INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The goals of asthma management plans recommended by U.S., Australian, U.K., Canadian, and Japanese consensus panels are similar (1). Patients with asthma should have as few symptoms and as little interference with normal activities as possible, no (or infrequent) need for short-acting inhaled -agonists, the best achievable lung function, and minimal concern about drug-related side effects. These consensus panels have all advised a stepped care pharmacotherapy approach to achieving these goals. Stepped care pharmacotherapy is based on two premises. In 1981 Aas emphasized that the choice of drugs in asthma therapy should take into account anticipated effects in different subsets of asthma patients; subsets of asthma severity could be distinguished by symptom assessments and physiologic lung function (7). Bernstein in 1985 recommended that asthma pharmacotherapy should be determined by balancing the risks of the drugs used against the possible benefits (8). In patients with more severe disease the risks from increasingly intensive therapy are warranted because the potential benefits are larger.

Balancing the risks and benefits of drugs used in asthma pharmacotherapy according to subsets of disease requires an objective method for assessing asthma severity. At present there is no agreement among asthma consensus panels on methods for categorizing asthma severity. The U.S. allergy consensus panel and the Canadian consensus panel state that categorizing asthma severity is not possible and provide only general descriptors of severe asthma (2, 5). The U.K. consensus panel defines severity in terms of the treatment needed to achieve asthma management goals, an approach which precludes balancing risks and benefits prior to initiation of therapy (4). Only recommendations from the Expert Panel II, convened by the National Asthma Education and Prevention Program (NAEPP) (1), and the Australian National Asthma Campaign (3) provide a sufficiently detailed approach to allow categorization of asthma severity. The Expert Panel II method for categorizing asthma severity relies on an assessment of asthma symptoms during the day, nocturnal asthma symptoms, and physiologic measures of lung function (1). Australian recommendations are based on similar variables, plus a history of hospitalizations or emergency room visits for asthma and previous life-threatening asthma attacks (3).

During the clinical development program for chlorofluorocarbon (CFC)-free beclomethasone dipropionate (BDP) (Qvar; 3M Pharmaceuticals, St. Paul, MN), data were collected from a large number of patients with asthma before randomization into clinical trials. This data included daily diary card records of daytime asthma symptoms, nighttime sleep disturbances caused by asthma symptoms, morning (A.M.) and evening (P.M.) peak expiratory flow (PEF), use of inhaled -agonists for symptom relief, and in-clinic spirometry measures of the forced expiratory volume in one second (FEV₁), but no information on prior hospitalizations for asthma. The availability of this large data set allowed an assessment of the contribution each variable individually made to asthma severity categorization in the Expert Panel II method. Because the worst individual variable categorization in the Expert Panel II approach determines overall classification, it was hypothesized that differences in categorization among variables might result in a disproportionate effect on overall asthma severity categorization by a single variable.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patient Characteristics

Data from patients entered into the run-in, pretreatment phase of five controlled clinical trials were pooled for this analysis. Before entry into these trials, all patients signed an informed consent approved by an appropriate institutional review board or ethics committee. The trials were performed in the United States, United Kingdom, France, Germany, Poland and Czechoslovakia between November 1994 and December 1996. Patients who had hospitalizations, emergency room visits, changes in asthma medications, or use of intramuscular, intravenous, or oral steroids in the 4 wk before the run-in period were not allowed into these studies. Details of asthma control were not obtained before this period. Other common inclusion criteria for these trials were: males and nonpregnant females, age 18 to 65 yr,  12% reversibility to a short-acting inhaled -agonist, and demonstrated proper press-and-breathe metered dose inhaler technique. Only one study allowed current smokers. In general, patients with  15 pack-years and a recent (within 6 mo of study entry) smoking history were excluded. Other significant medical or surgical conditions were also reasons for exclusion.

