LABA andLM.

METHODS

Data SourceA privately insured claims database covering over 6

million beneficiaries (including employees, spouses, and

293

J A

sthm

a D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

ichi

gan

Uni

vers

ity o

n 10

/30/

14Fo

r pe

rson

al u

se o

nly.

294 G. L. COLICE ET AL.

dependents) from 31 US-based companies for the years1999–2005 was available for review. The companies have op-erations nationwide in a broad array of industries and job clas-sifications. The database contains de-identified informationon patients’ demographics (e.g., age and gender), monthlyenrollment history, and medical and pharmacy claims. Uti-lization of medical services was recorded in the database withdate of service, associated diagnoses, performed procedures,billed charges, and actual payment amounts. The databasealso includes pharmacy claims with prescribed medicationsidentified by National Drug Code, date of prescription fill,days of supply, quantity, and actual payment amounts.

Sample Selection and Study CohortsPatients included in the study sample had an asthma di-

agnosis (ICD-9-CM code 493), but no diagnosis of chronicobstructive pulmonary disease (ICD-9-CM codes 491, 492,493.2, and 496) from January 1, 2000 to December 31, 2005.

Patients were required to have initiated regular use of ei-ther an ICS, ICS+LABA, or a LM. Regular use was definedas either having an initial prescription of 90 days supply orhaving a refill before a gap of 30 days from of the end of theinitial prescription supply. Patients with fills of other studycontroller medications during the first 90 days after controllerinitiation were excluded. The index date was defined as thedate when patients initiated regular use of controller medica-tion by filling a prescription for either an ICS, ICS+LABA,or a LM.

To ensure complete capture of medical and pharmacy uti-lization and costs, patients in the study sample were requiredto have continuous health coverage eligibility 1 year before(baseline period) and 1 year after the index date (follow-upperiod). Patients older than 64 at index date were excludedbecause information could not be obtained from Medicaresources. Patients younger than 6 at index date were excludedbecause treatment guidelines differ for children younger than6 years and those ages 6 and over.

Mild persistent asthma patients were identified from thedatabase using a two-step process as described previously(10, 11). The first step used the Healthcare Effectiveness Dataand Information Set (HEDIS) criteria that identify patientswith persistent asthma as those having four or more asthmamedication dispensings; or one or more acute inpatient dis-charge(s) with a primary diagnosis of asthma; or one or moreED visit(s) with a primary diagnosis of asthma; or four ormore outpatient visits with asthma listed anywhere as one ofthe diagnoses and two or more asthma medication dispens-ing (12). To ensure that patients with either mild intermittentasthma or moderate-to-severe persistent asthma were not in-cluded in the sample, the second step used Leidy’s criteria(13) as an additional screen in conjunction with establishedguideline recommendations for treating persistent asthma (2).

Leidy’s reliever and oral corticosteroid method categorizesasthma patients into severity groups based on their use ofshort-acting inhaled β2-agonists (SABA) and oral corticos-teroids. To be classified as having mild persistent asthmabased on Leidy’s criteria, patients were required to have(1) four to six inhaled β2-agonist use per year and zero oralsteroid prescriptions per year; or (2) two to three inhaled β2-

agonist use per year and less than two oral steroid prescrip-tions per year; or (3) ≤ one inhaled β2-agonist use per yearand one oral steroid prescription per year. Guideline treat-ment recommendations (2) indicate that patients with mildpersistent asthma would be treated with low-dose ICS andpatients with moderate-to-severe persistent asthma would betreated with either moderate to high doses of ICS or any doseof ICS plus a LABA. Patients were classified into a moresevere group if their treatment conflicted with Leidy’s cri-teria. For instance, all patients identified as mild persistentby the Leidy method who either used medium-to-high doseICS consistently (i.e., three or more claims) or who usedlow-dose ICS consistently and also had claims for salme-terol were classified into the moderate or severe persistentasthma group. Conversely, patients classified as mild persis-tent by the GINA method through either using low-dose ICSconsistently or not using ICSs consistently (i.e., zero to twoclaims) could be classified as moderate or severe by Leidy’smethod.

We further limited our sample to patients with use ofSABA, but no more than two prescriptions for controllersduring the 1-year baseline period. To ensure that patientswere stable at the index date, patients were excluded if theyhad an asthma-related emergency department (ED) visit orhospitalization within 30 days of the index date.

Measures—Resource Use and Comorbidity ProfileResource utilization during the 12-month period before

the index date and 12-month follow-up period and the base-line comorbidity profiles were described for patients in thethree study groups. Resource utilization included any inpa-tient visit, any ED visit, and any outpatient/other visit, as wellas number of visits for each type of service. Inpatient visitswere identified by claims with a place of service specifiedas hospital inpatient, rehabilitation center, residential treat-ment center, or psychiatric facility. ED visits were identifiedby claims with a place of service specified as emergencytreatment center or hospital emergency room or type of ser-vice specified as emergency first aid or emergency room,or emergency room-related procedure codes (CPT codes:99058, 99281, 99282, 99283, 99284, 99285). All other visitswere categorized as outpatient/other visits.

