Address correspondence to: William H. Crown, Vice President, Outcomes Research and Econometrics, The MEDSTAT Group, 125 Cambridge Park Drive, Cambridge, MA 02140. E-mail: email@example.com
Background: There are no published studies that have compared the medical costs of patients with seasonal and perennial allergic rhinitis symptomatology.
Objectives: The objectives of this study were to develop an algorithm for classifying patients into seasonal and perennial groups based on their patterns of allergy medication use and then compare the epidemiology and economics of the two groups.
Methods: Data for the study were obtained from the 1996 to 1998 MarketScan databases containing linked inpatient, outpatient, and pharmaceutical claims for approximately 2 million covered lives annually. Patients were classified into seasonal allergic rhinitis (SAR) or perennial allergic rhinitis (PAR) groups based on their pattern of allergy medication use over the course of 1 year and then compared using descriptive methods.
Results: Seventy-nine percent of the total study sample (80,534 allergy patients) was classified as SAR and 21% as PAR. PAR patients were found to have higher mean levels of allergy-related outpatient payments ($568 vs. $471) and higher mean costs for second-generation antihistamines ($552 vs. $162). PAR patients also had higher levels of comorbidities (asthma, sinusitis, depression, and migraine), higher numbers of concomitant medications (multiple second-generation antihistamines, nasal steroids, other antihistamines, asthma medications, and ophthalmic decongestants), and more immunization encounters.
Conclusions: Approximately 21% of allergic rhinitis patients have perennial symptoms as reflected in their patterns of medication use. Perennial patients have significantly higher allergy-related health-care costs and rates of comorbidities and greater use of concomitant medications. These distinct clinical and resource use profiles may have implications for therapy choices in the cost-effective management of perennial allergic rhinitis patients.
Estimates of the prevalence of allergic rhinitis range from 10% of the US population  to as high as 30% among adults and 40% among children . Regardless of the exact figure, allergic rhinitis is the most prevalent chronic allergic respiratory disease  and is one of the most common chronic health conditions in the United States [4,5]. Very little information is available on the prevalence of perennial allergic rhinitis, although Naclerio and Solomon  note that 21% of allergic rhinitis patients are bothered by their symptoms for at least 41 weeks per year.
Allergic rhinitis rarely results in hospitalization. Nevertheless, its high prevalence results in substantial direct medical costs. Using data from the 1987 National Health Interview Survey and the 1988 National Ambulatory Care Survey, McMenamin  estimated that the direct health-care costs of treating allergic rhinitis were $1.16 billion in 1990 dollars. Using data from the 1987 National Medical Expenditure Survey, Malone et al.  estimated the direct medical costs of allergic rhinitis to be $1.15 billion in 1994. Storms et al. , however, using data from a nationwide sample of 15,000 households, estimated that the prescription and office visit costs associated with allergy was $3.5 billion.
Allergic rhinitis can be either seasonal or perennial (i.e., year-round) in duration. Most studies of allergic rhinitis do not distinguish between seasonal allergic rhinitis (SAR) and perennial allergic rhinitis (PAR)—either reporting on both conditions as “allergic rhinitis” or focusing specifically on SAR. SAR and PAR are distinguished by the allergens that trigger symptoms and by the varying duration and time of year of these symptoms. SAR patients react to outdoor allergens (pollens and fungal spores) that fluctuate during the year. Thus, symptoms either occur or are increased during certain seasons. PAR patients exhibit symptoms continuously or intermittently throughout the year and are affected by indoor allergens (house dust, dust mites, animal dander, mold). PAR and SAR may coexist in the same individual with symptoms increasing in severity during certain seasons.