All patients used short-acting inhaled -agonists for control of asthma symptoms prior to and during the run-in. Inclusion criteria restricted the use of theophylline products, oral -agonists, inhaled anticholinergics, long-acting inhaled -agonists, nedocromil, and cromolyn during the run-in period. Use of inhaled steroids varied per specific trial protocols. Two trials specified that patients were to be inhaled steroid nonusers for at least 3 wk before study entry. One trial allowed either inhaled steroid nonusers or patients stabilized on up to 400 µg/d of CFC-BDP. Two trials required that patients had to be using at least 400 µg/d, but not more than either 800 or 1,000 µg/d of CFC-BDP or equivalent for at least 4 wk before run-in. Three trials required that patients had active asthma symptoms (documented by either regular use of short-acting inhaled -agonists or diary card recordings of asthma symptoms) during the run-in for study entry.

Clinical Data Collection

Each trial had at least a 1-wk run-in period during which patients remained on their previously prescribed asthma medications. Patients recorded on a diary card each evening of the run-in the severity of four separate asthma symptoms (wheeze, cough, chest tightness, and shortness of breath) experienced during the day based on a six-point scale (0 = none; 1 = symptom present but caused little or no discomfort; 2 = mild symptom that became annoying; 3 = moderate symptom that caused discomfort; 4 = severe symptom that interfered at least once with normal daily activities; and 5 = symptoms so severe that normal daily activities could not be performed). Each morning patients recorded on the same diary card the severity of asthma symptoms experienced during the previous night based on a five-point scale (0 = no asthma symptoms; 1 = asthma symptoms causing wakening once or early wakening; 2 = asthma symptoms causing wakening twice or more, including early wakening; 3 = asthma symptoms causing patient to be awake most of the night; 4 = asthma symptoms so severe that patient did not fall asleep). Patients also recorded on this diary card their daily use of short-acting inhaled -agonist bronchodilator. To enter each trial, compliance with diary card completion during the run-in period was required.

Patients performed three PEF maneuvers each morning and evening with a MiniWright peak flow meter and recorded the best value on the diary card. Spirometry was performed in the clinic, before noon, just prior to the start of the run-in period using equipment and reproducibility standards established by the American Thoracic Society (9). The FEV₁ was expressed as percent predicted using standard nomograms (10). Patients were advised not to use asthma medications for 6 h prior to in-clinic spirometry.

Asthma Severity Categorization

Rules for grading asthma symptoms and nocturnal symptoms were adapted from the Expert Panel II recommendations (1) to the diary card information available (Table 1). These recommendations did not specify which asthma symptom or lung function parameter should be used for classification purposes. Initial analyses using the kappa statistic were performed to determine agreement among the four different daytime asthma symptoms recorded on patient diary cards. Although agreement for classifying asthma severity between symptoms was low, wheeze seemed to agree most consistently with the other three symptoms (Table 2). Consequently, in the primary analysis wheeze was chosen as the symptom used for classification. In a secondary analysis, instead of wheeze, the worst grade from all four asthma symptoms tracked on the diary cards was used for categorization. Included within the Expert Panel II's method for assessing asthma symptoms was use of short-acting inhaled -agonist for symptom relief. Because information on -agonist use was collected separately on the diary cards, in this analysis -agonist use was considered an independent variable.

An initial analysis was also performed to determine agreement in severity categorization among physiologic measures of lung function. The FEV₁ and A.M. PEF performed similarly in classifying severity of asthma. In a previous analysis (data not shown) PEF variability had been found to underclassify the severity of asthma compared with FEV₁ and A.M. PEF. Because patients were to be assigned to the most severe grade according to the Expert Panel II classification recommendations, PEF variability was not used for classification purposes. To simplify the analysis, FEV₁ was selected as the physiologic measure of lung function for categorization.