Medical services were classified as asthma-related inthe presence of an asthma diagnosis (ICD-9-CM code493) associated with the claims. Asthma-related medica-tion use included prescription claims for study controllers(ICS, ICS+LABA, LM), other controllers (aminophyllineand methylxanthine), oral corticosteroids (hydrocortisone,methylprednisolone, prednisolone, and prednisone), SABA(albuterol, bitolterol, fenoterol, isoetharine, metaproterenol,pirbuterol, procaterol, terbutaline), and other relievers.

The baseline comorbidity profile was constructed basedon claims data during the 12-months before the index date.Asthma-related comorbidities included otitis media (ICD-9-CM codes 381 and 382), sinusitis (ICD-9-CM codes 461 and473), allergic rhinitis (ICD-9-CM code 477), angina (ICD-9-CM code 413), pneumonia (ICD-9-CM codes 480, 481,482, 484, 485, 486, 487, and 770.0), gastroesophageal refluxdisease (or GERD, ICD-9-CM code 530.81), and obesity

J A

sthm

a D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

ichi

gan

Uni

vers

ity o

n 10

/30/

14Fo

r pe

rson

al u

se o

nly.

COSTS OF MILD PERSISTENT ASTHMA PATIENTS 295

(ICD-9-CM code 278). As a measure of overall physical co-morbidities, we also calculated the Charlson Comorbidity In-dex (CCI), a weighted sum of 17 physical comorbidities (14).The 17 comorbidities were identified using ICD-9-CM codesselected by Romano et al. (15). Asthma diagnosis (ICD-9-CM493) was excluded when calculating the CCI.

Measures—CostsHealthcare costs were calculated for total (including any

medical services and drug use) and asthma-related treatments(including asthma-related medical services and drugs) dur-ing the 12-month follow-up period. Cost analyses were con-ducted from third-party payer’s perspective (i.e., costs weredefined as payments to providers by third-party payers). Allcosts were adjusted for inflation to 2005 U.S. dollars usingthe medical care component of Consumer Price Index.

Medical costs were calculated for inpatient, ED, and outpa-tient/other medical services (e.g., outpatient surgery, physi-cian services, laboratory, and other ancillary services). Totaldirect costs include medical and prescription drug costs.

Asthma-related medical costs were costs associated withclaims with a diagnosis of asthma. Asthma-related drug costsare defined as costs for all prescription claims for SABA,other relievers, ICS, LABA, ICS+LABA, LM, other con-trollers, and oral corticosteroids.

Statistical AnalysesBaseline characteristics, resource use, and costs during

the follow-up period were compared descriptively for mildpersistent asthma patients initiating regular use of ICS,ICS+LABA, and LM. Categorical variables were com-pared using Chi-squared tests. Costs and resource use countswere compared descriptively using Wilcoxon rank-sum tests.Other continuous variables (e.g., age and CCI) were com-pared using t tests. Multivariate generalized linear modelswith a log link and gamma distribution were also used tocompare costs controlling for age, gender, index year andindex quarter, baseline asthma comorbidities, the baselineCCI, prior resource utilization (i.e., any all-cause and asthma-related inpatient, ED, and outpatient visits), prior medica-tion use of any controller medications and oral steroids,and the number of SABA prescriptions and different classesof drugs used during the 12-month pre-index period. Risk-adjusted costs were then estimated for patients initiating ICS,ICS+LABA, or LM. All analyses were conducted using SASversion 9.1 (SAS Institute Inc., Cary, NC).

Sensitivity AnalysisIn the initial analysis, extreme outliers (three patients with

total costs greater than 10 standard deviations from the meantotal cost across the three groups) were excluded. We alsoconducted a sensitivity analysis that included these outliers.

RESULTS

Baseline CharacteristicsAmong a total of 1,283 mild-persistent asthma patients

who met the study criteria, 319 (24.9%) initiated regular ICSuse, 414 (32.3%) initiated ICS+LABA use, and 550 (42.9%)initiated LM use (Table 1). Patients in the ICS+LABA group

were on average older than patients in the ICS and LM groups.Patients in the ICS group were on average older than pa-tients in the LM group. The baseline comorbidity profileof patients in the three groups was similar with the excep-tion of the rates of otitis media and gastroesophageal refluxdisease (GERD). The rate of otitis media was significantlylower in the ICS+LABA group (8.5%) compared with theICS group (13.8%, p = 0.021) and the LM group (13.5%,p = 0.016). The rate of GERD was significantly higher inthe ICS+LABA group (4.1%) compared with the LM group(1.8%, p = 0.033).

