One-year claims analysis comparing inhaled fluticasone propionate with zafirlukast for the treatment of asthma

Transcription

1 One-year claims analysis comparing inhaled fluticasone propionate with zafirlukast for the treatment of asthma David A. Stempel, MD, a Jay W. Meyer, PhD, b Richard H. Stanford, PharmD, MS, c and Steve W.Yancey, MS c Seattle, Wash, Eden Prairie, Minn, and Research Triangle Park, NC Background: Randomized clinical trials have demonstrated that fluticasone propionate (FP) has better objective as well as subjective clinical outcomes than zafirlukast (ZA) in the treatment of asthma. Objective: The goal of this study was to determine whether the superiority of FP over ZA observed in clinical trials is supported under actual practice conditions. Methods: A retrospective cohort analysis of pharmacy and medical claims for asthma was performed. Patients were identified who had at least 1 ICD-9 (493.XX) claim for asthma and were recently prescribed inhaled FP or ZA. Subjects could not have had a claim for any inhaled corticosteroid or oral leukotriene modifier in the 9 months before initiation of FP or ZA. They were subsequently observed for 12 months. Results: A total of 725 persons were new users of FP and 309 of ZA. FP was associated with a 70% reduced risk for hospitalization (P =.0232), a 49% lower risk for an emergency department event (P =.0546), and a 51% reduction in combined emergency department events and hospitalizations (P =.0268) when compared with ZA. Adjusted annual asthma care costs declined significantly for FP and increased for ZA. The adjusted mean difference in annual asthma costs was $215 less per patient for FP (P <.0001). Conclusion: Asthma care costs decreased for patients treated with FP and increased for patients treated with ZA. Furthermore, FP-treated patients had significantly lower risks of asthma-related hospitalization than ZA patients. This study supports results seen in clinical trials comparing these two medications. (J Allergy Clin Immunol 2001;107:94-8.) Key words: Fluticasone propionate, zafirlukast, retrospective analysis, leukotriene modifiers, inhaled corticosteroids, asthma, claims data, asthma costs From a Virginia Mason Medical Center, Seattle; b Ingenix Pharmaceutical Services, Eden Prairie; and c Glaxo Wellcome Inc, Research Triangle Park. Supported by Glaxo Wellcome, Inc. Presented in part at the American College of Allergy, Asthma and Immunology, Chicago, Ill, November Received for publication February 15, 2000; revised September 14, 2000; accepted for publication September 18, Reprint requests: David A. Stempel, MD, Virginia Mason Medical Center, th Ave, C4-ALL, Seattle, WA Copyright 2001 by Mosby, Inc /2001 $ /81/ doi: /mai Abbreviations used ED: Emergency department FP: Fluticasone propionate FP T : All strengths of fluticasone propionate ICS: Inhaled corticosteroid OR: Odds ratio ZA: Zafirlukast Asthma is a chronic inflammatory disease that is increasing in prevalence and morbidity. 1 Two studies have reported that the annual cost of asthma care in the United States accounts for an estimated $5.1 billion to $6.2 billion in direct medical costs, with 35% to 50% of the direct costs of care being emergency department (ED) and hospital visits. 2,3 When adjusted to 1999 dollars this represents an annual cost of about $8 billion to $9.6 billion for total asthma care. Hospitalization and ED visits represent significant expenditures of resources: the cost of 1 asthma-related ED visit and 1 asthma-related hospitalization have been determined to be approximately $235 and $3100, respectively, with an average length of hospital stay of 3.8 days. 4 Several studies have demonstrated that pharmacologic intervention can improve morbidity and decrease hospitalization and ED utilization. 5-9 In response to the recent trends of increasing prevalence and morbidity of asthma, national and international guidelines have been created for the diagnosis and management of asthma. 10,11 Criteria have been established for starting maintenance medications, with the treatment of inflammation identified as the cornerstone of therapy. 12 These guidelines used an evidence-based approach to support the use of inhaled corticosteroids (ICSs) as the initial daily therapy for adults with mild-to-severe persistent asthma. Leukotriene modifiers are listed as an alternative maintenance treatment, although their position as a first-line therapy is unclear under the 1997 guidelines because of the lack of comparative studies demonstrating equivalent efficacy at the time of publication. 10 Since the distribution of these guidelines, several published studies and recent abstracts have directly compared specific ICSs to leukotriene antagonists In all of these studies there has been significantly greater improvement in lung function with the use of ICSs. Most of the other parameters that studied patient-oriented outcomes, including symptom improvement, nighttime awakenings, and use of rescue albuterol, have also significantly favored the ICSs studied Although clinical trials can establish differences in efficacy, the characteristics of the study subjects and the structured environment of the trial may decrease the generalizability of the results to clinical practice. 21,22 Observational study of claims databases may add confirmatory support to randomized clinical trials and enhance the generalizability of the findings to routine practice. 23,24

