- Email: [email protected]
Asthma rescue and allergy medication use among asthmatic children with prior allergy prescriptions who initiated asthma controller therapy
Asthma rescue and allergy medication use among asthmatic children with prior allergy prescriptions who initiated asthma controller therapy Allan Luskin, MD*; Don Bukstein, MD*; Vasilisa Sazonov Kocevar, PhD†; and Donald D. Yin, PhD†
Background: Asthma and allergic rhinitis are frequently comorbid conditions. Montelukast is effective in treating both diseases and may reduce total medication use among children with asthma and allergic rhinitis. Objective: To determine the differences in respiratory and allergy medication use and costs, as proxies for control, in pediatric patients with asthma and allergy who initiated asthma controller therapy. Methods: A 24-month, retrospective, pre-post cohort study using a pharmacy claims database of children (age ⬍16 years) with 2 or more consecutive asthma controller prescriptions and 1 or more allergy prescription (within 12 months before initial controller prescription). Children taking inhaled corticosteroids (ICSs) and montelukast were matched one to one based on age, days of prior allergic rhinitis therapy supply, duration of controller therapy, and propensity score. Differences in costs of rescue or acute asthma medications, prescription allergy medications, other respiratory medications, and the number of days of rescue or acute asthma medication use and allergy medication use were calculated. Results: A total of 1,236 children were matched into ICS and montelukast groups (n ⫽ 618 each). Montelukast patients had a smaller cost increase overall compared with ICS patients (combined cost for rescue or acute asthma medications, allergy medications, and other respiratory medications: $5.55 vs $12.08, P ⬍ .001). Cost increase for rescue or acute asthma medications was significantly lower in the montelukast group ($0.94 vs $3.82, P ⫽ .003). The cost increase for allergy medications ($5.29 vs $10.06, P ⬍ .001) was also significantly lower in the montelukast group. Patients taking montelukast also had fewer days of therapy with asthma rescue medication and allergy medication compared with patients taking ICSs. Conclusions: Initiating therapy with montelukast was associated with better asthma and allergy control demonstrated via lower increase in use and costs of asthma rescue and allergy medications compared with initiating ICS therapy. Ann Allergy Asthma Immunol. 2005;95:129–136.
INTRODUCTION Asthma is the most common chronic illness in childhood.1 Children with asthma use significantly more health care services than children in the general population, and among children with asthma, those with allergic rhinitis account for a large proportion of health care utilization in terms of outpatient care, pharmacy, and urgent care.2,3 Asthma and allergic rhinitis are frequently comorbid conditions.4 The prevalence of nasal allergy symptoms has been shown to range from 78%5 to more than 90%6 in patients with allergic asthma. This association is particularly evident in young patients.7 Epidemiologic, clinical, and pathophysiologic studies suggest that allergic rhinitis and asthma may be considered manifestations of a single disease.8 –10 Treatment of allergic rhinitis has been shown to improve symptoms of seasonal asthma,11,12 and high-dose inhaled corticosteroids
* Respiratory Institute, Dean Medical Center, and University of Wisconsin, Madison, Wisconsin. † Merck & Co Inc, Whitehouse Station, New Jersey. This study was supported by a grant from Merck and Co Inc. Received for publication November 4, 2004. Accepted for publication in revised form January 24, 2005.
VOLUME 95, AUGUST, 2005
(ICSs) used to treat subclinical asthma improve allergic rhinitis symptoms.13 However, despite associations between the 2 disorders, allergic rhinitis and asthma are generally treated as separate entities. Montelukast, a potent antagonist of the type 1 cysteinyl leukotriene receptor, is approved for treatment of both allergic rhinitis and asthma in the same oral tablet preparation. Since studies suggest that there is a clinically relevant association between asthma and allergic rhinitis, montelukast used as monotherapy for asthma may have a positive impact on resource use to treat symptoms of allergic rhinitis. The objective of this study was to compare differences in respiratory medication utilization by pediatric patients with asthma and allergic rhinitis who initiated montelukast or ICS therapy. METHODS Design and Data Sources A 24-month, retrospective, longitudinal cohort analysis was performed using 2 data sources. The Medco Health Solutions Inc Information Warehouse (Franklin Lakes, NJ) (⬎65 million lives), a large, geographically diverse administrative claims database that contains 3 years of longitudinal phar-
129
macy claims data, was used for prescription, physician, and patient identification, and published average wholesale prices (AWPs) were used to determine medication costs. The Medco cohort is evenly distributed by sex (50% male and 50% female) and represents covered lives from insurers, managed care organizations, employers, and retirement and government plans. The mean age of the members of the Medco Health Solutions database is similar to that of the US population (mean age of Medco Health Solutions members, 36 years; mean age of US population according to census data, 35 years). The age distribution is as follows: 1 to 19 years, 22%; 20 to 44 years, 28%; 45 to 64 years, 28%; and 65 years or older, 22%. Only data from clients who authorized use of their data for secondary research purposes were used. Approximately 1,300,000 unique practitioners have prescribed medications to patients covered in the data warehouse. This database is unique in that it contains true longitudinal, patient-centric information. Patient prescription continuity is maintained for the following patient subgroups, resulting in an accurate longitudinal view of prescribing behavior: patients who switch pharmacies (both within chain and between chains), patients who switch from retail to mail service (and vice versa), patients who receive prescriptions from multiple physicians, patients who switch health plans or employers but maintain pharmacy coverage through Medco Health Solutions, and patients who change their names. The AWP was the second data source used to evaluate the cost of asthma therapy in the prestudy and poststudy periods. Although it is recognized that the AWP may not reflect actual prices paid for prescriptions, it provides a constant costsupporting comparison across all agents. The AWP is also not subject to variation by benefit design, manufacturer discounts, or pharmaceutical benefit manager contracts. Other widely varying cost factors, such as member copay and benefit design structure, were not considered in this analysis. Cohort The analyzed cohort consisted of patients who were continuously benefit eligible, were new to asthma controller therapy, had initiated controller therapy (ⱖ2 consecutive fills) with either montelukast or an ICS from February 1, 1999, to January 31, 2002, and had 1 or more antihistamine or nasal steroid prescriptions during the 12 months before the first asthma controller prescription. Use of allergy medications defined presence of concomitant allergic rhinitis. Only patients initiating monotherapy and younger than 16 years were included in the current study, whereas patients who initiated long-acting -agonist therapy were not included, because long-acting -agonist monotherapy is not recommended for treatment of children with asthma and allergic rhinitis (Fig 1). Endpoints The mean monthly costs of rescue or acute asthma medications (short-acting -agonists, antibiotics, oral corticosteroids), allergy medications, including nasal corticosteroids and prescription antihistamines, and other respiratory medi-
130
cations (xanthines, mast cell stabilizers, other leukotrienes) for the 12 months before and after new controller therapy initiation and difference in costs (before compared with after the index date) were calculated. Evaluation of rescue and acute asthma medication use has been found to be a good surrogate for asthma control in addition to other measures.14 Allergy medication use with its seasonal patterns can also be considered as a proxy for allergy control. Medication costs, as opposed to prescription rates only, are commonly used metrics and particularly important and useful in pre-post study designs. Additional endpoints were the number of days with asthma rescue agents, short-acting -agonists, oral corticosteroids, allergy medication, and nasal steroids in the prestudy and poststudy periods. Statistical Analyses All statistical analyses were performed using SAS statistical software, version 8 (SAS Institute Inc, Cary, NC). Three analyses of outcomes were conducted. First, we created a probabilistic matched cohort of patients to compare outcomes in each group, ICS vs montelukast, as well as a post hoc analysis of fluticasone vs montelukast. The second analysis was a multivariate Tobit modeling of the outcomes. The third analysis evaluated the likelihood of increased or decreased days on therapy before and after controller introduction. Probabilistic matching was a 2-step process that included adjustment for observable and unobservable characteristics. Step 1 included a selection of observable variables. Univariate and bivariate analyses were first conducted to evaluate clinical relevance of potential variables and determine appropriate candidates for matching. Observed explanatory variables included age, sex, asthma severity index, prescriber specialty, index controller date, days undergoing controller therapy during study period, and prior 12-month antihistamine use. Prescriber specialty was defined based on the prescriber of the initial index controller prescription as follows: (1) specialists (allergy, immunology, internal medicine, pulmonology), (2) general practitioners (pediatrics, family practice), and (3) all others. Asthma severity was computed using an asthma severity algorithm15 based on prescription history for the 12 months preceding the index controller date, where severity is assigned based on the number of canisters of -agonist used and the number of courses of oral corticosteroids within a year. This algorithm has been validated for use in administrative databases studying asthma-related outcomes (resource utilization). Step 2 of the probabilistic matching adjusted for unobservable patient characteristics. The propensity scoring method is recommended to control for selection bias associated with controller agent selection.16 Since a number of clinical and demographic factors, some unobservable, may play a role in a prescriber’s decision to select one therapy regimen vs another, patients who were assigned and not randomly selected into treatment groups may differ (eg, may be more severe or older). Multiple logistic regression analysis was used to develop propensity scores that predicted the proba-
ANNALS OF ALLERGY, ASTHMA & IMMUNOLOGY
Figure 1. Cohort selection diagram. MON indicates montelukast; ICS, inhaled corticosteroid; LABA, long-acting -agonist; and AR, allergic rhinitis.