Patients in the ICS group had a lower rate of inpatient vis-its in the 12 months before the index date compared withICS+LABA and with LM. However, the percent of patientswith asthma-related hospitalizations, ED visits, and outpa-tient visits and the number of these events were similar amongthe three treatment groups. The ICS group had significantlymore frequent use of asthma controller medication during thebaseline period than the other two groups and the LM grouphad significantly more frequent use than the ICS+LABAgroup (62.7% in the ICS group, 29.2% in the ICS+LABAgroup, and 42.4% in the LM group). The ICS+LABA grouphad a significantly higher number of SABA prescriptions dur-ing the baseline period and a lower rate of oral corticosteroiduse than the LM group, but the differences were small.

Unadjusted Resource Use ComparisonsSignificant differences in resource utilization as measured

by the rates and counts of total and asthma-related inpatientvisits, ED visits, and outpatient visits and other visits werenot found across treatment medication groups during the 12-month follow-up period (Table 2).

During the 1-year treatment period, number of prescrip-tions filled for ICS, ICS+LABA, and LM were 7.3 ± 6.1(SD), 9.6 ± 6.3 (SD), and 10.8 ± 8.1 (SD), respectively. Thedifferences in the number of index controller prescriptionsbetween groups were all statistically significant, p < 0.01.

Unadjusted Cost ComparisonsTotal direct costs during the 12-month period after the in-

dex date were significantly lower in the ICS and LM groupsthan in the ICS+LABA group ($3,971 for ICS group, $4,733for ICS+LABA group, and $4,148 for LM group, p < 0.001for both comparisons), but costs were not significantly dif-ferent between the ICS and LM groups (Table 3). Direct costdifferences were attributable to differences in drug and out-patient visit costs, both of which were significantly lower forthe ICS and LM groups than the ICS+LABA group. Drugcosts were also lower for the ICS group than the LM group($1,499 for ICS and $1,819 for LM, p = 0.001)

Asthma-related direct costs were significantly lower for theICS and LM groups compared with the ICS+LABA groupand significantly lower with ICS compared with LM ($819for ICS, $1,094 for ICS+LABA, and $869 for LM, p = 0.001for all comparisons). This difference was largely due to differ-ences in drug costs. The ICS and LM groups had significantlylower drug costs compared with the ICS+LABA group, andthe ICS group had significantly lower drug costs than the LMgroup ($510 for ICS, $797 for ICS+LABA, and $655 for LM,

J A

sthm

a D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

ichi

gan

Uni

vers

ity o

n 10

/30/

14Fo

r pe

rson

al u

se o

nly.

296 G. L. COLICE ET AL.

TABLE 1.—Patient baseline characteristics.

ICS+LABA ICS vs. ICS+LABA ICS vs. LM ICS+LABAICS (N = 319) (N = 414) LM (N = 550) p-value p-value vs. LM p-value

Demographic factorsAge, mean (SD) 27.0 (19.1) 33.2 (17.2) 23.3 (17.3) <0.001 0.004 <0.001Gender, % male 52.4% 46.6% 47.5% 0.124 0.164 0.797

Asthma-related comorbidities, number (%)Otitis Media 44 (13.8%) 35 (8.5%) 74 (13.5%) 0.021 0.888 0.015Sinusitis 61 (19.1%) 98 (23.7%) 128 (23.3%) 0.138 0.153 0.885Allergic rhinitis 102 (32.0%) 129 (31.2%) 199 (36.2%) 0.814 0.209 0.103Angina 1 (0.3%) 3 (0.7%) 3 (0.6%) 0.454 0.626 0.726Pneumonia 23 (7.2%) 33 (8.0%) 40 (7.3%) 0.701 0.973 0.685GERD 9 (2.8%) 17 (4.1%) 10 (1.8%) 0.351 0.330 0.033Obesity 4 (1.3%) 8 (1.9%) 9 (1.6%) 0.473 0.654 0.730

Charlson Comorbidity Index, mean (SD) 0.08 (0.35) 0.13 (0.56) 0.11 (0.66) 0.122 0.419 0.564Resource utilization during the 1-year baselineNumber (%) of patients with:

All-cause inpatient stay 11 (3.5%) 30 (7.3%) 39 (7.1%) 0.027 0.026 0.926All-cause ED visits 76 (23.8%) 105 (25.4%) 161 (29.3%) 0.632 0.082 0.179All-cause outpatient visits 296 (92.8%) 396 (95.7%) 527 (95.8%) 0.095 0.055 0.899

Mean (SD) number of:All-cause inpatient stays 0.03 (0.18) 0.12 (0.69) 0.09 (0.36) 0.026 0.014 0.296All-cause ED visits 0.32 (0.71) 0.42 (1.02) 0.45 (0.91) 0.138 0.033 0.673All-cause outpatient visits 9.18 (9.55) 9.95 (11.34) 10.12 (11.78) 0.326 0.222 0.820