2 J ALLERGY CLIN IMMUNOL VOLUME 107, NUMBER 1 Stempel et al 95 TABLE I. Patient demographics Parameter ZA FP T FP 44 µg FP 110 µg FP 220 µg Total Age (y) (mean ± SD) 39.4 ± ± 17.8* 27.7 ± 18.5* 34.4 ± 17.1* 40.0 ± 15.3 Female (%) Male (%) *P <.05 compared with ZA. Retrospective cohort studies using administrative claims databases that report medical and pharmacy claims may be used to augment and validate the data from clinical studies Such studies are intended to complement the efficacy data from clinical studies with cost and effectiveness data generated in routine clinical practice. These studies may use inclusion criteria for the patients, create comparison populations, control for comorbidities, and observe groups for longer periods of time. In addition, retrospective cohort studies using administrative claims may use large populations of patients. Furthermore, because they are not controlled trials, they may reflect actual clinical practice conditions. For these reasons, retrospective database analysis is suitable for cost comparisons The objective of this study was to determine whether the differences between ICSs and leukotriene modifiers observed in the clinical trial data are supported under actual practice conditions, and if so, whether these differences translate into significant changes in direct medical and pharmacy utilization. Specifically, the ICS fluticasone propionate (FP) was compared with the leukotriene modifier zafirlukast (ZA) in patients beginning controller therapy for their asthma. ED and hospitalization rates were the primary effectiveness measures, whereas costs of asthma care and total cost of medical care were the primary cost comparisons. METHODS This was a retrospective administrative data analysis using physician, hospital, and pharmacy claims of 13 United Health- Care affiliated plans for the period of October 1, 1995, through December 31, The total database available included 1.4 million lives. Patients were selected only if they had an ICD-9-CM (493.XX; International Classification of Diseases, 9th revision clinical modification) coding for asthma. Study patients were required to be continuously enrolled in their health plan for at least 21 consecutive months around the index event. The index event was a first prescription for FP or ZA. During the 9-month period before the index date (preindex period) FP or ZA patients could not have been prescribed any ICS or leukotriene modifier. Patients were then observed for a subsequent 12-month postindex period. Patients were classified by their index prescription, but during this postindex period, FP patients could receive other ICS medications or a leukotriene modifier, and ZA patients could receive other ICS medications and leukotriene modifiers. FP is available in 3 doses (44, 110, and 220 µg), and the analysis looked at patients receiving any of these strengths of FP (FP T ); as a sensitivity analysis each dose was compare individually to ZA. Patients taking salmeterol were excluded from the analysis to reduce the confounding effect of this additional controller in the postindex period. Asthma-related events were tabulated over the preindex and postindex periods. These were categorized as hospitalization, ED visit, office visit, and outpatient (nonclinician) visit. Total asthma and total health care costs were calculated in addition to asthma pharmacy costs, total pharmacy costs, and annual prescription claims for the index medications. Total health care costs were the sum of all asthma and nonasthma health care costs reported for each patient. Because the preindex and postindex periods were not the same length, mean monthly hospitalization and ED visit rates and costs were compared. Statistical analysis Multivariate logistic and linear regression analyses were performed to determine whether the differences across cohorts with regard to the utilization and cost of asthma-related therapy were maintained after the differences in cohort characteristics were accounted for. This technique