bility of being assigned to one treatment group vs another based on known baseline characteristics, including age, sex, asthma severity index, prescriber specialty, index controller date, and prior 12-month antihistamine and nasal steroid use. Use of a combination of observable characteristics predicts differences in unobservable characteristics.
VOLUME 95, AUGUST, 2005
Following this 2-step process using univariate and logistic procedures, patients from the 2 groups (montelukast vs ICS) were matched one to one based on propensity score, age, prior 12-month antihistamine and nasal steroid use, and days undergoing controller therapy during the study period. The matched patients were then compared on prestudy (12 months
131
before initiation of controller) vs poststudy (12 months after initiation of controller) differences in unadjusted costs of rescue or acute medications for asthma, allergy, and other respiratory medications. These results were further confirmed using the multivariate Tobit model. The Tobit analysis,17 a regression method that adjusts regression for zero-cost items commonly seen in populations that do not incur any health care costs in specific categories of interest, was used to adjust the cost data, since for many cost categories there was a substantial proportion of patients with $0 cost in that category. Variables included in the Tobit modeling were age, sex, asthma severity index, prescriber specialty, index controller date, and prior 12month antihistamine and nasal steroid use. In the final analysis, patients were evaluated by the amount of asthma rescue medication or allergy medications used in the prestudy and poststudy periods and were categorized into more days vs fewer or the same number of days (⫾1 day). Logistic regression analysis was used to determine the likelihood of experiencing more days, compared with fewer or the same number of days, in the poststudy period compared with the prestudy period by treatment (ICS vs montelukast), age, sex, prescriber specialty, and patient severity category. RESULTS Patients A total of 3,217 pediatric patients younger than 16 years with asthma and allergic rhinitis initiated controller therapy with either an ICS or montelukast for asthma according to the
1999 to 2000 Medco Health Solutions database. Baseline characteristics of age, sex, propensity score, percentage of patients with allergic rhinitis, and use of allergy medications and asthma rescue medications, including antibiotics and oral corticosteroids, before one-to-one matching are presented in Table 1. Of the 3,217 pediatric patients, 1,236 (618 in the ICS group and 618 in the montelukast group) were included in this analysis after one-to-one matching. The average age was 9.9 years, approximately 60% were male, and more than 75% had asthma that was classified as mild; 5% or less of patients had severe asthma as defined by the asthma severity score. There were no significant differences between groups in any baseline characteristic after matching. Within the ICS group, most patients received fluticasone (46%), followed by beclomethasone (31%), triamcinolone (13%), budesonide (7%) and other ICSs (3%). Costs for Asthma and Allergic Rhinitis Medications The prestudy compared with poststudy mean change per member cost of all asthma and allergy medications, exclusive of the index drugs, prescribed to pediatric patients was significantly less in the montelukast group ($5.55) compared with the ICS group ($12.08) (P ⬍ .001) (Table 2). The prestudy compared with poststudy difference for all asthma rescue medication costs was also significantly less in the montelukast group ($0.94) compared with the ICS group ($3.82) (P ⫽ .003). This difference was primarily due to the difference in use of short-acting -agonists in the montelukast group ($1.79) compared with the ICS group ($3.34) (P ⫽ .008) (Table 2). The overall cost of all allergy media-
Table 1. Characteristics of Patient Populations Before One-to-One Matching Characteristics Mean age, y Sex, No. (%) Male Female Mean propensity score* Asthma severity,† No. (%) Mild Moderate Severe Asthma rescue medications, mean days‡ Short-acting -agonists Antibiotics Oral corticosteroids Allergy medications, mean days‡ Antihistamines Nasal steroids
Montelukast patients (n ⴝ 1,287)
ICS patients (n ⴝ 1,930)*
P value
9.9
10.2
.004 .77
772 (60) 515 (40) 0.44
1,167 (60) 763 (40) 0.41
1,047 (81) 183 (14) 57 (4)
1,577 (81) 263 (14) 90 (5)
37.7 24.1 10.8
38.2 22.1 10.7
.80 .09 .88
90.1 23.2
71.0 24.8
⬍.001 .29
⬍.001 .26
Abbreviation: ICS, inhaled corticosteroid. * The propensity score is the predicted probability of receiving montelukast based on patients characteristics.25 † Asthma severity defined by the asthma severity algorithm24 based on prescription history. ‡ During the 12 months before asthma controller introduction.
132
ANNALS OF ALLERGY, ASTHMA & IMMUNOLOGY
Table 2. Mean Difference Between Precontroller and Postcontroller Costs of Concomitant Asthma Drug Therapy and Allergy Medications (US Dollars per Member per Month) Cost category
Montelukast patients (n ⴝ 618)
ICS patients (n ⴝ 618)
P value
5.55 0.94 1.79 ⫺0.65* ⫺0.20 5.29 4.44 0.85 ⫺0.68
12.08 3.82 3.34 ⫺0.51 ⫺0.04 10.06 7.43 2.63 ⫺1.79
⬍.001 .003 .008 .10 .07 ⬍.001 .004 ⬍.001 .046
Combined cost Rescue or acute asthma medications Short-acting -agonists Antibiotics Oral corticosteroids Allergy medications Antihistamines Nasal steroids Other respiratory medications†
Abbreviation: ICS, inhaled corticosteroid. * Negative values represent decreased costs compared with the period before starting asthma controller therapy. † Xanthines, mast cell stabilizers, and leukotrienes other than montelukast.
