Economic Evaluation of Quality-of-Life Improvement with Second-Generation Antihistamines and Montelukast in Patients with Allergic Rhinitis

Allergic rhinitis (AR) has a significant detrimental effect on a patient's quality of life (QoL), affecting sleep, normal daily activities, and work.^1-4 Stempel and Woolf estimated that the 50 million Americans with AR spend more than $6 billion annually on prescription and nonprescription medications to relieve AR symptoms.^5,6 Despite a growing body of literature on the effect of AR on QoL, there is a paucity of studies that use QoL findings to assess the cost-effectiveness of various agents used to treat AR. The recent US Food and Drug Administration (FDA) approval of the second-generation antihistamine (SGA) levocetirizine, and the lack of studies that compare the cost-effectiveness of oral AR treatments in terms of QoL led to the present analysis.

Although estimates of economic impact vary widely, AR clearly has important economic implications for managed care, employers, and patients. Specifically, estimates of the direct costs to US payers range from $1.4 billion to $6 billion annually.^5,7,8 These estimates may be low, however, because they exclude substantial spending on over-the-counter (OTC) allergy medications, a cost that is largely borne by the patient.⁵ In addition, indirect costs realized by employers through presenteeism and absenteeism represent a significant burden. When these indirect costs are considered in conjunction with direct healthcare costs, allergies are ranked as the fifth most expensive chronic condition for employers.⁹

Lack of rigorous economic evidence has not slowed the adoption of newer SGAs, which are among the States. SGAs have the advantage of fewer side effects, including less sedation, than older antihistamines,¹⁰ making them more acceptable to many patients. The purpose of this study was to estimate the comparative cost-effectiveness of QoL improvements associated with major prescription agents used for AR management.

To date, no cost-effectiveness studies have compared individual SGAs to one another or to alternative oral AR treatments based on QoL improvements. One costeffectiveness study compared SGAs with older, firstgeneration antihistamines that produced a significant sedating effect.¹⁰ However, current treatment patterns call for more advanced modeling that directly compares economic outcomes of treatment patterns with newer agents.

A recent study by some of the present authors compared the cost-effectiveness of SGAs to one another and to alternative AR treatments based on clinical symptom improvement.¹¹ Our current analysis builds on that study by assessing the cost-effectiveness of levocetirizine relative to other prescription SGAs and the widely used leukotriene receptor antagonist montelukast in terms of QoL improvements reported by clinical trials. The present study was conducted from the perspective of prescription benefit managers from US health plans; therefore, OTC products were not included.

Methods
We used a decision-analytic cost-effectiveness model developed from the perspective of managed care decision makers and using costs over a 1-year time period. US payers typically do not include OTC products in their benefit design; therefore, our treatment comparators were limited to prescription products, including the SGAs levocetirizine, desloratadine, and fexofenadine, and the leukotriene receptor antagonist montelukast, which is also FDA approved for AR treatment.

The study population included patients with AR who had been treated with an SGA monotherapy or with montelukast. Although combination therapy with an SGA and montelukast is sometimes used for AR symptoms, combination therapy was excluded in this analysis, because clinical trial evidence supporting such combination therapy is limited, which would limit the generalizability of our findings. Patients with asthma requiring daily corticosteroid treatment were excluded, to preserve homogeneity of the population.

The Rhinoconjunctivitis Quality of Life Questionnaire (RQLQ) was selected for our model, because it is specific to AR, widely used, validated, and has a specific clinical interpretation.^1,2,12-14 The RQLQ includes 28 items on 7 disease-specific domains: activity limitations, practical problems, nonrespiratory symptoms, nasal symptoms, eye symptoms, emotional function, and sleep. The outcome measure in our model was the composite of all 28 items.