permits evaluation of differences in utilization rates, asthma costs, and total health care costs while adjusting for other patient characteristics (covariates). These various models were used to determine the differences in the odds of having a particular asthma event, and in comparing the changes in costs between each of the groups, they compared (1) FP T versus ZA and (2) each different dose of FP versus ZA. Risk-adjusted odds ratios (ORs) for ED or hospitalization events were estimated to assess the change in likelihood of an event occurring in the FP users compared with the ZA users. These ORs were adjusted for age, sex, preindex health care costs, preindex comorbidities (chronic obstructive airway disease), preindex concomitant asthma medications (β-agonists and oral corticosteroids), and preindex ED visits or hospitalization. The logistic regression analysis, which estimates the incremental OR, was based on the number of patients with at least 1 event in the postindex period. An adjusted OR less than 1 indicates a reduction in odds for FP use compared with ZA use. Risk-adjusted changes in asthma and total health care costs were estimated to assess the differences in the change in cost for FP users compared with ZA users. Changes in costs were adjusted for age, sex, preindex health care costs, preindex comorbidities (chronic obstructive airway disease), and preindex concomitant asthma medications (β-agonists and oral corticosteroids). The ordinary leastsquares regression analysis, which estimates the changes in costs, was based on the difference in per treated sample per month costs between the postindex period and the preindex period. A negative risk-adjusted change in cost indicates a reduction in cost for FP use compared with ZA use. RESULTS Demographics and baseline utilization A total of 725 subjects were identified as FP T patients, and 309 were identified as ZA patients. Of the 725 patients treated with fluticasone (FP T ), 234 (32.3%) received FP 44 µg, 368 (50.6%) FP 110 µg, and 123 (17.0%) FP 220 µg. Demographic information of the patients is presented in Table I.

3 96 Stempel et al J ALLERGY CLIN IMMUNOL JANUARY 2001 TABLE II. Adjusted OR* for ED and hospitalization events: FP versus ZA Asthma-related ED visits Asthma-related hospitalization FP dose Adjusted OR P value Adjusted OR P value FP T FP 44 µg NE NE FP 110 µg FP 220 µg NE, Not estimable because of low event frequency. *Adjusted for age, sex, preindex health care costs, preindex comorbidities, preindex concomitant medications (β-agonists and oral corticosteroids), and preindex ED visits and hospitalizations. The percentage of patients with a preindex asthma ED event for FP T was 8.7%, with a range of 6.0% to 10.6% for the different FP doses. The percentage of patients with a preindex asthma ED event for ZA was 6.1%. The percentage of patients with a preindex asthma hospitalization event was 3.3% for FP T (range 1.7%-7.3%) and 2.9% for ZA. Mean baseline ED and hospital rates were higher with increasing doses of FP, as presented in Table II. The total cost of asthma care per patient per month before the index date was approximately $46.06 for FP T (range $30.11-$82.17) and $45.49 for ZA (Table III). The numbers of office visits and nonclinician outpatient visits were similar at baseline among the cohorts. Pharmacy costs for asthma and total medications were higher at baseline for ZA users than for FP users at baseline. Postperiod utilization by treatment group In the 12-month postindex period, patients in the FP T group had a 56% decline in ED visits and 71% decrease in hospitalizations. Patients treated by their providers with ZA had an 11% decrease in ED visits and no change in their hospitalization rate. The adjusted OR for hospitalization was 0.30 (95% CI, ; P =.0232) and for ED visits was 0.51 (95% CI, ; P =.0546). This significant result indicates that the use of FP T lowers the odds of a hospitalization by 70% compared with the use of ZA. If an asthma exacerbation is defined as an asthma-related ED visit or hospitalization, the adjusted OR of either event was 0.49 (95% CI, ; P =.0268). The ORs for ED visits consistently favored FP 44 µg, FP 110 µg, and FP 220 µg (0.7, 0.58, and 0.19, respectively). Only FP 220 µg was statistically significant (P =.0289) because of the overall low event frequency in the other dose groups. The ORs for hospitalization also consistently favored FP for the 3 doses. The adjusted OR of an outpatient visit (total or asthma related) was not statistically significant, indicating no reduction in outpatient visits for FP users versus ZA users. Changes in office visits (total or asthma related) from preindex to postindex also showed no statistically significant difference