Allergic rhinitis has been linked to a variety of other conditions. Patients with allergic rhinitis often experience fatigue, irritability, mood swings, and cognitive disturbances . A clinical trial reported consistently lower health-related quality of life for PAR patients than for healthy controls . Specifically, PAR takes its toll on sleep, social functioning, and productivity; is associated with nonrhinitis problems of thirst and fatigue; and includes practical problems of frequent nose blowing. Furthermore, untreated allergic rhinitis may lead to more serious diseases in upper and lower airways. Allergic rhinitis is often associated with asthma, sinusitis, otitis media, and polyps . An extensive body of literature documents that allergic rhinitis and asthma frequently coexist [3,12,13]. Although the prevalence of asthma in the general population is 3% to 5%, asthma affects approximately 38% of patients diagnosed with allergies  and, among those patients diagnosed with asthma, 60% to 78% have allergic rhinitis . Not only do allergic rhinitis and asthma coexist, but asthma patients with symptomatic rhinitis use more asthma medications than do asthmatics without rhinitis . It has been postulated that comorbid allergic rhinitis may be related to increased asthma severity. Similarly, investigators have noted links between allergies and sinusitis [14,15], otitis media , and other conditions.
Despite the large literature on comorbidities in the allergic rhinitis population, virtually nothing is known about differences in the patterns of comorbidities between patients with SAR and those with PAR. In addition to being a marker of disease severity, different patterns of comorbidities could have substantial implications for the cost burden of illness associated with SAR and PAR.
This study provides important new information on the epidemiology and economics of PAR and SAR by examining the characteristics and health-care use of patients with seasonal versus perennial patterns of prescription allergy medication use patterns. We first present an algorithm for classifying patients treated with a second-generation antihistamine into seasonal or perennial cohorts based upon patient-level allergy prescription histories. Once this classification is accomplished, we conduct a detailed retrospective analysis of concomitant medication use, clinical profile, and medical resource use patterns of SAR and PAR patients.
The analytic file used in this study was constructed from the 1996 to 1998 MarketScan Fee-For-Service and Encounter databases. The Private Pay Fee-For-Service database contains the health-care experience of privately insured individuals enrolled in several types of health plans, including exclusive provider organizations, preferred provider organizations, point-of-service plans, and indemnity plans. The Private Pay Encounter database contains the health-care experience of individuals covered under fully and partially capitated health plans. Together, the MarketScan Fee-for-Service and Encounter databases contain information on approximately two million covered lives annually. Using unique patient identifiers, outpatient pharmaceutical claims can be linked with the inpatient services, inpatient admissions, and outpatient claims files. The outpatient pharmaceutical claims file provides information on drugs used, therapeutic class, days of intended therapy, number of units prescribed, and pharmacy service payments.
Patients were identified as having allergic rhinitis if they had a minimum of 30 days of therapy on one of the nondecongestant versions of the three leading second-generation antihistamines (cetirizine, fexofenadine, or loratadine). We could not rely on identifying the study population by the ICD-9 code 477.xx because only 47% of patients who had claims for these drugs also had that diagnosis code. Many of these patients may have received a diagnosis of allergic rhinitis prior to 1996, which would not be reflected in the claims data that we analyzed. We recognize that cetirizine, fexofenadine, and loratadine are also sometimes used to treat conditions other than allergy (e.g., atopic eczema) but we found very little evidence of these diagnoses in our data.
Our methodology for identifying patients with allergic rhinitis differs from other studies in the peer-reviewed literature. Most previous studies have identified allergic rhinitis patients based on the presence of an allergic rhinitis diagnosis, either patient self-report or physician coded [7,8]. However, because we found that 47% of the patients with second-generation antihistamine prescriptions lacked a corresponding medical claim containing an allergic rhinitis diagnosis we felt that excluding such a large segment of the sample would dramatically underestimate allergic rhinitis prevalence, as well as potentially introduce bias into our analysis. Thus, we elected to select the study sample on the basis of medication utilization, rather than diagnosis codes alone.
Enrollment data and a claims-based proxy for enrollment were used to identify whether patients were continuously enrolled in their health plan for the study period. Using the enrollment data available for a subset of the population, we identified patients who had a positive enrollment flag between January 1, 1996, and December 31, 1998. For the rest of the sample, we coded patients as continuously enrolled if they incurred service claims before and during the first quarter of the study period as well as during and after the last quarter of the study period. Patients who were not continuously enrolled were excluded from the study sample. After applying the continuous enrollment, preperiod, and study-period criteria, 80,534 patients were included in the final study sample (Table 1).