tions was significantly less in the montelukast group ($5.29) compared with the ICS group ($10.06) (P ⬍ .001) (Table 2). The monthly cost difference associated with antihistamines and nasal steroids was significantly less in the montelukast group compared with the ICS group (Table 2). The cost of antibiotics and oral corticosteroids decreased in both the montelukast group and the ICS group. The differences were not statistically significant (Table 2). The cost of other respiratory medications, including xanthines, mast cell stabilizers, and leukotrienes other than montelukast, was reduced in both groups but was significantly less in the ICS group (⫺$1.79) compared with the montelukast group (⫺$0.68) (P ⫽ .046) (Table 2). The Tobit-adjusted monthly costs between prestudy and poststudy periods by asthma controller therapy confirmed the significant reduction in costs of other medications associated with the use of montelukast vs ICSs in this pediatric patient population. The cost was $3.61 in the montelukast group and $6.07 in the ICS group for total rescue agents (P ⬍ .001) and $8.04 in the montelukast group and $12.00 in the ICS group for allergy medications (P ⬍ .001), favoring montelukast. A post hoc analysis evaluated outcomes in a matched cohort of patients who initiated fluticasone (n ⫽ 282) or montelukast (n ⫽ 282) therapy. Results suggest that combined costs of other asthma and allergy medications increased at a numerically lower rate in montelukast group ($6.18 vs $10.1, P ⫽ .18), whereas the costs of allergy medications increased significantly less in the montelukast group ($4.50 vs $9.45, P ⫽ .005), which was primarily driven by less use of nasal steroids (P ⫽ .001) in the post– controller introduction period. Post– controller introduction antibiotic use remained stable in both groups, whereas oral corticosteroid use declined in both groups (⫺$0.19 in montelukast group, ⫺$0.11 in fluticasone group, P ⫽ .25). These results suggest similar levels of asthma control in both groups and a slight economic advantage for montelukast.
VOLUME 95, AUGUST, 2005
Days of Therapy Patients taking ICSs (compared with patients taking montelukast) were 66% more likely to have a greater number of days of therapy with asthma rescue agents and a greater number of days with use of allergy medications in the poststudy period when compared with the prestudy period. The increase in asthma rescue agents was due mainly to increased use of short-acting -agonists. Compared with patients taking montelukast, patients taking ICSs were 2.3 times more likely to have a greater number of days taking short-acting -agonists (Table 3). The odds of using oral corticosteroids and antibiotics as acute asthma medications were similar for the 2 treatment groups. For both groups, the odds of having more days of rescue agent therapy in the poststudy period increased as age increased. Additionally, there was approximately a 5% increase in the odds of using more antibiotics for each increase in 1 year of age (data not shown) in the poststudy period compared with the prestudy period.
Table 3. Adjusted Logistic Regression Analysis for the Likelihood of Experiencing More Days With Asthma or Allergy Therapy in the Poststudy Compared With the Prestudy Period* Therapeutic agent Asthma rescue agents All Short-acting -agonists Allergy medication All Nasal steroids
OR for ICS vs montelukast (95% CI)† 1.66 (1.32–2.10) 2.31 (1.83–2.92 1.27 (1.02–1.59) 1.78 (1.40–2.27)
Abbreviations: CI, confidence interval; ICS, inhaled corticosteroid; OR, odds ratio. * Models adjusted for age, sex, prescriber specialty, and asthma severity. † Differences were significant between treatment groups (P ⫽ .05).
133
DISCUSSION This retrospective pre-post cohort study based on a large pharmacy claims database analyzed the cost of rescue or acute asthma medication and allergy medications in patients 16 years and younger with a history of allergy medication use and who were new to controller medication for their asthma. Patients who were taking montelukast had significantly lower overall cost increases for asthma rescue medications and allergy medications compared with patients taking ICSs during the 12-month poststudy period vs the 12-month prestudy period. The overall costs of antibiotics and oral corticosteroids were numerically less in the montelukast group compared with the ICS group, but these differences were not statistically significant. Patients taking ICSs as their controller therapy were 66% more likely to have more days of asthma rescue agent use after initiation (odds ratio of 2.31 for having more days taking short-acting -agonists compared with montelukast). The use of allergy medications was also different between groups; patients taking ICSs were 27% more likely to have more days of nasal steroid use for their allergies than patients taking montelukast. Because montelukast is approved for treatment of asthma and allergic rhinitis, these findings broadly support the concept of “one airway” disease in that asthma and allergic rhinitis are manifestations of a single disorder and that treating one disease can affect the treatment required to control the other. To our knowledge, this is the first cost analysis from a US pharmacy database that investigates asthma and allergy medication use in pediatric patients with comorbid disease. Previous retrospective studies have evaluated the costs of montelukast vs ICSs, assessing multiple parameters of comparison.18 –21 A similar study by Bukstein and colleagues22 in new users of controller therapy showed that patients undergoing montelukast therapy required fewer -agonist prescriptions compared with those taking fluticasone. Similar to our findings, the study by Bukstein et al22 showed that pediatric patients taking montelukast required fewer -agonist prescriptions in addition to prescriptions for other asthma and allergy–related medications compared with fluticasone. In the current study, asthmatic patients had prescriptions for allergy medications before initiation of controller therapy for asthma. Because of this prescription history, we considered these patients to have had concomitant allergic rhinitis. Although our analysis selected patients with both diseases, it is probably representative of the larger population of asthmatic patients. Allergic rhinitis and asthma are often comorbid conditions. At least 85% of patients with asthma reported symptoms of allergic rhinitis, and even patients who report no allergy symptoms have evidence of chronic nasal inflammation.23 One disease affects the other in the increased cost of services related to pediatric patient care2 and the cost of medication.3 Respiratory-related medical costs in children with asthma and allergic rhinitis may be almost twice those incurred by patients with asthma.7
134
The data from a large pharmacy system on prescription fills represent a relatively effective means of evaluating medication needs in a patient population, because the data are not subject to reporting bias24 and may represent “real-world” effectiveness of medications related to a particular disease state. However, utilization of data contained in pharmacy systems lacks the randomization of controlled clinical trials and may be subject to confounding errors. To reduce confounding, we used the matching and propensity scores method16 to reduce imbalance in the baseline characteristics of patients in the montelukast and ICS groups. To be accurate, the propensity score must contain variables that accurately describe the model and are predictors of the outcome.25 In this study, the propensity score contained variables for age, sex, asthma severity index, prescriber specialty, index controller date, and prior 12-month antihistamine and nasal steroid use. Adjusting the patients by propensity score ensured that patients would be accurately matched between cohort groups as evidenced by the lack of significant differences in observed baseline characteristics after matching. However, caution should be exercised when drawing conclusions, because causal relationships cannot be determined from retrospective studies. There are other limitations of analyses of pharmacy claims databases. Patients in this study were identified by prescriptions filled for allergy medications. It is possible that asthmatic children were being treated with over-the-counter drugs for their allergic symptoms and were not captured in this analysis. Allergic rhinitis is frequently undertreated in children,26 and patients who sought physician counseling on their disease may have had more severe allergies. However, asthmatic children with comorbid allergic rhinitis use more health care services than other children,2,3 so it is likely that the patients included in this analysis are typical of many asthmatic children with allergic rhinitis. Most patients included in this analysis were mild asthmatic patients as judged by -agonist use. Therefore, the selection criteria for this study were probably not biased toward patients with more severe disease, and the results are more likely representative of the larger pediatric asthmatic population. Total pharmacy cost analysis, which would include the costs of montelukast and ICS, was not pursued because of wide variability in amounts both charged and paid for any particular medication. Variables that are directly relevant to any cost analysis include contractual details related to patients, employers and health plans, pharmacy benefit management services, and the manufacturers of the drugs, since discounts and cost sharing would shift the economic impact for any individual stakeholder. No analysis performed to date purports to adjust for all of these variables. In addition, pharmacy costs are only a small proportion of total asthmarelated costs (including asthma-related physician visits and hospitalizations). The primary purpose of this analysis was to use rescue or acute asthma and allergy medication use and costs as proxies for asthma control, which is also influenced
ANNALS OF ALLERGY, ASTHMA & IMMUNOLOGY
by patient preferences, corresponding compliance, and access to health care. The shared pathophysiology of asthma and allergic rhinitis and the impact the presence of one disease has on the other in terms of morbidity and health care costs make the use of one medication that treats both diseases an attractive form of therapy. Montelukast is the only medication that has one formulation (oral tablet) approved for both indications and is proven efficacious in treating asthma and allergic rhinitis.27–33 The current study clearly demonstrated that introduction of montelukast in children with asthma and allergic rhinitis was associated with significantly lower increase in the overall costs of rescue or acute asthma medications and allergy medications compared with introduction of ICSs. These data suggest that montelukast use may be associated with better asthma control than ICS use in this population by reducing the need for asthma rescue medications and also reduced need for acute asthma and allergy therapies. ACKNOWLEDGMENTS We thank Carol A. Tozzi, PhD, for scientific input, writing, and editorial assistance and Arlene Swern, PhD, for statistical review.