between FP and ZA users. The effect on total asthma costs statistically favored all 3 doses of FP and their combination. The mean unadjusted costs for preindex and postindex events are presented in Table III. For all 3 doses of FP and their combination, the adjusted incremental total asthma costs were negative, indicating a reduction in cost for FP users compared with ZA users (Table IV). The addition of either FP or ZA may have had an effect on the total cost of health care. The unadjusted mean data presented in Table III show that all doses of FP either reduced or were cost neutral to total health care costs as compared with the increase noted with ZA. More importantly, the adjusted incremental total health care costs, per treated patient per month, were negative, which favored FP but never reached statistical significance between the groups. The mean number of index claims demonstrated that patients treated with ZA had an average of 5.06 claims in the postindex period. Patients treated with FP T had 3.07 claims in this period. These are reported claims and may represent 1 or more months supply for each of the index prescriptions. In addition, approximately 10% of the FP users and 20% of the ZA users had pharmacy claims for other leukotriene modifiers or other ICS medications during the follow-up period. The smaller increase in pharmacy costs observed with FP may be related to the combination of the lower cost of this medication, the lower number of refill claims compared with ZA, or the lower use of additional asthma-control therapy. DISCUSSION This study demonstrated that the use of inhaled FP in an asthma population produced statistically greater reductions in hospitalizations than the use of ZA in a cohort of patients with asthma. Consistent reductions in ED visits were also noted with all doses of FP. These findings were seen with all doses of FP, although they did not reach statistical significance because of the low number of events in the baseline and follow-up periods. In addition, FP at all doses was associated with a significant reduction in total asthma costs. Compared with ZA, the annualized cost savings was $215 per treated patient. Consistently lower total medical costs of care were also noted with all doses of FP, for an annualized cost savings of $530 per treated patient. Patients with asthma requiring controller medication are categorized as having mildly, moderately, or severely persistent disease. Asthma varies in intensity for many patients, and classification is often a moving target. Providers at the time of prescribing controller therapy use their best clinical knowledge and judgment when selecting a primary controller medication. The preindex demographics in this study suggest that physicians may

4 J ALLERGY CLIN IMMUNOL VOLUME 107, NUMBER 1 Stempel et al 97 TABLE III. Unadjusted mean costs per treated patient in the preindex and postindex periods Asthma pharmacy costs Asthma costs Total health care costs ($/member/mo) ($/member/mo)* ($/member/mo) Medication Pre Post Pre Post Pre Post ZA FP T FP 44 µg FP 110 µg FP 220 µg *Includes asthma-related pharmacy and medical costs. Includes asthma-related and non-asthma-related pharmacy and medical costs. TABLE IV. Adjusted incremental asthma and total health care costs Comparison Adjusted incremental monthly asthma costs ($) Adjusted incremental monthly health care costs ($) ZA vs FP T (P <.0001) (P <.33)* ZA vs FP 44 µg (P <.0004) (P <.17)* ZA vs FP 110 µg (P <.0015) (P <.55)* ZA vs FP 220 µg (P <.0219) (P <.5)* *Nonsignificant comparisons. Note: a negative incremental cost favors Flovent use. have selected ZA for patients most similar to those treated with FP 110 µg for hospitalizations, total asthma costs, and total health care costs. The risk of an asthmarelated hospitalization was 70% less (P =.0232) for patients treated with FP T than for those treated with ZA. The OR for ED visits was 0.51 for FP T but did not achieve statistical significance (P =.0546). When both ED visits and hospitalizations are combined, patients treated with FP had statistically lower rates during the follow-up period even though their preindex rates were most often higher. Two desirable attributes of a therapeutic intervention are improvement in clinical outcomes and cost-effectiveness of the medication. This study has demonstrated reduced resource utilization in the inpatient treatment of asthma. For all doses of FP, mean adjusted asthma costs were significantly lower than for ZA. Although statistical significance was not achieved for differences in total health care costs, the incremental costs were negative, indicating costs were always lower for FP than for ZA. This study has demonstrated greater clinical