Number (%) of patients with:Asthma-related inpatient stay 3 (0.9%) 3 (0.7%) 6 (1.1%) 0.748 0.833 0.558Asthma-related ED visits 17 (5.3%) 28 (6.8%) 45 (8.2%) 0.423 0.115 0.410Asthma-related outpatient visits 201 (63.0%) 275 (66.4%) 336 (61.1%) 0.337 0.575 0.089Mean (SD) number of:Asthma-related inpatient stays 0.01 (0.10) 0.01 (0.15) 0.01 (0.10) 0.779 0.833 0.885Asthma-related ED visits 0.06 (0.26) 0.08 (0.32) 0.10 (0.35) 0.426 0.101 0.385Asthma-related outpatient visits 1.35 (2.24) 1.30 (1.92) 1.25 (1.89) 0.777 0.504 0.677Medication use during the 1-year baselineAny use of asthma controller medication 200 (62.7%) 121 (29.2%) 233 (42.4%) <0.001 <0.001 <0.001Number of SABA prescriptions, mean (SD) 3.2 (1.4) 3.3 (1.5) 3.0 (1.4) 0.132 0.103 0.001Any use of oral corticosteroids 109 (34.2%) 131 (31.6%) 216 (39.3%) 0.470 0.134 0.015

ICS = inhaled corticosteroids; LABA = long-acting inhaled β2-agonists; LM = leukotriene modifiers; SABA = short-acting inhaled β2-agonists; ED = emergency department; SD =standard deviation.

p = 0.001 for all comparisons). The LM group had signifi-cantly lower costs for outpatient visits than the ICS+LABAgroup (p = 0.014).

Risk-Adjusted Cost ComparisonsMultivariate analysis using generalized linear regressions

provided similar results to the descriptive comparisons of to-tal direct costs and asthma-related direct costs (Table 4). Risk-adjusted direct costs were significantly lower in the ICS groupcompared with the ICS+LABA group ($4,305 for ICS and$4,997 for ICS+LABA, p = 0.016). Risk-adjusted asthma-

TABLE 2.—Resource use in the 12-month follow-up period.

ICS ICS+LABA LM ICS vs. ICS+LABA ICS vs. LM ICS+LABA vs. LM(N = 319) (N = 414) (N = 550) p-value p-value p-value

Number (%) of patients with:All-cause inpatient stay 21 (6.6%) 31 (7.5%) 29 (5.3%) 0.217 0.415 0.589All-cause ED visits 91 (28.5%) 99 (23.9%) 148 (26.9%) 0.840 0.688 0.506All-cause outpatient visits 303 (95.0%) 399 (96.4%) 523 (95.1%) 0.851 0.727 0.554

Mean (SD) number of:All-cause inpatient stays 0.09 (0.35) 0.09 (0.34) 0.07 (0.32) 0.167 0.226 0.783All-cause ED 0.43 (0.85) 0.38 (0.96) 0.44 (1.01) 0.447 0.963 0.380All-cause outpatient visits 11.02 (13.81) 12.32 (13.19) 12.00 (14.57) 0.176 0.958 0.121

Number (%) of patients with:Asthma-related inpatient stay 5 (1.6%) 5 (1.2%) 5 (0.9%) 0.088 0.863 0.054Asthma-related ED visits 26 (8.2%) 25 (6.0%) 40 (7.3%) 0.219 0.353 0.685Asthma-related outpatient visits 217 (68.0%) 299 (72.2%) 341 (62.0%) 0.898 0.892 1.000

Mean (SD) number of:Asthma-related inpatient stays 0.02 (0.12) 0.01 (0.11) 0.01 (0.09) 0.067 0.864 0.073Asthma-related ED 0.10 (0.37) 0.07 (0.30) 0.08 (0.30) 0.436 0.791 0.512Asthma-related outpatient visits 1.90 (2.56) 1.76 (2.23) 1.68 (3.48) 0.375 0.439 0.808

Any use of oral corticosteroids 108 (33.9%) 112 (27.1%) 166 (30.2%) 0.457 0.925 0.455

ICS = inhaled corticosteroids; LABA = long-acting inhaled β2-agonists; LM = leukotriene modifiers; ED = emergency department; SD = standard deviation.

related direct costs were significantly lower with ICS treat-ment compared with both ICS+LABA and LM ($782 forICS, $1,126 for ICS+LABA and $871 for LM, p < 0.001for ICS and LM versus ICS+LABA andp = 0.034 for ICSversus LM).

Sensitivity AnalysisUnadjusted and multivariate cost comparisons including

the three patients with total costs greater than 10 standarddeviations from the mean total costs produced similar results(data not shown).

J A

sthm

a D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

ichi

gan

Uni

vers

ity o

n 10

/30/

14Fo

r pe

rson

al u

se o

nly.

COSTS OF MILD PERSISTENT ASTHMA PATIENTS 297

TABLE 3.—Comparison of observed costs in the 12-month follow-up period.