12.
13. 14. 15. 16. 17. 18. 19. 20.
REFERENCES 1. Taylor WR, Newacheck PW. Impact of childhood asthma on health. Pediatrics. 1992;90:657– 662. 2. Grupp-Phelan J, Lozano P, Fishman P. Health care utilization and cost in children with asthma and selected comorbidities. J Asthma. 2001;38:363–373. 3. Thomas M, Zhang Q, Sazonov Kocevar V, Yin D, Price DB. Health care resource use by children with asthma and co-morbid allergic rhinitis in general practice in the United Kingdom. Pediatrics. 2005;115:129 –134. 4. Leynaert B, Neukrich F, Demoly P, Bousquet J. Epidemiologic evidence for asthma and rhinitis comorbidity. J Allergy Clin Immunol. 2000;106(5 suppl):S201–S205. 5. Pedersen PA, Weeke ER. Asthma and allergic rhinitis in the same patients. Allergy. 1983;38:25–29. 6. Kapsali T, Horowitz E, Diemer F, Togias A. Rhinitis is ubiquitous in allergic asthmatics. J Allergy Clin Immunol. 1997;99: S138. 7. Yawn BP, Yunginger JW, Wollan PC, et al. Allergic rhinitis in Rochester Minnesota residents with asthma: frequency and impact on health care charges. J Allergy Clin Immunol. 1999;103: 54 –59. 8. Simons FE. Allergic rhinobronchitis: the asthma-allergic rhinitis link. J Allergy Clin Immunol. 1999;104:534 –540. 9. Meltzer EO. Role for cysteinyl leukotriene receptor antagonist therapy in asthma and their potential role in allergic rhinitis based on the concept of “one linked airway disease”. Ann Allergy Asthma Immunol. 2000;84:176 –185. 10. Bousquet J, Van Cauwenberge P, Khaltaev N. Allergic rhinitis and its impact on asthma. J Allergy Clin Immunol. 2001; 108(suppl):S147–S334. 11. Welsh PW, Stricker WE, Chu C-P, et al. Efficacy of beclomethasone nasal solution, flunisolide, and cromolyn in re-
VOLUME 95, AUGUST, 2005
21. 22.
23.
24. 25. 26. 27. 28. 29.
30.
lieving symptoms of ragweed allergy. Mayo Clin Proc. 1987; 62:125–134. Foresi A, Pelucchi A, Gherson G, et al. Once daily intranasal fluticasone propionate (200 micrograms) reduces nasal symptoms and inflammation but also attenuates the increase in bronchial hyperresponsiveness during pollen season in allergic rhinitis. J Allergy Clin Immunol. 1996;98:274 –282. Greiff L, Andersson M, Svensson C, et al. Effects of orally inhaled budesonide in seasonal allergic rhinitis. Eur Respir J. 1998;11:1268 –1273. Vollmer WM, Markson LE, O’Connor E, et al. Association of asthma control with health care utilization. Am J Respir Crit Care Med. 2002;165:195–199. Leidy NK, Paramore LC, Wartrous, et al. Development of algorithm for scoring severity of asthma in administrative databases. Value Health. 1999;2:394. Rubin DB. Estimating causal effects from large data sets using propensity scores. Ann Intern Med. 1997;127(8 pt 2):757–763. Maddala GS. Introduction to Econometrics. New York, NY: Macmillan Publishing Company; 1988. Allen-Ramey FC, Duong PT, Goodman DC, et al. Outcomes of patients treated with ICS propionate or montelukast sodium. Managed Care Int. 2003;16:30 –35. Price DB, Ben-Joseph RH, Zhang Q. Changes in asthma drug therapy costs for patients receiving chronic montelukast therapy in the UK. Respir Med. 2001;95:83– 89. Stempel DA, Mauskopf J, McLaughlin T, et al. Comparison of asthma costs in patients starting fluticasone propionate compared to patients starting montelukast. Respir Med. 2001;95: 227–234. Stempel DA, O’Donnell JC, Meyer JW. Inhaled corticosteroids plus salmeterol or montelukast: effects on resource utilization and costs. J Allergy Clin Immunol. 2002;109:433– 439. Bukstein DA, Henk HJ, Luskin AT. A comparison of asthmarelated expenditures for patients started on montelukast versus fluticasone propionate as monotherapy. Clin Ther. 2001;23: 1589 –1600. Gaga M, Lanbrou P, Papaageorgiou N, et al. Eosinophils are a feature of upper and lower airway pathology in no-atopic asthma, irrespective of the presence of rhinitis. Clin Exp Allergy. 2000;30:663– 669. Schmier JK, Leidy NK. The complexity of treatment adherence in adults with asthma: challenges and opportunities. J Asthma. 1998;35:455– 472. Michels KB, Braunwald E. Estimating treatment effects from observational data: dissonant and resonant notes from the SYMPHONY trials. JAMA. 2002;287:3130 –3132. Stone KD. Atopic diseases of childhood. Curr Opin Pediatr. 2002;14:634 – 646. Reiss TF, Chervinsky P, Dockhorn RJ, et al. Montelukast, a once-daily leukotriene receptor antagonist, in the treatment of chronic asthma. Arch Intern Med. 1998;158:1213–1220. Knorr B, Matz J, Bernstein JA, et al. Montelukast for chronic asthma in 6- to 14-year-old children. JAMA. 1998;279: 1181–1186. Knorr B, Franchi LM, Bisgaard H, et al. Montelukast, a leukotriene receptor antagonist, for the treatment of persistent asthma in children aged 2 to 5 years. Pediatrics. 2001; 108(3):E48. Philip G, Malmstrom K, Hampel FC Jr, et al. Montelukast for
135
treating seasonal allergic rhinitis: a randomized, double-blind, placebo-controlled trial performed in the spring. Clin Exp Allergy. 2002;32:1020 –1028. 31. Nayak AS, Philip G, Lu S, et al. Efficacy and tolerability of montelukast alone or in combination with loratadine in seasonal allergic rhinitis: a multicenter, randomized, double-blind, placebo-controlled trial performed in the fall. Annals Allergy Asthma Immunol. 2002;88:592– 600. 32. van Adelsberg J, Philip G, Pedinoff AJ, et al. Montelukast improves symptoms of seasonal allergic rhinitis over a 4-week treatment period. Allergy. 2003;58:1268 –1276. 33. van Adelsberg J, Philip G, LaForce CF, et al. Randomized
136
controlled trial evaluating the clinical benefit of montelukast for treating spring seasonal allergic rhinitis. Ann Allergy Asthma Immunol. 2003;90:214 –222. Requests for reprints should be addressed to: Allan Luskin, MD Pediatric and Adult Asthma and Allergic Diseases Pediatric Lung Diseases 10 Tower Dr Sun Prairie, WI 53590 E-mail: [email protected]
ANNALS OF ALLERGY, ASTHMA & IMMUNOLOGY
Background: Asthma and allergic rhinitis are frequently comorbid conditions. Montelukast is effective in treating both diseases and may reduce total medication use among children with asthma and allergic rhinitis. Objective: To determine the differences in respiratory and allergy medication use and costs, as proxies for control, in pediatric patients with asthma and allergy who initiated asthma controller therapy. Methods: A 24-month, retrospective, pre-post cohort study using a pharmacy claims database of children (age ⬍16 years) with 2 or more consecutive asthma controller prescriptions and 1 or more allergy prescription (within 12 months before initial controller prescription). Children taking inhaled corticosteroids (ICSs) and montelukast were matched one to one based on age, days of prior allergic rhinitis therapy supply, duration of controller therapy, and propensity score. Differences in costs of rescue or acute asthma medications, prescription allergy medications, other respiratory medications, and the number of days of rescue or acute asthma medication use and allergy medication use were calculated. Results: A total of 1,236 children were matched into ICS and montelukast groups (n ⫽ 618 each). Montelukast patients had a smaller cost increase overall compared with ICS patients (combined cost for rescue or acute asthma medications, allergy medications, and other respiratory medications: $5.55 vs $12.08, P ⬍ .001). Cost increase for rescue or acute asthma medications was significantly lower in the montelukast group ($0.94 vs $3.82, P ⫽ .003). The cost increase for allergy medications ($5.29 vs $10.06, P ⬍ .001) was also significantly lower in the montelukast group. Patients taking montelukast also had fewer days of therapy with asthma rescue medication and allergy medication compared with patients taking ICSs. Conclusions: Initiating therapy with montelukast was associated with better asthma and allergy control demonstrated via lower increase in use and costs of asthma rescue and allergy medications compared with initiating ICS therapy. Ann Allergy Asthma Immunol. 2005;95:129–136.