We conducted a MEDLINE search to identify eligible articles between January 1950 and May 2007, using the comparator names levocetirizine, fexofenadine, desloratadine, and montelukast in combination with the terms RQLQ, QoL, and Rhinoconjunctivitis Quality of Life Questionnaire. Additional studies were identified by searching the references listed in these studies. Eligible studies had to (1) include monotherapy with an FDAapproved dose of one of the model comparator agents; (2) be randomized, blinded, and placebo-controlled; (3) exclude patients with asthma requiring daily corticosteroids; and (4) include RQLQ as an outcome. Study length had to be 14 to 90 days. Patients' age had to be 11 years or older. A total of 12 studies were included in the analysis.^15-26

For each study, the standardized mean difference (SMD) between the comparator and placebo was calculated as a ratio of the RQLQ outcome to the standard deviation (see Goodman and colleagues¹¹). To convert the pooled SMD into a usable measure for the costeffective ratio, the baseline mean and standard deviation of the RQLQ for placebo and the comparator were combined. Using the baseline mean and the standard deviation as the untreated standard, the mean marginal RQLQ score, assuming treatment with each comparator after removing the effect of placebo observed in the trials, was calculated as the baseline mean plus the SMD multiplied by the standard deviation.

A clinically relevant improvement was defined as a 0.5-point reduction from baseline in marginal RQLQ score—after removing the effect of placebo—which was used as the threshold for clinically relevant improvement. In contrast, Juniper and colleagues' definition does not remove the effect of placebo.¹ Our definition, although conservative, is consistent with the standards for cost-effectiveness analysis. The proportion of the population 0.5 points below the baseline mean was calculated using standard formulas for computing the area under the normal curve. The probability of clinically relevant improvement was defined as the marginal difference in the proportion of the population below the threshold for clinically relevant improvement.¹¹

Drug costs were calculated as the expected days of therapy in a year multiplied by the daily wholesale acquisition cost.²⁷ The model assumed 90 days of therapy for a calendar year. Medical costs for allergy-related physician office visits were calculated from an analysis of the proprietary PharMetrics dataset for a 1-year period for each model comparator agent. Medical costs were inflated to 2007 dollars, using the Bureau of Labor Statistics.²⁸

Because levocetirizine was not available in the United States when the PharMetrics data were captured, its costs were imputed using a linear fit of the RQLQ effect size to the physician's office visit costs for the other model comparator agents, based on a simple linear regression. Indirect costs, such as productivity, were not included, because they were not assessed in the original trials used for our analysis, and because the model's perspective was that of a third-party payer.

The comparative cost-effectiveness of the agents was calculated as the ratio of costs to the probability of clinically relevant improvement. Incremental costeffective ratio (ICER) between agents was evaluated as the ratio of difference in cost to difference in probability of clinically relevant improvement for any alternative therapy relative to levocetirizine.

A Monte Carlo simulation, which varied the total cost by ±10%, was used to obtain 95% confidence interval (CI) values. The effectiveness measure (SMD) for each product varied within the 95% CI range for the pooled SMD across each comparator. Random draws were run 1000 times (using Microsoft Excel 2003, Service Pack 2). Exact 95% CI values for the cost-effective ratios were calculated as the 26th and 974th ordered values in the simulation.

Significance for a cost-effective ratio was defined by a CI that did not overlap the point estimate; significance for an ICER was defined as a CI that did not overlap 0.0.

Results
Effects on Quality of Life

Table 1

Table 1 compares the mean QoL effect size for each comparator agent versus placebo. All comparators were significantly better than placebo in improving QoL.15-26 Levocetirizine demonstrated greater QoL improvement (–0.418; 95% CI, –0.573, –0.262), as measured by the pooled RQLQ SMD, than desloratadine (–0.360; 95% CI, –0.539, –0.180), montelukast (–0.213; 95% CI, –0.267, –0.159), or fexofenadine (–0.201; 95% CI, –0.301, –0.101).
 

Table 2

Table 2 summarizes the annual drug and medical expenditures for each comparator, as well as each agent's efficacy, expressed as the probability of a clinically relevant improvement in RQLQ. The medical costs for AR physician office visits were highest for montelukast. Annual AR drug costs, assuming 90 days of therapy, ranged from $168 to $275. Column 5 of Table 2 translates the SMDs into a probability of clinically relevant improvement. Applying these probability estimates to a population of 10,000 AR patients, levocetirizine would lead to clinically relevant improvement in an additional 810 patients compared with montelukast, or 231 additional patients compared with desloratadine.

Cost-Effectiveness

Table 3

Table 3 outlines the results of this study. Levocetirizine had the lowest average cost ($3255) for a clinically relevant RQLQ improvement, followed by desloratadine ($4165). Montelukast had the highest cost ($7871) per clinically relevant RQLQ improvement and lower efficacy compared with the other comparators.