ICS ICS+LABA LM(N = 319) (N = 414) (N = 550)

ICS vs. ICS+LABAICS+LABA ICS vs. LM vs. LM

Mean SD Median Mean SD Median Mean SD Median p value p value p value

Total direct costs $3,971 $6,895 $2,077 $4,733 $6,549 $2,678 $4,148 $6,797 $2,029 <0.001 0.192 0.001Drug costs $1,499 $2,643 $848 $2,053 $2,690 $1,259 $1,819 $3,592 $1,023 <0.001 0.001 <0.001Medical costs $2,472 $5,327 $821 $2,680 $4,975 $1,046 $2,330 $4,884 $797 0.040 0.879 0.028

Hospital inpatient costs $514 $3,016 $0 $454 $3,135 $0 $433 $3,181 $0 0.632 0.395 0.138ED costs $159 $426 $0 $127 $438 $0 $146 $484 $0 0.199 0.700 0.279Outpatient/other costs $1,799 $3,113 $667 $2,099 $3,099 $926 $1,751 $2,665 $712 0.016 0.657 0.026

Asthma-related direct costs $819 $1,068 $513 $1,094 $1,349 $878 $869 $712 $722 <0.001 <0.001 <0.001Drug costs $510 $411 $384 $797 $431 $732 $655 $397 $600 <0.001 <0.001 <0.001Medical costs $309 $973 $76 $296 $1,261 $84 $214 $542 $64 0.669 0.130 0.023

Hospital inpatient costs $75 $825 $0 $70 $1,129 $0 $10 $176 $0 0.675 0.373 0.646ED costs $43 $198 $0 $16 $80 $0 $25 $120 $0 0.232 0.586 0.435Outpatient/other costs $191 $373 $71 $211 $552 $75 $179 $476 $476 0.502 0.159 0.014

ICS = inhaled corticosteroids; LABA = long-acting inhaled β2-agonists; LM = leukotriene modifiers; ED = emergency department; SD = standard deviation

DISCUSSION

This analysis has several important findings. Despiteguideline recommendations that ICS are the preferred treat-ment for mild persistent asthma, clinicians more commonlychose LM and ICS+LABA to treat these patients. Over the1-year baseline period before initiation of regular controllermedications, meaningful differences in asthma control (as as-sessed by asthma-related hospitalizations, ED visits or out-patient visits, number of SABA prescriptions filled, and useof oral corticosteroids) were not found among the three dif-ferent treatment groups to explain the difference in treatmentapproaches. Treatment with ICS provided cost advantagesover use of either ICS+LABA or LM, as demonstrated bylower asthma-related direct costs. Measures of asthma con-trol were similar over the 1-year follow-up period for thethree treatment groups.

Asthma management guidelines recommend ICS as thepreferred treatment for mild persistent asthma rather than ei-ther LM or ICS+LABA based on the efficacy results obtainedfrom controlled clinical trials (1, 2). Multiple clinical trialshave shown that ICS provide more effective symptom controland improvements in lung function than LM (3, 4). AlthoughLM such as montelukast have an overall more favorablesafety profile than ICS, at the low doses recommended fortreatment of mild persistent asthma ICS should also be quitesafe (1, 2). Although the use of once -aily ICS, such as flutica-sone, might provide even a greater safety margin, the efficacyreported in some studies (16–18) has not been sufficientlyconvincing to lead to FDA approval of this dosing regimen(8). In the OPTIMA trial patients with mild persistent asthmawho had not previously been on an ICS were randomized totreatment with placebo, ICS, or ICS+LABA (19). Significantimprovements in symptom control, lung function, and exacer-bation rates were found with both ICS and ICS+LABA treat-ment. There were slightly greater increases in lung function

TABLE 4.—Comparison of risk-adjusted costs in the 12-month follow-up period.

ICS ICS+LABA LMICS vs. ICS+LABA ICS vs. LM ICS+LABA vs. LM

Mean SD Mean SD Mean SD p-value p-value p-value

Total direct costs $4,305 $8,083 $4,997 $9,382 $4,562 $8,567 0.016 0.319 0.086Total asthma-related direct costs $782 $199 $1,126 $286 $871 $221 <0.001 0.034 <0.001

ICS = inhaled corticosteroids; LABA = long-acting inhaled β2-agonists; LM = leukotriene modifiers; SD = standard deviation.

with ICS+LABA treatment than with ICS therapy alone, butthe reductions in symptoms and exacerbation rates were sim-ilar for the ICS and ICS+LABA groups. Because outcomesare generally similar for ICS and ICS+LABA treatment ofmild persistent asthma, management guidelines recommenduse of an ICS alone for treatment to minimize safety concernsrelated to exposure to multiple drugs. This study could notaddress efficacy differences in terms of daily symptom con-trol and lung function among the three treatments. However,the finding that asthma-related outcomes, such as hospital-izations, ED visits, and oral corticosteroid use, were similarfor the three treatment groups during the 1-year follow-up pe-riod supports the value of ICS treatment for mild persistentasthma. Whether recent approval of a new ICS+LABA com-bination product with the rapid acting formoterol will leadto greater adherence to ICS therapy and improved overallasthma control is presently uncertain.