INTRODUCTION Asthma is the most common chronic illness in childhood.1 Children with asthma use significantly more health care services than children in the general population, and among children with asthma, those with allergic rhinitis account for a large proportion of health care utilization in terms of outpatient care, pharmacy, and urgent care.2,3 Asthma and allergic rhinitis are frequently comorbid conditions.4 The prevalence of nasal allergy symptoms has been shown to range from 78%5 to more than 90%6 in patients with allergic asthma. This association is particularly evident in young patients.7 Epidemiologic, clinical, and pathophysiologic studies suggest that allergic rhinitis and asthma may be considered manifestations of a single disease.8 –10 Treatment of allergic rhinitis has been shown to improve symptoms of seasonal asthma,11,12 and high-dose inhaled corticosteroids
* Respiratory Institute, Dean Medical Center, and University of Wisconsin, Madison, Wisconsin. † Merck & Co Inc, Whitehouse Station, New Jersey. This study was supported by a grant from Merck and Co Inc. Received for publication November 4, 2004. Accepted for publication in revised form January 24, 2005.
VOLUME 95, AUGUST, 2005
(ICSs) used to treat subclinical asthma improve allergic rhinitis symptoms.13 However, despite associations between the 2 disorders, allergic rhinitis and asthma are generally treated as separate entities. Montelukast, a potent antagonist of the type 1 cysteinyl leukotriene receptor, is approved for treatment of both allergic rhinitis and asthma in the same oral tablet preparation. Since studies suggest that there is a clinically relevant association between asthma and allergic rhinitis, montelukast used as monotherapy for asthma may have a positive impact on resource use to treat symptoms of allergic rhinitis. The objective of this study was to compare differences in respiratory medication utilization by pediatric patients with asthma and allergic rhinitis who initiated montelukast or ICS therapy. METHODS Design and Data Sources A 24-month, retrospective, longitudinal cohort analysis was performed using 2 data sources. The Medco Health Solutions Inc Information Warehouse (Franklin Lakes, NJ) (⬎65 million lives), a large, geographically diverse administrative claims database that contains 3 years of longitudinal phar-
129
macy claims data, was used for prescription, physician, and patient identification, and published average wholesale prices (AWPs) were used to determine medication costs. The Medco cohort is evenly distributed by sex (50% male and 50% female) and represents covered lives from insurers, managed care organizations, employers, and retirement and government plans. The mean age of the members of the Medco Health Solutions database is similar to that of the US population (mean age of Medco Health Solutions members, 36 years; mean age of US population according to census data, 35 years). The age distribution is as follows: 1 to 19 years, 22%; 20 to 44 years, 28%; 45 to 64 years, 28%; and 65 years or older, 22%. Only data from clients who authorized use of their data for secondary research purposes were used. Approximately 1,300,000 unique practitioners have prescribed medications to patients covered in the data warehouse. This database is unique in that it contains true longitudinal, patient-centric information. Patient prescription continuity is maintained for the following patient subgroups, resulting in an accurate longitudinal view of prescribing behavior: patients who switch pharmacies (both within chain and between chains), patients who switch from retail to mail service (and vice versa), patients who receive prescriptions from multiple physicians, patients who switch health plans or employers but maintain pharmacy coverage through Medco Health Solutions, and patients who change their names. The AWP was the second data source used to evaluate the cost of asthma therapy in the prestudy and poststudy periods. Although it is recognized that the AWP may not reflect actual prices paid for prescriptions, it provides a constant costsupporting comparison across all agents. The AWP is also not subject to variation by benefit design, manufacturer discounts, or pharmaceutical benefit manager contracts. Other widely varying cost factors, such as member copay and benefit design structure, were not considered in this analysis. Cohort The analyzed cohort consisted of patients who were continuously benefit eligible, were new to asthma controller therapy, had initiated controller therapy (ⱖ2 consecutive fills) with either montelukast or an ICS from February 1, 1999, to January 31, 2002, and had 1 or more antihistamine or nasal steroid prescriptions during the 12 months before the first asthma controller prescription. Use of allergy medications defined presence of concomitant allergic rhinitis. Only patients initiating monotherapy and younger than 16 years were included in the current study, whereas patients who initiated long-acting -agonist therapy were not included, because long-acting -agonist monotherapy is not recommended for treatment of children with asthma and allergic rhinitis (Fig 1). Endpoints The mean monthly costs of rescue or acute asthma medications (short-acting -agonists, antibiotics, oral corticosteroids), allergy medications, including nasal corticosteroids and prescription antihistamines, and other respiratory medi-
130
cations (xanthines, mast cell stabilizers, other leukotrienes) for the 12 months before and after new controller therapy initiation and difference in costs (before compared with after the index date) were calculated. Evaluation of rescue and acute asthma medication use has been found to be a good surrogate for asthma control in addition to other measures.14 Allergy medication use with its seasonal patterns can also be considered as a proxy for allergy control. Medication costs, as opposed to prescription rates only, are commonly used metrics and particularly important and useful in pre-post study designs. Additional endpoints were the number of days with asthma rescue agents, short-acting -agonists, oral corticosteroids, allergy medication, and nasal steroids in the prestudy and poststudy periods. Statistical Analyses All statistical analyses were performed using SAS statistical software, version 8 (SAS Institute Inc, Cary, NC). Three analyses of outcomes were conducted. First, we created a probabilistic matched cohort of patients to compare outcomes in each group, ICS vs montelukast, as well as a post hoc analysis of fluticasone vs montelukast. The second analysis was a multivariate Tobit modeling of the outcomes. The third analysis evaluated the likelihood of increased or decreased days on therapy before and after controller introduction. Probabilistic matching was a 2-step process that included adjustment for observable and unobservable characteristics. Step 1 included a selection of observable variables. Univariate and bivariate analyses were first conducted to evaluate clinical relevance of potential variables and determine appropriate candidates for matching. Observed explanatory variables included age, sex, asthma severity index, prescriber specialty, index controller date, days undergoing controller therapy during study period, and prior 12-month antihistamine use. Prescriber specialty was defined based on the prescriber of the initial index controller prescription as follows: (1) specialists (allergy, immunology, internal medicine, pulmonology), (2) general practitioners (pediatrics, family practice), and (3) all others. Asthma severity was computed using an asthma severity algorithm15 based on prescription history for the 12 months preceding the index controller date, where severity is assigned based on the number of canisters of -agonist used and the number of courses of oral corticosteroids within a year. This algorithm has been validated for use in administrative databases studying asthma-related outcomes (resource utilization). Step 2 of the probabilistic matching adjusted for unobservable patient characteristics. The propensity scoring method is recommended to control for selection bias associated with controller agent selection.16 Since a number of clinical and demographic factors, some unobservable, may play a role in a prescriber’s decision to select one therapy regimen vs another, patients who were assigned and not randomly selected into treatment groups may differ (eg, may be more severe or older). Multiple logistic regression analysis was used to develop propensity scores that predicted the proba-
ANNALS OF ALLERGY, ASTHMA & IMMUNOLOGY
Figure 1. Cohort selection diagram. MON indicates montelukast; ICS, inhaled corticosteroid; LABA, long-acting -agonist; and AR, allergic rhinitis.