Negative ICERs (Table 3) can reflect either lower cost and higher effectiveness or higher cost and lower effectiveness, which are generally reported as "dominated." All the comparators in this study are less effective in terms of RQLQ improvement than levocetirizine. Generic fexofenadine is less costly than levocetirizine but results in lower RQLQ improvement. The other comparators have lower RQLQ improvement and are more costly than levocetirizine.

The statistical significance of the ICERs is shown with CI values. The wide CI values reflect the relatively small sample sizes and the resulting large variation in the RQLQ measures in the Monte Carlo simulation. Only the comparison between levocetirizine and montelukast is significant (95% CI, –7471, –212). Because the CIs overlapped zero for the other comparators, another way to understand ICERs is to examine how many times the simulated ICER is negative, which indicates that levocetirizine dominated the comparator in that simulation. The number of negative ICERs of 1000 simulations is 601 for desloratadine, 273 for generic fexofenadine, 644 for branded fexofenadine, and 993 for montelukast. In the case of desloratadine, negative ICERs indicate that 60% of the time levocetirizine has greater RQLQ improvement and is less costly.

When the ICERs are positive, a tradeoff is required between cost and effectiveness. The tradeoff for a positive ICER is between lower cost and lower effectiveness greater clinical benefit for the other. For example, the positive ICER for generic fexofenadine (Table 3) reflects its lower cost and lower impact on RQLQ compared with the greater clinical benefit of levocetirizine. Decision makers will need to weigh the additional clinical benefit against the additional cost.

Discussion
The 4 model comparators vary substantially in their ability to improve QoL for patients with uncomplicated AR. This variation, combined with significant variation in the costs of the comparators, make this costeffectiveness analysis important for formulary decision makers, clinicians treating AR, and patients undergoing treatment. The results favor levocetirizine, which has the lowest average cost-effective ratio.

To our knowledge, this is the first published analysis of the cost per patient with clinically relevant improvement in RQLQ of the individual SGAs and the only leukotriene receptor antagonist indicated for the treatment of AR. This is surprising given the prevalence and the large economic impact of AR in direct and indirect costs. Sullivan and colleagues noted that the lack of a standard outcome measure across AR studies could contribute to the scant number of studies about cost-effectiveness of AR therapies.²⁹

The effectiveness measure in the present study is important clinically and is consistent with the need for economic decisions that focus on QoL.³⁰ The choice to use the RQLQ as the basis of our cost-effectiveness model was based on several factors. The RQLQ is reproducible, can assess the impact of treatment over multiple dimensions, and has a strict clinical interpretation that has been validated in many studies.^1,12-14 Our results, however, are conservative. To make the denominator of the cost-effectiveness ratio consistent with standards in economic modeling, we removed the effect observed for placebo from the calculation of clinically relevant improvement. This reduces the number of persons in a population whom we classify as having clinically relevant improvement. A change that is smaller than the clinically relevant threshold may nevertheless represent meaningful improvement in symptom relief for many patients. An alternative measure, the number needed to treat to achieve 1 person with improvement, has been suggested for interpreting the RQLQ.³¹

Our analysis was designed specifically to assess the cost-effectiveness of levocetirizine relative to other oral prescription medications for the management of AR symptoms, where effectiveness is defined as clinically relevant improvement in RQLQ. Our model indicates that levocetirizine has greater RQLQ improvement and is less costly than montelukast for the management of AR in patients without asthma who require daily corticosteroids. However, because the 95% CI of the ICERs comparing levocetirizine with the other comparators cross zero, levocetirizine does not have complete dominance over the SGA comparators.

The 95% CI of the ICER comparing levocetirizine with desloratadine was wide (–$10,138 and $17,275, respectively; Table 3) for 2 reasons. First, only 2 trials had usable data for changes in RQLQ for levocetirizine, and only 1 study for desloratadine. With such few studies, the resulting variability from the SMD in RQLQ score was large. Second, the difference in probability of a clinically relevant improvement between the 2 drugs was small (16.1% and 13.8%, respectively). When the denominator in an ICER is a probability, a small difference between the 2 comparators leads to very large CI estimates. As additional comparator-specific information on QoL becomes available, the precision of these estimates will be increased and judgments about their relative effectiveness and cost-effectiveness will become more accurate.