benefit at significantly lower cost. The results of this study, which used administrative claims data, are consistent and complementary with those of previous studies using clinical trial data to compare ICSs versus leukotriene inhibitors. Clinical studies have consistently demonstrated that ICS treatment is associated with better clinical outcomes and improved lung function than leukotriene modifier therapy Bleecker et al 19 have reported that FP 88 µg taken twice daily when compared with ZA 20 mg twice daily significantly improved FEV 1 and both morning and evening peak expiratory flow rates. In addition, symptom scores, symptom-free days, albuterol use, rescue-free days, and nighttime awakenings all improved significantly more with FP than with ZA. The finding of superiority for ICSs is consistent with findings of most other direct head-to-head comparisons of ICSs and leukotriene modifiers Although it was beyond the scope of a retrospective claims analysis to measure symptom improvement, there was a significantly greater decrease in combined utilization of ED and inpatient hospital services for patients treated with FP than for those treated with ZA, which suggests that exacerbations were better controlled in the FP groups. Furthermore, this study presents results consistent with previous estimates of the effect of ICS use on hospitalization rates for asthma. Donahue et al 5 assessed hospital rates in patients treated with ICSs in the Harvard Pilgrim health care program. They reported that patients receiving ICSs as maintenance therapy had reductions in hospitalization rates of up to 50% compared with patients receiving short-acting β-agonists alone, whereas patients treated with cromolyn sodium had a reduced relative risk of hospitalization of approximately 20%. These results are consistent with the data reported in this study showing a reduction in the risk of hospitalization of 70% in FP users and no change in ZA users. Retrospective cohort studies using claims data, such as this one, have significant limitations. Selection bias by prescribing physicians and managed care and assessment of severity of disease often pose difficulties in analysis of data from administrative claims. Appropriate statistical methods were used in this study to help in interpretation of the data. Adjustment for severity based on the number of claims is a technique widely used in administrative claims-based research and is generally accepted as a reliable measure of disease severity. 22,27-30 However, other factors not accounted for may affect outcomes, such as misclassification of diagnosis. This misclassification is probably nondifferential, and thus differences between

5 98 Stempel et al J ALLERGY CLIN IMMUNOL JANUARY 2001 the groups should be negated. In addition, the results from observational studies such as this one show associations between drug and outcomes, although causality may be difficult to determine. Despite these limitations, these data provide important information on the utilization of medical resources including ED visits, hospitalizations, pharmacy expenses, and physician visits occurring in clinical practice. Controlled clinical trials have many strengths. They are randomized, the population is well defined, there are careful controls regarding study design, and they have specific measurable outcomes such as lung-function measurement. However, controlled clinical trials have their own limitations. The populations studied are exclusionary; patients with very mild or very severe disease are frequently not included. Study lengths are usually 12 weeks or less, and hospital and ED events are uncommon. These limitations of randomized controlled clinical trials are the strengths of retrospective claims analysis that are more inclusionary, generalizable, longer in duration, and have the ability to capture rare ED and hospital events. The results of this study demonstrated that inhaled FP produced significantly better outcomes than ZA in the treatment of patients with a diagnosis of asthma who were beginning maintenance therapy, as measured by the reduction in combined hospitalizations and ED visits. This study corroborates the clinical trial findings of ICSs versus leukotriene modifiers and expands the body of literature to include actual practice conditions. In addition to better outcomes, FP was also associated with significantly lower asthma-related costs compared with ZA. These results, along with the findings of previous studies, complement the position of the NIH Guidelines for the Diagnosis and Management of Asthma 10 that ICSs should be selected as first-line maintenance treatment for asthma. REFERENCES 1. Centers for Disease Control and Prevention. CDC Surveillance Summaries. April 24, 1998; 47 (No. SS-1). 