bility of being assigned to one treatment group vs another based on known baseline characteristics, including age, sex, asthma severity index, prescriber specialty, index controller date, and prior 12-month antihistamine and nasal steroid use. Use of a combination of observable characteristics predicts differences in unobservable characteristics.
VOLUME 95, AUGUST, 2005
Following this 2-step process using univariate and logistic procedures, patients from the 2 groups (montelukast vs ICS) were matched one to one based on propensity score, age, prior 12-month antihistamine and nasal steroid use, and days undergoing controller therapy during the study period. The matched patients were then compared on prestudy (12 months
131
before initiation of controller) vs poststudy (12 months after initiation of controller) differences in unadjusted costs of rescue or acute medications for asthma, allergy, and other respiratory medications. These results were further confirmed using the multivariate Tobit model. The Tobit analysis,17 a regression method that adjusts regression for zero-cost items commonly seen in populations that do not incur any health care costs in specific categories of interest, was used to adjust the cost data, since for many cost categories there was a substantial proportion of patients with $0 cost in that category. Variables included in the Tobit modeling were age, sex, asthma severity index, prescriber specialty, index controller date, and prior 12month antihistamine and nasal steroid use. In the final analysis, patients were evaluated by the amount of asthma rescue medication or allergy medications used in the prestudy and poststudy periods and were categorized into more days vs fewer or the same number of days (⫾1 day). Logistic regression analysis was used to determine the likelihood of experiencing more days, compared with fewer or the same number of days, in the poststudy period compared with the prestudy period by treatment (ICS vs montelukast), age, sex, prescriber specialty, and patient severity category. RESULTS Patients A total of 3,217 pediatric patients younger than 16 years with asthma and allergic rhinitis initiated controller therapy with either an ICS or montelukast for asthma according to the
1999 to 2000 Medco Health Solutions database. Baseline characteristics of age, sex, propensity score, percentage of patients with allergic rhinitis, and use of allergy medications and asthma rescue medications, including antibiotics and oral corticosteroids, before one-to-one matching are presented in Table 1. Of the 3,217 pediatric patients, 1,236 (618 in the ICS group and 618 in the montelukast group) were included in this analysis after one-to-one matching. The average age was 9.9 years, approximately 60% were male, and more than 75% had asthma that was classified as mild; 5% or less of patients had severe asthma as defined by the asthma severity score. There were no significant differences between groups in any baseline characteristic after matching. Within the ICS group, most patients received fluticasone (46%), followed by beclomethasone (31%), triamcinolone (13%), budesonide (7%) and other ICSs (3%). Costs for Asthma and Allergic Rhinitis Medications The prestudy compared with poststudy mean change per member cost of all asthma and allergy medications, exclusive of the index drugs, prescribed to pediatric patients was significantly less in the montelukast group ($5.55) compared with the ICS group ($12.08) (P ⬍ .001) (Table 2). The prestudy compared with poststudy difference for all asthma rescue medication costs was also significantly less in the montelukast group ($0.94) compared with the ICS group ($3.82) (P ⫽ .003). This difference was primarily due to the difference in use of short-acting -agonists in the montelukast group ($1.79) compared with the ICS group ($3.34) (P ⫽ .008) (Table 2). The overall cost of all allergy media-
Table 1. Characteristics of Patient Populations Before One-to-One Matching Characteristics Mean age, y Sex, No. (%) Male Female Mean propensity score* Asthma severity,† No. (%) Mild Moderate Severe Asthma rescue medications, mean days‡ Short-acting -agonists Antibiotics Oral corticosteroids Allergy medications, mean days‡ Antihistamines Nasal steroids
Montelukast patients (n ⴝ 1,287)
ICS patients (n ⴝ 1,930)*
P value
9.9
10.2
.004 .77
772 (60) 515 (40) 0.44
1,167 (60) 763 (40) 0.41
1,047 (81) 183 (14) 57 (4)
1,577 (81) 263 (14) 90 (5)
37.7 24.1 10.8
38.2 22.1 10.7
.80 .09 .88
90.1 23.2
71.0 24.8
⬍.001 .29
⬍.001 .26
Abbreviation: ICS, inhaled corticosteroid. * The propensity score is the predicted probability of receiving montelukast based on patients characteristics.25 † Asthma severity defined by the asthma severity algorithm24 based on prescription history. ‡ During the 12 months before asthma controller introduction.
132
ANNALS OF ALLERGY, ASTHMA & IMMUNOLOGY
Table 2. Mean Difference Between Precontroller and Postcontroller Costs of Concomitant Asthma Drug Therapy and Allergy Medications (US Dollars per Member per Month) Cost category
Montelukast patients (n ⴝ 618)
ICS patients (n ⴝ 618)
P value
5.55 0.94 1.79 ⫺0.65* ⫺0.20 5.29 4.44 0.85 ⫺0.68
12.08 3.82 3.34 ⫺0.51 ⫺0.04 10.06 7.43 2.63 ⫺1.79
⬍.001 .003 .008 .10 .07 ⬍.001 .004 ⬍.001 .046
Combined cost Rescue or acute asthma medications Short-acting -agonists Antibiotics Oral corticosteroids Allergy medications Antihistamines Nasal steroids Other respiratory medications†
Abbreviation: ICS, inhaled corticosteroid. * Negative values represent decreased costs compared with the period before starting asthma controller therapy. † Xanthines, mast cell stabilizers, and leukotrienes other than montelukast.