Overall, the ICER CI values were very wide, which again reflects the lack of multiple studies for the comparators. For example, the variation for montelukast is smaller than for desloratadine, because there are more montelukast trials with RQLQ data. Although the results are valid for the specific sample sizes, the relatively small number of studies results in CI values that might have obscured effects of clinical significance for improved QoL.³²

Several prescription and OTC medications are approved for the relief of AR symptoms, including nasal corticosteroids, antihistamines, decongestants, and leukotriene receptor antagonists. Although the use of these agents and combination therapy is common for AR management, it was unreasonable to include all these options in our analysis, because of the lack of RQLQ outcomes data and because of the sheer number of potential comparators. We therefore opted to include comparators available only by prescription, because our goal was to address issues relevant to prescription drug benefit managers. Many health plans do not reimburse for OTC medications.³³

Our model included direct costs only, specifically drug costs and outpatient physician office visit costs for AR. The literature indicates that a sizable amount of the economic burden associated with AR is because of lost productivity from absenteeism and presenteeism.³⁴ An employee survey showed that employees with AR were absent roughly 3.6 days annually due to this condition and lost 2.3 hours per workday in productivity when symptoms were present. In our analysis, indirect costs were excluded due to lack of data substantiating differences in indirect costs between individual model comparators, which may be related to the short duration of most AR clinical trials. Other costs not accounted for in our model include costs associated with lack of effectiveness and adverse events.

Limitations
Several potential limitations of this study warrant discussion. Our rigorous inclusion criteria might have led to unintended bias. For example, Meltzer and colleagues only reported on the significance of a single RQLQ domain, so this study was not included in our analysis.³⁵

In addition, the AR patient population is broad, ranging from young children to the elderly and from relatively healthy patients to patients with severe respiratory conditions. Our decision to exclude clinical trials that enrolled children younger than 11 years old or AR patients with concomitant asthma treated with daily corticosteroids does not accurately reflect the overall population with AR. It is estimated that nearly 40% of individuals with allergies have asthma.³⁶ Furthermore, 60% to 78% of individuals with asthma have AR.³⁶ Montelukast is FDA-approved for the management of asthma and AR, whereas the SGAs are not approved for the treatment of asthma.^37-40 Therefore, we chose to eliminate studies involving patients with asthma requiring daily corticosteroids so as to not bias the results with respect to montelukast. About 40% of individuals younger than age 12 suffer from AR.6 Therefore, although our results are robust to the general population, specific subpopulations are not represented, and the cost-effectiveness in the young and in those with significant comorbidities remains to be examined in future research.

We also did not include OTC SGAs, such as loratadine and cetirizine, in our analysis. We believe that this is justified, because the audience for this study is health plan decision makers. Historically, when drugs have been moved to an OTC status, this typically has removed them from coverage by managed care.⁴¹ Therefore, our intended audience is served by this exclusion. In addition, cost and QoL data are not available for all OTC agents.

We only included physician office visits and AR drug costs, and excluded costs for emergency department visits or hospitalizations. Patients with AR without asthma are unlikely to be hospitalized, visit the emergency department, or use additional drugs.

There may be some limitations to using the PharMetrics database to obtain the cost inputs for the analysis. For example, diagnosis codes from claims are based on payment rather than on clinical practice. Therefore, diagnoses from claims data may be less specific than diagnoses or narrative in a medical record. This limitation, however, is offset by the coding bias being consistent across all comparators. In addition, the PharMetrics data contain few elderly patients; however, as with the coding bias noted, this should result in a consistent effect across all comparators.

We also did not separate seasonal and perennial AR. All the agents included in our analysis have an indication for both seasonal and perennial AR, except for fexofenadine, which is only indicated for seasonal AR. In addition, the availability of the clinical trial data would have been limited if we separated our analysis by seasonal and perennial AR, potentially reducing the robustness of the QoL estimates.

Finally, we did not formally analyze statistical heterogeneity in our calculation of comparative effectiveness. Because our goal was to model the economics rather than conduct a full-fledged meta-analysis of comparative effectiveness of AR medications, it was beyond this study's scope to rigorously exclude studies that might introduce heterogeneity. Estimates derived from a full meta-analysis would provide tighter CI values for effectiveness measures, but they would also limit the generalizability of the results.