Table 1. Descriptive statistics for the entire sample population and by seasonal versus perennial allergic rhinitis status
Type of allergic rhinitis
Total number of patients
Mean age (years) in the illness category
Type of benefit plan
Basic/major medical plan
Percentage that were encounter plans
Seasonal Allergic Rhinitis and Perennial Allergic Rhinitis Classification
People with seasonal allergic rhinitis commonly present with seasonal symptoms in the spring or fall because SAR is caused by either mold spores or tree, grass, or weed pollens. In general, trees pollinate in the spring, grasses pollinate in the summer, and weeds pollinate in the fall. The timing of these different seasons varies somewhat by region. Conversely, people with perennial allergic rhinitis usually have allergy symptoms throughout the year. PAR is most often due to sensitivity to dust, dust mites, animal dander, or mold spores. Because of the timing differences in the occurrence of symptoms for SAR and PAR patients, we were able to classify patients according to the patterns of their claims for allergy medication. We hypothesize that PAR patients will have claims for allergy medications throughout the year, while SAR patients will have claims for allergy medications predominantly around pollen allergy seasons: April, May, and June for the spring allergy season and August, September, and October for the fall allergy season. We observe that the empirical markers of allergy seasons indicated by the patterns of claims over time differed by region—a finding consistent with the geographic variation in seasonality of allergens .
Patients were classified as having a seasonal persistent allergy pattern if they had at least one period of continuous therapy lasting a minimum of 45 days and fewer than eight prescriptions during the 12-month follow-up period; those having 4 months or less of prescription therapy and no period of continuous therapy of at least 45 days were classified as having a seasonal intermittent pattern (Fig. 1). The two seasonal groups were combined to form the SAR group.
Patients were classified as having a perennial persistent allergy pattern if they had eight or more prescriptions during the 12-month period following initial treatment; those having more than 4 months of prescription therapy and no period of continuous therapy of at least 45 days were classified as perennial intermittent (Fig. 1). The two perennial groups were combined to form the PAR group. Although our classification method for patients with perennial allergic rhinitis does not distinguish between those patients who have year-long symptoms with seasonal exacerbations from those without seasonal exacerbations, it does enable us to compare the characteristics and health-care use patterns of patients with year-long symptoms to those with seasonal symptoms alone, as indicated by patterns of prescription allergy medication use.
Chi-square tests were used to compare the distributions of categorical variables between the SAR and PAR groups; two-tailed t tests were used to compare mean visits and mean expenditures.
Regional Distributions of Prescription Volume by SAR/PAR Status
After identifying SAR and PAR patients, we used the regional definitions of the US Bureau of the Census (Fig. 2) to analyze the distribution of allergy prescription claims over a 12-month follow-up period. As expected, the allergy medication claims for SAR patients were concentrated during the months between April and June (Fig. 3). There was another, smaller spike in SAR allergy prescriptions in September, but prescription counts fell in October, November, and December. In contrast, allergy medication claims for PAR patients were distributed evenly over the year.
Comparisons by SAR/PAR Status
Demographic and health insurance characteristics of the SAR and PAR samples are reported in Table 1. In both samples, the majority of the patients were women but the proportion of women was higher in PAR patients than SAR patients (65 vs. 60%, P < .0001). Among the five classifications of health plans, a plurality of the patients (35.4%) had comprehensive indemnity plans, followed by preferred provider organizations (24.4%); both types of health plans were more common among PAR than SAR patients (P < .0001). PAR patients tended to be older than SAR patients (mean age 44 years for PAR vs. 39 years for SAR, p < .0001).
Overall, the sample was concentrated in the North Central and Southern regions, similar to the distribution of the MarketScan population in general. However, the distribution of SAR and PAR patients differed across regions. The percentage of the national SAR patient population located in the Northeast was higher than that of PAR patients (15.6 vs. 11.8%, P < .0001), lower in the North Central (40.0 vs. 46.2%, P < .0001), higher in the South (32.7 vs. 30.3%, P < .0001), and not statistically different in the West. Within each region, SAR patients dominated—ranging from 77% of the allergic rhinitis population in the North Central region to 84% in the Northeast.