tions was significantly less in the montelukast group ($5.29) compared with the ICS group ($10.06) (P ⬍ .001) (Table 2). The monthly cost difference associated with antihistamines and nasal steroids was significantly less in the montelukast group compared with the ICS group (Table 2). The cost of antibiotics and oral corticosteroids decreased in both the montelukast group and the ICS group. The differences were not statistically significant (Table 2). The cost of other respiratory medications, including xanthines, mast cell stabilizers, and leukotrienes other than montelukast, was reduced in both groups but was significantly less in the ICS group (⫺$1.79) compared with the montelukast group (⫺$0.68) (P ⫽ .046) (Table 2). The Tobit-adjusted monthly costs between prestudy and poststudy periods by asthma controller therapy confirmed the significant reduction in costs of other medications associated with the use of montelukast vs ICSs in this pediatric patient population. The cost was $3.61 in the montelukast group and $6.07 in the ICS group for total rescue agents (P ⬍ .001) and $8.04 in the montelukast group and $12.00 in the ICS group for allergy medications (P ⬍ .001), favoring montelukast. A post hoc analysis evaluated outcomes in a matched cohort of patients who initiated fluticasone (n ⫽ 282) or montelukast (n ⫽ 282) therapy. Results suggest that combined costs of other asthma and allergy medications increased at a numerically lower rate in montelukast group ($6.18 vs $10.1, P ⫽ .18), whereas the costs of allergy medications increased significantly less in the montelukast group ($4.50 vs $9.45, P ⫽ .005), which was primarily driven by less use of nasal steroids (P ⫽ .001) in the post– controller introduction period. Post– controller introduction antibiotic use remained stable in both groups, whereas oral corticosteroid use declined in both groups (⫺$0.19 in montelukast group, ⫺$0.11 in fluticasone group, P ⫽ .25). These results suggest similar levels of asthma control in both groups and a slight economic advantage for montelukast.
VOLUME 95, AUGUST, 2005
Days of Therapy Patients taking ICSs (compared with patients taking montelukast) were 66% more likely to have a greater number of days of therapy with asthma rescue agents and a greater number of days with use of allergy medications in the poststudy period when compared with the prestudy period. The increase in asthma rescue agents was due mainly to increased use of short-acting -agonists. Compared with patients taking montelukast, patients taking ICSs were 2.3 times more likely to have a greater number of days taking short-acting -agonists (Table 3). The odds of using oral corticosteroids and antibiotics as acute asthma medications were similar for the 2 treatment groups. For both groups, the odds of having more days of rescue agent therapy in the poststudy period increased as age increased. Additionally, there was approximately a 5% increase in the odds of using more antibiotics for each increase in 1 year of age (data not shown) in the poststudy period compared with the prestudy period.
Table 3. Adjusted Logistic Regression Analysis for the Likelihood of Experiencing More Days With Asthma or Allergy Therapy in the Poststudy Compared With the Prestudy Period* Therapeutic agent Asthma rescue agents All Short-acting -agonists Allergy medication All Nasal steroids
OR for ICS vs montelukast (95% CI)† 1.66 (1.32–2.10) 2.31 (1.83–2.92 1.27 (1.02–1.59) 1.78 (1.40–2.27)
Abbreviations: CI, confidence interval; ICS, inhaled corticosteroid; OR, odds ratio. * Models adjusted for age, sex, prescriber specialty, and asthma severity. † Differences were significant between treatment groups (P ⫽ .05).
133
DISCUSSION This retrospective pre-post cohort study based on a large pharmacy claims database analyzed the cost of rescue or acute asthma medication and allergy medications in patients 16 years and younger with a history of allergy medication use and who were new to controller medication for their asthma. Patients who were taking montelukast had significantly lower overall cost increases for asthma rescue medications and allergy medications compared with patients taking ICSs during the 12-month poststudy period vs the 12-month prestudy period. The overall costs of antibiotics and oral corticosteroids were numerically less in the montelukast group compared with the ICS group, but these differences were not statistically significant. Patients taking ICSs as their controller therapy were 66% more likely to have more days of asthma rescue agent use after initiation (odds ratio of 2.31 for having more days taking short-acting -agonists compared with montelukast). The use of allergy medications was also different between groups; patients taking ICSs were 27% more likely to have more days of nasal steroid use for their allergies than patients taking montelukast. Because montelukast is approved for treatment of asthma and allergic rhinitis, these findings broadly support the concept of “one airway” disease in that asthma and allergic rhinitis are manifestations of a single disorder and that treating one disease can affect the treatment required to control the other. To our knowledge, this is the first cost analysis from a US pharmacy database that investigates asthma and allergy medication use in pediatric patients with comorbid disease. Previous retrospective studies have evaluated the costs of montelukast vs ICSs, assessing multiple parameters of comparison.18 –21 A similar study by Bukstein and colleagues22 in new users of controller therapy showed that patients undergoing montelukast therapy required fewer -agonist prescriptions compared with those taking fluticasone. Similar to our findings, the study by Bukstein et al22 showed that pediatric patients taking montelukast required fewer -agonist prescriptions in addition to prescriptions for other asthma and allergy–related medications compared with fluticasone. In the current study, asthmatic patients had prescriptions for allergy medications before initiation of controller therapy for asthma. Because of this prescription history, we considered these patients to have had concomitant allergic rhinitis. Although our analysis selected patients with both diseases, it is probably representative of the larger population of asthmatic patients. Allergic rhinitis and asthma are often comorbid conditions. At least 85% of patients with asthma reported symptoms of allergic rhinitis, and even patients who report no allergy symptoms have evidence of chronic nasal inflammation.23 One disease affects the other in the increased cost of services related to pediatric patient care2 and the cost of medication.3 Respiratory-related medical costs in children with asthma and allergic rhinitis may be almost twice those incurred by patients with asthma.7
134
The data from a large pharmacy system on prescription fills represent a relatively effective means of evaluating medication needs in a patient population, because the data are not subject to reporting bias24 and may represent “real-world” effectiveness of medications related to a particular disease state. However, utilization of data contained in pharmacy systems lacks the randomization of controlled clinical trials and may be subject to confounding errors. To reduce confounding, we used the matching and propensity scores method16 to reduce imbalance in the baseline characteristics of patients in the montelukast and ICS groups. To be accurate, the propensity score must contain variables that accurately describe the model and are predictors of the outcome.25 In this study, the propensity score contained variables for age, sex, asthma severity index, prescriber specialty, index controller date, and prior 12-month antihistamine and nasal steroid use. Adjusting the patients by propensity score ensured that patients would be accurately matched between cohort groups as evidenced by the lack of significant differences in observed baseline characteristics after matching. However, caution should be exercised when drawing conclusions, because causal relationships cannot be determined from retrospective studies. There are other limitations of analyses of pharmacy claims databases. Patients in this study were identified by prescriptions filled for allergy medications. It is possible that asthmatic children were being treated with over-the-counter drugs for their allergic symptoms and were not captured in this analysis. Allergic rhinitis is frequently undertreated in children,26 and patients who sought physician counseling on their disease may have had more severe allergies. However, asthmatic children with comorbid allergic rhinitis use more health care services than other children,2,3 so it is likely that the patients included in this analysis are typical of many asthmatic children with allergic rhinitis. Most patients included in this analysis were mild asthmatic patients as judged by -agonist use. Therefore, the selection criteria for this study were probably not biased toward patients with more severe disease, and the results are more likely representative of the larger pediatric asthmatic population. Total pharmacy cost analysis, which would include the costs of montelukast and ICS, was not pursued because of wide variability in amounts both charged and paid for any particular medication. Variables that are directly relevant to any cost analysis include contractual details related to patients, employers and health plans, pharmacy benefit management services, and the manufacturers of the drugs, since discounts and cost sharing would shift the economic impact for any individual stakeholder. No analysis performed to date purports to adjust for all of these variables. In addition, pharmacy costs are only a small proportion of total asthmarelated costs (including asthma-related physician visits and hospitalizations). The primary purpose of this analysis was to use rescue or acute asthma and allergy medication use and costs as proxies for asthma control, which is also influenced
ANNALS OF ALLERGY, ASTHMA & IMMUNOLOGY
by patient preferences, corresponding compliance, and access to health care. The shared pathophysiology of asthma and allergic rhinitis and the impact the presence of one disease has on the other in terms of morbidity and health care costs make the use of one medication that treats both diseases an attractive form of therapy. Montelukast is the only medication that has one formulation (oral tablet) approved for both indications and is proven efficacious in treating asthma and allergic rhinitis.27–33 The current study clearly demonstrated that introduction of montelukast in children with asthma and allergic rhinitis was associated with significantly lower increase in the overall costs of rescue or acute asthma medications and allergy medications compared with introduction of ICSs. These data suggest that montelukast use may be associated with better asthma control than ICS use in this population by reducing the need for asthma rescue medications and also reduced need for acute asthma and allergy therapies. ACKNOWLEDGMENTS We thank Carol A. Tozzi, PhD, for scientific input, writing, and editorial assistance and Arlene Swern, PhD, for statistical review.