Asthma is a common and expensive disease (20–22). Costsincrease with the severity of the disease, but patients withmild persistent asthma will generate increased overall costsfor health care (10, 11, 23–25). Previous studies, using botha cost-effectiveness and willingness-to-pay approach, haveshown benefits with regular use of ICS in mild persistentasthma rather than relying on symptom control with SABAalone (26, 27). Others have attempted to compare cost savingsachieved with ICS and LM in asthma, but these studies wereflawed by the lack of methods to identify the appropriate pa-tient population, mild persistent asthma patients, for analysis(28–30). Friedman and Yawn compared resource utilizationin asthma patients using either an ICS or an ICS+LABA(31). They applied a less rigorous algorithm than used in thisstudy to identify patients with mild persistent asthma, andfound, as we did, that health care costs were lower for pa-tients receiving ICS than ICS+LABA and health outcomeswere similar. More widespread compliance with guideline

J A

sthm

a D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

ichi

gan

Uni

vers

ity o

n 10

/30/

14Fo

r pe

rson

al u

se o

nly.

298 G. L. COLICE ET AL.

recommendations to use ICS, rather than LM or ICS+LABA,for the treatment of mild persistent asthma might have im-portant cost benefits for payers.

Of note, nonadherence with use of controllers by asthmapatients is well recognized and clearly related to poor out-comes (32, 33). In this analysis we found that patients in allthree treatment groups frequently refilled their prescriptionsover the 1-year follow-up period.

There are two important issues regarding study methodsthat must be addressed. The first relates to whether the ap-proach used for severity categorization accurately identifiedpatients with mild persistent asthma. Guidelines recommendusing clinical criteria, i.e., symptoms, functional limitations,and lung function, to categorize asthma severity (1). However,accumulating information suggests that information availablein an administrative database, specifically medication claims,ED visits and hospitalizations, can be reliably used for thispurpose. Schatz and colleagues used the National Commit-tee for Quality Assurance HEDIS administrative databasecriteria (12) to identify persistent asthma patients. Using vali-dated asthma symptom questionnaires, they found that SABAclaims significantly reflected asthma symptom control. Pa-tients with more SABA claims reported worse asthma con-trol and were more likely to have an asthma-related ED visitsand hospitalization (34, 35). They also found that six or moreSABA claims per year was a marker of more severe asthma(36). Cabana et al. interviewed the parents of 896 childrenwith asthma, determined asthma severity based on guidelinemethods, and evaluated how well the HEDIS method per-formed in identifying persistent asthma (37). They found thatthe HEDIS method was sensitive but did misclassify a minor-ity of children with intermittent asthma as having persistentdisease. Others have made the same observation (38).

To minimize the likelihood of misclassifying intermittentasthma as persistent with the HEDIS method, some authorshave suggested using a 2-year observation period (39, 40). In-stead of this approach, in this and our previous work (10, 11)we incorporated the Leidy criteria, which effectively excludesintermittent asthma patients by establishing minimal require-ments for SABA claims, as a secondary screen to the HEDISmethod. The Leidy criteria also tends to exclude moderateand severe persistent asthma patients, per observations bySchatz et al. (36), by limiting SABA claims to six or fewer.Incorporating the guideline treatment recommendations oncontroller medications as an additional secondary screen alsotends to exclude moderate and severe persistent asthma pa-tients. Supporting the validity of our approach to identifyingmild persistent asthma patients was the finding that patientsincluded in our sample had about three SABA claims duringthe baseline period. Also, about 30% to 40% had receivedoral corticosteroids. In the OPTIMA trial in mild persistentasthma, 79 of 237 (33%) patients in the placebo arm of theICS free group required oral corticosteroids during the 1-yeartrial (16). In the recent American Lung Association trial inmild persistent asthma, 28.4% to 31.5% of patients receivedoral corticosteroids in the year before the study (6).

The second issue regarding methods used in this studyinvolves possible differences among the three different treat-ment groups at baseline. To address this issue we examinedmultiple baseline factors. There were no differences amonggroups in asthma-related comorbidities except for higher

rates of otitis media in the ICS and LM groups than the ICS+ LABA group and more GERD in the ICS + LABA groupthan the LM group. However, these differences were smalland not clinically relevant. Although there were differencesin hospitalizations for any cause among groups during the1 year before the study, again the variation among groupswas small. Differences among groups in the Charlson co-morbidity index were not found. Most important, differencesamong groups in asthma-related hospitalizations and ED vis-its, SABA claims, and oral corticosteroid use were not found.The ICS group was significantly more likely to have used acontroller medication than the other groups, but this suggeststhat they might actually have had more symptomatic disease.Although these findings reassure us that the groups were wellmatched, we cannot exclude the possibility that there wereother unmeasured differences in severity factors among thetreatment groups at baseline.