2. Smith DH, Malone KA, Lawson LJ, Okatmoto LJ, Battista C, Saunders W. A national estimate of the economic costs of asthma. Am J Respir Crit Care Med 1997;156: Weiss KB, Gergen PJ, Hodgson TA. An economic evaluation of asthma in the United States. N Engl J Med 1992;326: Stanford R, McLaughlin T, Okamoto LJ. The cost of asthma in the emergency department and hospital. Am J Respir Crit Care Med 1999;160: Donahue JG, Weiss ST, Livingston JM, Goetsch MA, Greineder DK, Platt R. Inhaled steroids and the risk of hospitalization for asthma. JAMA 1997;277: Balkrishnan R, Norwood GJ, Anderson A. Effect of corticosteroid therapy introduction in asthmatic Medicaid-enrolled children. Drug Benefit Trends 1998;10: Ishihara K, Hasegawa T, Nishimura T, Okazaki M, Katakami N, Umeda B. Increased use of inhaled corticosteroids and reduced hospitalizations in adult asthmatics: 11 years experience in a Japanese hospital. Respirology 1998;3: Blais L, Suissa S, Boivin JF, Ernst P. First treatment with inhaled corticosteroids and the prevention of admissions to hospital for asthma. Thorax 1998;53: Goldman M, Rachmiel M, Gendler L, Katz Y. Decrease in asthma mortality rate in Israel from : is it related to the use of inhaled corticosteroids? J Allergy Clin Immunol 2000;105: National Institutes of Health (National Heart, Lung, and Blood Institute). Highlights of the expert panel report II: guidelines for the diagnosis and management of asthma. Bethesda (MD): NIH; p (NIH Publication No A). 11. Higgins G. New US asthma guidelines stress early, aggressive treatment. Inpharma 1997;1077: Keeley D, Rees J. New guidelines on asthma management. BMJ 1997;314: Busse WW. Inflammation in asthma: the cornerstone of the disease and target of therapy. J Allergy Clin Immunol 1998;102:S Westbroek J, Pasma HR. The effect of inhaled fluticasone propionate 100 µ bid compared with oral zafirlukast 20 mg bid on bronchial hyperresponsiveness [abstract]. Eur Respir J 1997;10(Suppl 25):243S. 15. Westbroek J, Pasma HR, James MH, Thwaites RMA. Inhaled fluticasone propionate and oral zafirlukast in moderate asthma: a clinical and cost comparison [abstract]. Am J Respir Crit Care Med 1998;157:A Mullen B, Driskel JE, Yancey SW. A statistical comparison of the efficacy of leukotriene modifiers and inhaled corticosteroids [abstract]. J Allergy Clin Immunol 1998;101:S Busse WW, Srebro SH, Edwards L, Johnson MC, Rickard K. Low-dose inhaled fluticasone propionate versus oral zafirlukast in asthma patients [abstract]. Am J Respir Crit Care Med 1999;159:A Storms WW, Srebro SH, Edwards L, Rickard K. Better asthma symptom control with low-dose inhaled fluticasone propionate than with oral zafirlukast. Am J Respir Crit Care Med 1999;159: Bleecker ER, Welch MJ, Weinstein SF, Kalberg C, Johnson M, Edwards L, et al. Low-dose inhaled fluticasone propionate versus oral zafirlukast in the treatment of persistent asthma. J Allergy Clin Immunol 2000;105: Malmstrom K, Rodriguez-Gomez G, Guerra J, Villaran C, Pineiro A, Wei LX, et al. Oral montelukast, inhaled beclomethasone, and placebo for chronic asthma. A randomized, controlled trial. Montelukast/Beclomethasone Study Group. Ann Intern Med 1999;130: Jones AJ, Sanchez LA. Pharmacoeconomic evaluation: applications in managed care health care formulary decision-making. Drug Benefit Trends 1995;7:12,15,19-22, Goldberg-Arnold R, Kotsanos JG, Motheral B, Ramsey S, Crown W, Poaler K, et al. Methodological issues in conducting pharmacoeconomic evaluations: retrospective and claims database studies. Value in Health 1999;2: Concato J, Shah N, Horwitz RI. Randomized, controlled trials, observational studies, and the hierarchy of research designs. N Engl J Med 2000;342: Pocock SJ, Elbourne DR. Randomized trials or observational tribulations. N Engl J Med 2000;342: Association for Pharmacoeconomics and Outcomes Research (APOR) Consensus Development Committee. Draft consensus development on pharmacoeconomic analysis: a working paper. Drug Benefit Trends 1997;9: Blaiss MS. Outcomes analysis in asthma. JAMA 1997;278: Stempel DA, Hedblom EC, Durcanin-Robbins JF, Sturm LL. Use of a pharmacy and medical claims database to document cost centers for 1993 annual asthma expenditures. Arch Fam Med 1996;5: Guyatt G, Drummond M, Feeny D, Tugwell P, Stoddart G, Haynes RB, et al. Guidelines for the clinical and economic evaluation of health care technologies. Soc Sci Med 1986;22: Grana J, Preston S, McDermott PD, Hanchak NA. The use of administrative data to risk-stratify asthmatic patients. Am J Med Qual 1997;12: Minshall ME, Kody MC, Mosbacher F. Pharmacoeconomic research credibility: a controversial and recurring theme in health outcomes research. Med Care 1999;37:AS12-9.