12.
13. 14. 15. 16. 17. 18. 19. 20.
REFERENCES 1. Taylor WR, Newacheck PW. Impact of childhood asthma on health. Pediatrics. 1992;90:657– 662. 2. Grupp-Phelan J, Lozano P, Fishman P. Health care utilization and cost in children with asthma and selected comorbidities. J Asthma. 2001;38:363–373. 3. Thomas M, Zhang Q, Sazonov Kocevar V, Yin D, Price DB. Health care resource use by children with asthma and co-morbid allergic rhinitis in general practice in the United Kingdom. Pediatrics. 2005;115:129 –134. 4. Leynaert B, Neukrich F, Demoly P, Bousquet J. Epidemiologic evidence for asthma and rhinitis comorbidity. J Allergy Clin Immunol. 2000;106(5 suppl):S201–S205. 5. Pedersen PA, Weeke ER. Asthma and allergic rhinitis in the same patients. Allergy. 1983;38:25–29. 6. Kapsali T, Horowitz E, Diemer F, Togias A. Rhinitis is ubiquitous in allergic asthmatics. J Allergy Clin Immunol. 1997;99: S138. 7. Yawn BP, Yunginger JW, Wollan PC, et al. Allergic rhinitis in Rochester Minnesota residents with asthma: frequency and impact on health care charges. J Allergy Clin Immunol. 1999;103: 54 –59. 8. Simons FE. Allergic rhinobronchitis: the asthma-allergic rhinitis link. J Allergy Clin Immunol. 1999;104:534 –540. 9. Meltzer EO. Role for cysteinyl leukotriene receptor antagonist therapy in asthma and their potential role in allergic rhinitis based on the concept of “one linked airway disease”. Ann Allergy Asthma Immunol. 2000;84:176 –185. 10. Bousquet J, Van Cauwenberge P, Khaltaev N. Allergic rhinitis and its impact on asthma. J Allergy Clin Immunol. 2001; 108(suppl):S147–S334. 11. Welsh PW, Stricker WE, Chu C-P, et al. Efficacy of beclomethasone nasal solution, flunisolide, and cromolyn in re-
VOLUME 95, AUGUST, 2005
21. 22.
23.
24. 25. 26. 27. 28. 29.
30.
lieving symptoms of ragweed allergy. Mayo Clin Proc. 1987; 62:125–134. Foresi A, Pelucchi A, Gherson G, et al. Once daily intranasal fluticasone propionate (200 micrograms) reduces nasal symptoms and inflammation but also attenuates the increase in bronchial hyperresponsiveness during pollen season in allergic rhinitis. J Allergy Clin Immunol. 1996;98:274 –282. Greiff L, Andersson M, Svensson C, et al. Effects of orally inhaled budesonide in seasonal allergic rhinitis. Eur Respir J. 1998;11:1268 –1273. Vollmer WM, Markson LE, O’Connor E, et al. Association of asthma control with health care utilization. Am J Respir Crit Care Med. 2002;165:195–199. Leidy NK, Paramore LC, Wartrous, et al. Development of algorithm for scoring severity of asthma in administrative databases. Value Health. 1999;2:394. Rubin DB. Estimating causal effects from large data sets using propensity scores. Ann Intern Med. 1997;127(8 pt 2):757–763. Maddala GS. Introduction to Econometrics. New York, NY: Macmillan Publishing Company; 1988. Allen-Ramey FC, Duong PT, Goodman DC, et al. Outcomes of patients treated with ICS propionate or montelukast sodium. Managed Care Int. 2003;16:30 –35. Price DB, Ben-Joseph RH, Zhang Q. Changes in asthma drug therapy costs for patients receiving chronic montelukast therapy in the UK. Respir Med. 2001;95:83– 89. Stempel DA, Mauskopf J, McLaughlin T, et al. Comparison of asthma costs in patients starting fluticasone propionate compared to patients starting montelukast. Respir Med. 2001;95: 227–234. Stempel DA, O’Donnell JC, Meyer JW. Inhaled corticosteroids plus salmeterol or montelukast: effects on resource utilization and costs. J Allergy Clin Immunol. 2002;109:433– 439. Bukstein DA, Henk HJ, Luskin AT. A comparison of asthmarelated expenditures for patients started on montelukast versus fluticasone propionate as monotherapy. Clin Ther. 2001;23: 1589 –1600. Gaga M, Lanbrou P, Papaageorgiou N, et al. Eosinophils are a feature of upper and lower airway pathology in no-atopic asthma, irrespective of the presence of rhinitis. Clin Exp Allergy. 2000;30:663– 669. Schmier JK, Leidy NK. The complexity of treatment adherence in adults with asthma: challenges and opportunities. J Asthma. 1998;35:455– 472. Michels KB, Braunwald E. Estimating treatment effects from observational data: dissonant and resonant notes from the SYMPHONY trials. JAMA. 2002;287:3130 –3132. Stone KD. Atopic diseases of childhood. Curr Opin Pediatr. 2002;14:634 – 646. Reiss TF, Chervinsky P, Dockhorn RJ, et al. Montelukast, a once-daily leukotriene receptor antagonist, in the treatment of chronic asthma. Arch Intern Med. 1998;158:1213–1220. Knorr B, Matz J, Bernstein JA, et al. Montelukast for chronic asthma in 6- to 14-year-old children. JAMA. 1998;279: 1181–1186. Knorr B, Franchi LM, Bisgaard H, et al. Montelukast, a leukotriene receptor antagonist, for the treatment of persistent asthma in children aged 2 to 5 years. Pediatrics. 2001; 108(3):E48. Philip G, Malmstrom K, Hampel FC Jr, et al. Montelukast for
135
treating seasonal allergic rhinitis: a randomized, double-blind, placebo-controlled trial performed in the spring. Clin Exp Allergy. 2002;32:1020 –1028. 31. Nayak AS, Philip G, Lu S, et al. Efficacy and tolerability of montelukast alone or in combination with loratadine in seasonal allergic rhinitis: a multicenter, randomized, double-blind, placebo-controlled trial performed in the fall. Annals Allergy Asthma Immunol. 2002;88:592– 600. 32. van Adelsberg J, Philip G, Pedinoff AJ, et al. Montelukast improves symptoms of seasonal allergic rhinitis over a 4-week treatment period. Allergy. 2003;58:1268 –1276. 33. van Adelsberg J, Philip G, LaForce CF, et al. Randomized
136
controlled trial evaluating the clinical benefit of montelukast for treating spring seasonal allergic rhinitis. Ann Allergy Asthma Immunol. 2003;90:214 –222. Requests for reprints should be addressed to: Allan Luskin, MD Pediatric and Adult Asthma and Allergic Diseases Pediatric Lung Diseases 10 Tower Dr Sun Prairie, WI 53590 E-mail: [email protected]
ANNALS OF ALLERGY, ASTHMA & IMMUNOLOGY