In this analysis we found that physicians, despite guidelinerecommendations, chose to treat patients with mild persistentasthma more often with LM and ICS+LABA than with ICS.Differences in asthma-related hospitalizations, ED visits, andSABA use did not explain the choice of LM or ICS+LABArather than ICS as the initial treatment choice. After initiationof controller therapy, the three treatment groups had similaroutcomes. Therapy with ICS was less costly than treatmentwith either LM or ICS+LABA, primarily due to differencesin drug costs.

REFERENCES

1. National Asthma Education and Prevention Program. Expert Panel ReportII: Guidelines for the Diagnosis and Management of Asthma. NHLBI, 1997.

2. Global Initiative for Asthma. Global Strategy for Asthma Management andPrevention. NHLBI, 2006.

3. Ducharme FM. Inhaled glucocorticoids versus leukotriene receptor antag-onists as single agent asthma treatment. BMJ 2003; 326:621–625.

4. Ducharme FM, Di Salvio F. Anti-leukotriene agents compared to inhaledcorticosteroids in the management of recurrent and/or chronic asthma inadults and children. Cochrane Database of Systemic Reviews 00075320,2004.

5. Boushey HA, Sorkness CA, King TS, Sullivan SD, Fahy JV, Lazarus SC,et al. Daily versus as-needed corticosteroids for mild persistent asthma. NEngl J Med 2005; 352:1519–1528.

6. The American Lung Association Asthma Clinical Research Centers. Ran-domized comparison of strategies for reducing treatment in mild persistentasthma. N Engl J Med 2007; 356:2027–2039.

7. Kluijver J, Evertse CE, Schrumpf JA, van der Veen H, Zwinderman AH,Hiemstra PS, et al. Asymptomatic worsening of airway inflammation duringlow-dose allergen exposure in asthma. Am J Respir Crit Care Med 2002;166:294–300.

8. Purucker ME, Rosebraugh CJ, Zhou F, Meyer RJ. Inhaled fluticasone pro-pionate by Diskus in the treatment of asthma. Chest 2003; 124:1584–1593.

9. Thomas M, Cleland J, Price D. Database studies in asthma pharmacoeco-nomics. Expert Opin Pharmocother 2003; 4:351–358.

10. Colice G, Wu EQ, Birnbaum H, Daher M, Marynchenko MB, VargheseS. Healthcare and workloss costs associated with patients with persistentasthma in a privately insured population. J Occup Environ Med 2006;48:794–802.

11. Colice G, Wu EQ, Birnbaum H, Daher M, Marynchenko MB. Use of inhaledcorticosteroids and health care costs in mild persistent asthma. J Asthma2007; 44:479–483.

12. National Committee for Quality Assurance. Use of appropriate medica-tions for people with asthma. HEDIS 2005, vol 2. Technical Specifications.Washington, DC.

J A

sthm

a D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

ichi

gan

Uni

vers

ity o

n 10

/30/

14Fo

r pe

rson

al u

se o

nly.

COSTS OF MILD PERSISTENT ASTHMA PATIENTS 299

13. Leidy NK, Paramore LC, Watrous M, Doyle J, Zeiger RS. Developmentof an algorithm for estimating asthma severity from an administrative costdatabase. Value Health 1999; 2:394.

14. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of clas-sifying prognostic comorbidity in longitudinal studies: development andvalidation. J Chron Dis 1987; 40:373–383.

15. Romano PS, Roos LL, Jollis JG. Adapting a clinical comorbidity indexfor use with ICD-9-CM administrative data: differing perspectives. J ClinEpidemiol 1993;46:1075–1079.

16. Wolfe J, Rooklin A, Grady J, Munk ZM, Stevens A, Prillaman B, Duke S,Harding S. Comparison of once- and twice-daily dosing of fluticasone pro-pionate 200 micrograms per day administered by diskus device in patientswith asthma treated with or without inhaled corticosteroids. J Allergy ClinImmunol. 2000; 105:1153–1161.

17. Nathan RA, Li JT, Finn A, Jones R, Payne JE, Wolford JP, Harding SM.A dose-ranging study of fluticasone propionate administered once dailyvia multidose powder inhaler to patients with moderate asthma. Chest2000;118:296–302.

18. ZuWallack R, Adelglass J, Clifford DP, Duke SP, Wire PD, Faris M, Hard-ing SM. Long-term efficacy and safety of fluticasone propionate powder ad-ministered once or twice daily via inhaler to patients with moderate asthma.Chest 2000; 118:303–312.

19. O’Byrne PM, Barnes PJ, Rodriguez-Roisin R, Runnerstrom E, SandstromT, Svensson K, et al. Low dose inhaled budesonide and formoterol in mildpersistent asthma. Am J Respir Crit Care Med 2001;164:1392–1397.