We searched for indications of several comorbid conditions of interest: asthma, depression, migraine, otitis media, and sinusitis (Table 2). Patients were classified as having comorbid conditions if they had claims with ICD-9 codes associated with the conditions during the 12-month study period or in the 12 months prior to the index prescription date. SAR patients were more likely to be free of comorbidities than PAR patients (42.0 vs. 35.9%). The two groups were about equally likely to have one comorbid condition (35.8% for SAR, 36.4% for PAR), but PAR patients had a higher likelihood of having two or more comorbidities (27.3 vs. 22.1%).
Table 2. Comorbidities among patients with allergic rhinitis, by seasonal versus perennial allergic rhinitis status
Type of allergic rhinitis (%)
Seasonal (N = 63,960)
Perennial (N = 16,574)
Number of comorbidities
Presence of comorbid conditions
Prevalence rates of most comorbid conditions varied significantly by PAR/SAR status. We found statistically significant and higher rates of comorbidities among PAR patients relative to SAR patients in asthma (21.6 vs. 14.8%, P < .0001), sinusitis (45.7 vs. 41.1%, P < .0001), depression (13.4 vs. 11.6%, P < .0001), migraine (6.1 vs. 5.5%, P < .0001), and nasal polyps (0.5 vs. 0.4%, P < .0303). Only otitis media did not have significantly different rates in the SAR and PAR samples.
Patients were coded as taking a concomitant medication of interest if they had at least one claim for those medications that overlapped with a prescription for one of the three-second generation antihistamines during the study period (Table 3 and Fig. 4). The window for overlap was defined as the prescription date for the study drug plus the number of days of therapy plus 30 days. We did this to account for small gaps in therapy that might lead one to conclude that patients were not being treated with concomitant medications when, in fact, the gaps were due to delays in refills. A significantly higher percentage of PAR patients than SAR patients had claims for one or more of the concomitant medications (68.4 vs. 46.8%). More SAR than PAR patients used only one second-generation antihistamine during the 12-month study period (86.9 vs. 81.5%, P < .0001). A higher percentage of PAR patients than SAR patients used second-generation antihistamines with nasal steroids (50.2 vs. 31.4%, P < .0001), with asthma medications (30.6 vs. 17.6%, P < .0001), or with other antihistamines (18.3 vs. 10.5%, P < .0001).
Table 3. Use of concomitant medications among patients with allergic rhinitis, by seasonal versus perennial allergic rhinitis status
Type of allergic rhinitis (%)
Seasonal (N = 63,960)
Perennial (N = 16,574)
Number of concomitant medications
Use of medications of interest
One second-generation antihistamine
More than one second-generation antihistamine
Second-generation antihistamines with nasal steroids
Second-generation antihistamine with other antihistamines
Second-generation antihistamine with asthma medications
Second-generation antihistamines with ophthalmic decongestants
PAR patients also tended to use more allergy-related services and had higher allergy-related payments (Table 4 and Fig. 5). Table 5 indicates that PAR patients on average had more allergy-related outpatient visits than SAR patients (15.1 vs. 10.0, P < .0001), as well as more immunization visits for allergy desensitization (22.1 vs. 17.8, P < .0001). Mean immunization visits exceed mean allergy-related outpatient visits for the PAR and SAR groups because these means were calculated for service utilizers only. PAR patients also had higher average total payments for allergy-related outpatient visits ($568 vs. $471, P < .0001). Not surprisingly, PAR patients had higher average payments for the three second-generation antihistamines ($531 vs. $162, P < .0001).
Table 4. Health-care utilization and expenditures among patients with allergic rhinitis, by seasonal versus perennial allergic rhinitis status
Allergy-related health-care utilization and expenditures
Type of allergic rhinitis
Number of allergy-related outpatient visits
Number of immunization visits for allergy desensitization
Total payments for allergy-related outpatient visits
Total payments for fexofenadine, loratadine, and cetirizine
Total payments for nasal steroids
The seasonal claims patterns varied by region, although the strong seasonality of claims overall suggests that our method of identifying SAR and PAR patients was sound. The volume of SAR prescription claims peaked in the spring and with a smaller increase in the early fall, offering face validity to the classification algorithm. The spring spike in prescription volumes was especially pronounced in the Northeast and the West. The apparent fall spike in prescription volume among SAR patients in the North Central region is actually due to lower volume in the spring allergy season relative to other regions. The lower spring volume exaggerates the spike in the fall. In both the Northeast and North Central regions SAR prescription volume declines in the winter months.