Conclusion
Our analysis shows that levocetirizine is a cost-effective option for the treatment of AR that produces clinically meaningful improvement in QoL based on the RQLQ. Levocetirizine is cost-effective compared with montelukast, and its cost-effective ratios are favorable compared with the other comparators in this analysis. Further research is warranted to assess patients with AR and comorbid asthma who require daily cortico steroids and in younger populations, as well as to address the effect of different therapies on indirect costs of AR.

Acknowledgments
Laurie Kozbelt and Jeanne Hawkins of Xcenda, LLC, Palm Harbor, FL, provided technical assistance in conducting this study and in preparing this manuscript.

Disclosure Statement
This study was funded by UCB, Inc, and sanofi-aventis US, LLC. Ms Saverno and Dr Goodman were employed and Dr Meyer is currently employed by Xcenda, which provides consulting services to various pharmaceutical companies, including UCB and sanofiaventis. Dr Seal is employed by sanofi-aventis.

References

1. Juniper EF, Guyatt GH, Griffith LE, Ferrie PJ. Interpretation of rhinoconjunctivitis quality of life questionnaire data. J Allergy Clin Immunol. 1996;98:843-845.
2. Juniper EF, O'Byrne PM, Guyatt GH, et al. Development and validation of a questionnaire to measure asthma control. Eur Respir J. 1999;14:902-907.
3. Nathan RA. The burden of allergic rhinitis. Allergy Asthma Proc. 2007;28:3-9.
4. Schoenwetter WF, Dupclay L Jr, Appajosyula S, et al. Economic impact and quality-of-life burden of allergic rhinitis. Curr Med Res Opin. 2004;20:305-317.
5. Stempel DA, Woolf R. The cost of treating allergic rhinitis. Curr Allergy Asthma Rep. 2002;2:223-230.
6. American Academy of Allergy, Asthma, and Immunology. Allergy statistics. www.aaaai.org/media/resources/media_kit/allergy_statistics.stm. Accessed January 22, 2008.
7. Malone DC, Lawson KA, Smith DH, et al. A cost of illness study of allergic rhinitis in the United States. J Allergy Clin Immunol. 1997;99(1 pt 1):22-27.
8. Reed SD, Lee TA, McCrory DC. The economic burden of allergic rhinitis: a critical evaluation of the literature. Pharmacoeconomics. 2004;22:345-361.
9. Goetzel RZ, Long SR, Ozminkowski RJ, et al. Health, absence, disability, and presenteeism cost estimates of certain physical and mental health conditions affecting U.S. employers. J Occup Environ Med. 2004;46:398-412.
10. Sullivan PW, Nichol MB. The economic impact of payer policies after the Rxto-OTC switch of second-generation antihistamines. Value Health. 2004;7:402-412.
11. Goodman MJ, Jhaveri M, Saverno K, et al. Cost-effectiveness of second-generation antihistamines and montelukast in relieving nasal symptoms of allergic rhinitis. Am Health Drug Benefits. 2008;1(8):26-34.
12. Juniper EF, Guyatt GH. Development and testing of a new measure of health status for clinical trials in rhinoconjunctivitis. Clin Exp Allergy. 1991;21:77-83.
13. Juniper EF, Thompson AK, Roberts JN. Can the standard gamble and rating scale be used to measure quality of life in rhinoconjunctivitis? Comparison with the RQLQ and SF-36. Allergy. 2002;57:201-206.
14. Juniper EF, Rohrbaugh T, Meltzer EO. A questionnaire to measure quality of life in adults with nocturnal allergic rhinoconjunctivitis. J Allergy Clin Immunol. 2003; 111:484-490.
15. Bachert C, Bousquet J, Canonica GW, et al. Levocetirizine improves quality of life and reduces costs in long-term management of persistent allergic rhinitis. J Allergy Clin Immunol. 2004;114:838-844.
16. Lombardo G, Quattrocchi P, Lombardo GR, et al. Concomitant levocetirizine and montelukast in the treatment of seasonal allergic rhinitis: influence on clinical symptoms. IT J Allergy Clin Immunol. 2006;16:63-68.