20. Morbidity and Mortality Weekly Report. Asthma prevalence and controlcharacteristics by race/ethnicity—United States, 2002. Morbid Mortal WklyRep MMWR 2004; 53:7.

21. Beasley R. The burden of asthma with specific reference to the United States.J Allergy Clin Immunol 2002; 109:S482–S489.

22. Birnbaum HG, Berger WE, Greenberg PE, Holland M, Auerbach R, AtkinsKM, et al. Direct and indirect costs of asthma to an employer. J Allergy ClinImmunol 2002; 109:264–270.

23. Serra-Batlles J, Plaza V, Morejon E, Comella A, Brugues J. Costs of asthmaaccording to the degree of severity. Eur Respir J 1998; 12:1322–1326.

24. Godard P, Chanez P, Siraudin L, Nicoloyannis N, Duru G. Costs of asthmaare correlated with severity. Eur Respir J 2002; 19:61–67.

25. Cisternas MG, Blanc PD, Yen IH, Katz PP, Earnest G, Eisner MD, et al.A comprehensive study of the direct and indirect costs of adult asthma. JAllergy Clin Immunol 2003; 111:1212–1218.

26. Sullivan SD, Buxton M, Andersson F, Lamm CJ, Liljas B, Chen YZ, et al.Cost-effectiveness analysis of early intervention with budesonide in mildpersistent asthma. J Allergy Clin Immunol 2003; 12:1229–1236.

27. O’Byrne P, Cuddy L, Taylor DW, Birch S, Morris J, Syrotuik J. Efficacy andcost benefit of inhaled corticosteroids in patients considered to have mildasthma in primary care practice. Can Respir J 1996; 3:169–175.

28. O’Connor RD, Parasuraman B, Roberts C, Leibman C. Inhaled corticos-teroids vs leukotriene receptor antagonists. Ann Allergy Asthma Immunol2006; 97:236–243.

29. Orsini L, Limpa-Amara S, Crown WH, Stanford RH, Kamal K. Asthmahospitalization risk and costs for patients treated with fluticasone propionatevs montelukast. Ann Allergy Asthma Immunol 2004; 92:523–529.

30. Balkrishnan R, Nelson LM, Kulkarni AS, Pleasants RA, Whitmire JT,Schechter MS. Outcomes associated with initiation of different controllertherapies in a Medicaid asthmatic population. J Asthma 2005; 42:35–40.

31. Friedman HS, Yawn BP. Resource utilization in asthma. Curr Med Res Opin2007; 23:427–434.

32. Rand C, Bilderback A, Schiller K, Edelman JM, Hustad CM, Zeiger RS,et al. Adherence with montelukast or fluticasone in a long-term clinical trial:results from the mild asthma montelukast versus inhaled corticosteroid trial.J Allergy Clin Immunol 2007; 119:916–923.

33. Milgrom H, Bender B, Ackerson L, Bowry P, Smith B, Rand C. Noncompli-ance and treatment failure in children with asthma. J Allergy Clin Immunol1996;98:1051–1057.

34. Schatz M, Zeiger RS, Vollmer WM, Mosen D, Apter AJ, Stibolt TB,et al. Validation of a β-agonist long-term asthma control scale derivedfrom computerized pharmacy data. J Allergy Clin Immunol 2006;117:995–1000.

35. Schatz M, Zeiger RS, Vollmer WM, Mosen D; Mendoza G, Apter AJ, et al.The controller-to-total asthma medication ratio is associated with patient-centered as well as utilization outcomes. Chest 2006;130:43–50.

36. Schatz M, Nakahiro R, Crawford W, Mendoza G, Mosen D, StiboltTB. Asthma quality-of-care markers using administrative data. Chest2005;128:1968–1973.

37. Cabana MD, Slish KK, Nan B, Clark NM. Limits of the HEDIS criteria indetermining asthma severity for children. Pediatrics 2004; 114:1049–1055.

38. Berger WE, Legorreta AP, Blaiss MS, Schneider EC, Luskin AT, StempelDA, et al. The utility of the Health Plan Employer Data and Information Set(HEDIS) asthma measure to predict asthma-related outcomes. Ann AllergyAsthma Immunol 2004; 93:538–545.

39. Fuhlbrigge AL, Carey VJ, Finkelstein JA, Lozano P, Inui TS, Weiss ST, et al.Validity of the HEDIS criteria to identify children with persistent asthmaand sustained high utilization. Am J Manag Care 2005; 11:325–330.

40. Mosen DM, Macy E, Schatz M, Mendoza G, Stibolt TB, McGaw J, et al.How well do the HEDIS asthma inclusion criteria identify persistent asthma.Am J Manag Care 2005; 11:650–654.

J A

sthm

a D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

ichi

gan

Uni

vers

ity o

n 10

/30/

14Fo

r pe

rson

al u

se o

nly.