The allergy seasons in the Northeast and North Central appeared to lag those in the South and West by about 1 month. The spring allergy season in the West and South appeared to start in March and peak in April. Although there was some evidence that prescriptions began to increase in March for SAR patients in the Northeast and North Central regions, prescription activity there accelerated strongly in April and peaked in May.
In contrast to the SAR graphs, the distribution of PAR prescriptions is not only remarkably constant across months but also across regions (Fig. 3). Both results indicate that the prescriptions of PAR patients are insensitive to the timing of allergy seasons—a finding consistent with properly classifying patients into the SAR and PAR categories.
After assigning patients to the SAR or PAR categories we found a greater incidence of comorbid conditions of interest and higher health-care use and allergy-related expenditures for PAR patients relative to SAR patients. Since expenditures were measured using total payments, differences in allergy-related payments between PAR and SAR patients reflect the perspective of third party payers. One possible explanation for the observed differences in health-care use and allergy-related expenditures between the SAR and PAR groups is differences in diagnostic procedures associated with the two categories of allergic rhinitis. SAR and PAR are diagnosed by a combination of clinical tests and patient medical history. The cyclic nature of SAR symptoms enables relative ease of identification. In contrast, the overlap of PAR symptoms with chronic sinusitis and vasomotor rhinitis may complicate detection. In general, the greater difficulty in diagnosing PAR would be expected to require more extensive clinical testing to confirm the perennial diagnosis, to uncover allergens, and to guide treatment . In turn, this more extensive clinical testing would be expected to result in higher diagnostic costs.
A second potential explanation for the observed differences in health-care resource use and allergy-related expenditures between the SAR and PAR groups is simply that the therapies used to treat allergic rhinitis are used in greater quantities by patients with PAR than those with SAR. The cost of second-generation antihistamine therapy for PAR patients was comparable to the cost associated with outpatient visits. However, second-generation antihistamine costs for PAR patients were more than three times those for SAR patients. There are numerous pharmaceutical treatments available to combat the symptoms of allergic rhinitis including decongestants to relieve nasal congestion and pruritus of the eye; antihistamines to relieve sneezing, itching, and prevention of nasal congestion prior to attack; and corticosteroid nasal sprays to reduce inflammation. The year-round nature of PAR is likely to increase expenses for all types of therapy because patients experience symptoms more often.
The findings are consistent with PAR patients having a greater average disease severity than SAR patients. This possibility has significant economic implications. Several studies have estimated substantial economic costs from allergies and the treatment of allergy symptoms with first-generation antihistamines [7–9,18]. Storms et al.  estimated that direct allergy-related medical costs for prescriptions and outpatient visits were $3.5 billion.
Estimates of the direct medical costs associated with allergic rhinitis, however, almost surely understate the economic burden of the condition. A number of studies have shown that allergic rhinitis is frequently comorbid with other conditions [3,11–13]. This pattern of comorbid conditions was found in the current study as well. Although we did not find evidence of a substantial comorbidity of allergic rhinitis with otitis media, this result was expected because we examined only solid oral dosage forms of medications. As a result, patients treated with pediatric forms of the medications were not included in our sample. Moreover, it was found that the prevalence of comorbidities was higher among the PAR group than the SAR group (Table 2). For example, the prevalence of diagnosed asthma was 21.6% in the PAR group and 14.8% in the SAR group. The higher prevalence of certain comorbidities among PAR patients suggests that, not only their allergy-related costs, but also their total direct medical costs, will be greater than those of SAR patients. Crystal-Peters et al.  found that treatment for allergic rhinitis can reduce emergency room use and hospitalizations for patients with comorbid asthma. Thus, the economic implications of important comorbidities with allergic rhinitis, and how these implications may vary for SAR and PAR patients, are areas in need of further study.
Collis and Pellegrini  report that as many as 75% of workers with allergies are either absent from work or perform less productively at work for 2 or more weeks per year. Studies of the economic impacts of allergic rhinitis, however, have implicitly assumed that the impacts are confined to spring and fall allergy seasons. Our algorithm resulted in 21% of patients being classified as PAR patients. For these patients, the potential economic implications could be year-long, suggesting that previous studies may have substantially underestimated the economic implications of allergic rhinitis and the higher indirect costs incurred when first-generation antihistamines are utilized.