17. Meltzer EO, Casale TB, Nathan RA, Thompson AK. Once-daily fexofenadine HCl improves quality of life and reduces work and activity impairment in patients with seasonal allergic rhinitis. Ann Allergy Asthma Immunol. 1999;83:311-317.
18. Meltzer EO, Malmstrom K, Lu S, et al. Concomitant montelukast and loratadine as treatment for seasonal allergic rhinitis: a randomized, placebo-controlled clinical trial. J Allergy Clin Immunol. 2000;105:917-922.
19. 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. Ann Allergy Asthma Immunol. 2002;88:592-600.
20. Okubo K, Gotoh M, Shimada K, et al. Fexofenadine improves the quality of life and work productivity in Japanese patients with seasonal allergic rhinitis during the peak cedar pollinosis season. Int Arch Allergy Immunol. 2005;136:148-154.
21. Patel P, Philip G, Yang W, et al. Randomized, double-blind, placebo-controlled study of montelukast for treating perennial allergic rhinitis. Ann Allergy Asthma Immunol. 2005;95:551-557.
22. Philip G, Malmstrom K, Hampel FC, et al. Montelukast for treating seasonal allergic rhinitis: a randomized, double-blind, placebo-controlled trial performed in the spring. Clin Exp Allergy. 2002;32:1020-1028.
23. Pradalier A, Neukirch C, Dreyfus I, Devillier P. Desloratadine improves quality of life and symptom severity in patients with allergic rhinitis. Allergy. 2007;62:1331-1334.
24. Tanner LA, Reilly M, Meltzer EO, et al. Effect of fexofenadine HCl on quality of life and work, classroom, and daily activity impairment in patients with seasonal allergic rhinitis. Am J Manag Care. 1999;5(suppl):S235-S247.
25. 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.
26. van Adelsberg J, Philip G, LaForce CF, et al. Randomized controlled trial evaluating the clinical benefit of montelukast for treating spring seasonal allergic rhinitis. Ann Allergy Asthma Immunol. 2003;90:214-222.
27. RED BOOK for Windows. Volume 46; 2007. [Requires password to access online.]
28. US Department of Labor, Bureau of Labor Statistics. Consumer price index. US medical care, 1982-84=100-CUUR0000SAM. http://data.bls.gov/cgi-bin/surveymost?cu. Accessed January 22, 2008.
29. Sullivan PW, Follin SL, Nichol MB. Transitioning the second-generation antihistamines to over-the-counter status. A cost-effectiveness analysis. Med Care. 2003; 41:1382-1395.
30. Gold MR. Cost-effectiveness in Health and Medicine. New York, NY: Oxford University; 1996.
31. Guyatt GH, Juniper EF, Walter SD, et al. Interpreting treatment effects in randomised trials. BMJ. 1998;316:690-693.
32. Kirk RE. Practical significance: a concept whose time has come. Educ Psychol Meas. 1996;56:746-759.
33. Hay JW, Leahy M. Cost and utilization impacts of oral antihistamines in the California Medi-Cal program. Value Health. 2005;8:506-516.
34. Lamb CE, Ratner PH, Johnson CE, et al. Economic impact of workplace productivity losses due to allergic rhinitis compared with select medical conditions in the United States from an employer perspective. Curr Med Res Opin. 2006;22: 1203-1210.
35. Meltzer EO, Jalowayski AA, Vogt K, et al. Effect of desloratadine therapy on symptom scores and measures of nasal patency in seasonal allergic rhinitis: results of a single-center, placebo-controlled trial. Ann Allergy Asthma Immunol. 2006;96: 363-368.
36. Crown WH, Olufade A, Smith MW, Nathan R. Seasonal versus perennial allergic rhinitis: drug and medical resource use patterns. Value Health. 2003;6:448-456.
37. Clarinex [package insert]. Kenilworth, NJ: Schering Co; 2005.
38. Xyzal [package insert]. Smyrna, GA: UCB; 2007.
39. Singulair [package insert]. Whitehouse Station, NJ: Merck & Co; 2007.
40. Allegra [package insert]. Bridgewater, NJ: sanofi-aventis US; 2008.
41. Sullivan PW, Nair KV, Patel BV. The effect of the Rx-to-OTC switch of loratadine and changes in prescription drug benefits on utilization and cost of therapy. Am J Manag Care. 2005;11:374-382.