Findings from this study should be evaluated in light of several potential limitations. We have not attempted to identify those patients who have both PAR and SAR. These patients would have been classified as per our algorithm into the PAR group. Identifying patients with both PAR and SAR would require clinical and diagnostic information unavailable to us in the claims data. It is not possible to distinguish perennial allergic rhinitis from nonallergic rhinitis or their combination without the patient's medical history and skin tests. Nevertheless, we feel that the prescription-based algorithm for classifying patients is useful for distinguishing between patients with seasonal versus perennial symptomotology. Future work that attempts to identify patients with both PAR and SAR would be helpful in arriving at improved prevalence estimates for the two populations.
The database analyzed in this study, although very large, is not necessarily nationally representative. The health-care claims contained in the database represent the health-care experience of employees or their dependents primarily of Fortune 200 companies. As such, the health benefit coverage that these individuals have is relatively generous compared to the national norm. Moreover, the geographic distribution of the individuals represented in the claims differs somewhat from the geographic distribution of the US population. This could affect the relative balance of SAR and PAR patients identified by our algorithm if the geographical distribution of patients in our analysis differs from the national distribution of the population in regions with strong allergy seasons.
This article uses prescriptions and prescription refills to identify patients likely to have allergic rhinitis. This approach represents an alternative to relying on diagnosis codes to identify allergic rhinitis patients but it has its strengths and weaknesses. Its primary strength is that it likely to capture nearly everyone with moderate to severe allergic rhinitis symptoms. In our data, we found that patients having a series of prescriptions indicative of allergic rhinitis often did not have a diagnosis code for the condition in their medical claims. There are several possible reasons for this—in particular, that these patients may have received an allergic rhinitis diagnosis prior to the study period. As a consequence, to rely on ICD-9 codes could result in a substantial undercount of patients. The pattern of missing data would probably be unbalanced as well. SAR patients experience symptoms in fewer months per year and thus would be less likely to receive an allergic rhinitis diagnosis than PAR patients.
Yet there is no bright line between PAR and SAR in terms of prescriptions patterns. To help distinguish between them, we required that patients classified as having SAR needed to have at least one 45-day period of therapy but less than 8 months of therapy in total over the course of a year. The results in Fig. 3 imply that the algorithm worked reasonably well: the PAR patients have very steady levels of prescriptions in every month, whereas the SAR patients show strong seasonality in prescription use.
Although use of inhaled corticosteroids and other drugs are also used to treat allergic rhinitis, in the absence of an associated diagnosis code for allergic rhinitis, we would run the danger of including patients with other conditions for which these drugs are frequently used (e.g., asthma) in our SAR/PAR categorizations. In contrast, during the time period of this study second-generation antihistamines were indicated only for allergic rhinitis. The estimated prevalence rates for the allergic rhinitis group identified in this fashion is approximately 10%. This is the same as that obtained from recent estimates from the National Health Interview Survey.
A related issue is the reliability of diagnosis codes for identifying the presence of comorbidities. Just as allergic rhinitis diagnoses were apparently missing for a substantial proportion of the observations, so too may diagnoses for various comorbidities be missing. Since SAR and PAR patients are both observed for a 1-year period in the study there is no apparent reason to expect that diagnosis codes indicative of comorbidities would be differentially missing for SAR versus PAR patients. However, it is likely that the degree of comorbidity is understated for both groups.
This study presents data on the epidemiology and health-care resource use of patients with seasonal or perennial patterns of prescription use. Patients with seasonal patterns of allergy prescription refills had different demographic characteristics, clinical profiles, and prescription patterns than patients with perennial refill patterns. Perennial patients had a higher number of allergy-related visits, higher allergy-related expenditures, and greater use of concomitant medications than seasonal patients. Perennial patients were also found to have a greater incidence of comorbid conditions (asthma, sinusitis, depression, and migraine) than seasonal patients. In particular, higher proportions of perennial patients than seasonal patients used second-generation antihistamines with nasal steroids or with other antihistamines. Such differences in the clinical characteristics and health-care use have important implications for the choice of therapy and cost of treating seasonal and perennial patients.