In the primary analysis asthma severity categorization was performed on inhaled steroid nonusers, because the recommendations specified that categorization should be performed before treatment. The analysis was repeated for the combined data set of inhaled steroid users and nonusers. A further analysis was performed on U.S. and European patients separately. As a secondary analysis, patients were classified by both asthma severity and their dose of inhaled steroids. Daily dosages of inhaled steroids (expressed as ex-actuator) were categorized as low, medium, and high according to criteria established by the Expert Panel II (1). For CFC-BDP low dose was 168 to 504 µg/d, medium dose was 504 to 840 µg/d, and high dose was > 840 µg/d. The low dose for flunisolide was 500 to 1,000 µg/d, the medium dose was 1,000 to 2,000 µg/d, and the high dose was > 2,000 µg/d. Triamcinolone was categorized similarly to flunisolide, except the low dose was 400 to 1,000 µg/d. Very few patients used either budesonide or fluticasone.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Inhaled steroid nonusers had similar demographic and baseline characteristics as the patients in the combined data set (Table 3). European and U.S. patients were similar in most respects, except slightly more European patients used nasal steroids and long-acting inhaled -agonists. The majority of patients were female. Patients were mostly young, white adults with moderate airway obstruction who regularly used short-acting inhaled -agonists for asthma symptom relief. Use of other asthma medications was uncommon in accordance with protocol stipulations.

In the primary analysis, the individual variables performed differently in classifying asthma severity. Nighttime asthma symptoms classified the majority of inhaled steroid nonusers as severe, persistent, but wheeze classified 27.3% as mild, intermittent and 25.7% as mild, persistent (Table 4). -Agonist use and FEV₁ percent predicted were intermediate in severity categorization; both classified most as moderate, persistent. When the worst categorization from all four daytime asthma symptoms was used, most patients were categorized as severe, persistent. However, wheeze alone contributed little to overall categorization; categories were not affected by excluding wheeze. Similar results were found in the combined data set (Table 5) and for asthma severity categorization performed separately for U.S. and European patients (Table 6). There was poor correlation between individual variables in categorizing severity of asthma in both the inhaled steroid nonusers (Table 7) and the combined data set (data not shown).

Patients in the combined data set were classified by both overall asthma severity and dose of inhaled steroid (Table 8). The vast majority of patients were receiving lower doses of inhaled steroids for their asthma severity category than recommended, usually by two or three dose levels. In other words, patients with moderate severity asthma were frequently not on inhaled steroids (two dose level disparity) and patients with severe asthma were often either not on inhaled steroids (three dose level disparity) or on low-dose inhaled steroids (two dose level disparity).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The results of this analysis support the hypothesis that a single variable, in this case nocturnal symptoms, might be dominant in determining overall asthma severity categorization using the NAEPP Expert Panel II approach. Because at present there is no agreed upon "gold standard" for categorizing asthma severity, it is not possible to validate the asthma severity categorization results obtained in this analysis. However, methodological issues suggest that the severity categorization results should be viewed cautiously. These methodological issues relate to the categorization variables recommended by the Expert Panel II and the stipulation that the worst individual variable category should determine overall categorization.

The NAEPP Expert Panel II recommends the use of three variables, asthma symptoms during the day, nocturnal asthma awakenings, and physiologic measures of lung function, for categorizing asthma severity (1). Use of short-acting inhaled -agonists for symptom relief was included within the variable of daytime asthma symptoms in the Expert Panel II methods, but was treated as an independent variable in this analysis. Each of these variables provides valuable information about asthma severity. Patient perception of asthma symptoms is fundamentally important, but reporting of asthma symptoms may not correlate well with physiologic measures of airway obstruction (11, 12). FEV₁ is an agreed upon measure of disability and impairment in asthma (13). Including both asthma symptoms and physiologic lung function as separate variables takes this disparity into account. Separating -agonist use from asthma symptoms is appropriate because patients self- titrate their asthma symptoms with the use of short-acting inhaled -agonist on an as-needed basis. Consequently, as asthma worsens, patients could either experience more troublesome symptoms or use -agonists more regularly. In this analysis, there was poor agreement between -agonist use and symptoms. Clinicians recognize the independent value of monitoring -agonist use in assessing asthma severity. Nocturnal asthma symptoms have been recognized as an important marker of severe asthma (14). The Australian guidelines include other variables for asthma severity categorization purposes (3). A recognized marker of severity is history of a near-fatal asthma episode (2, 5). Asthma severity may fluctuate over time, but near-fatal episodes should indicate a higher level of concern. Prior hospitalizations for asthma (15) and prior use of oral or systemic steroids (2) also suggest more severe asthma. Emergency room visits for asthma management may indicate severe asthma, but may also simply reflect inadequate delivery of health care (2). Neither the physician assessent (16), nor the physical examination (17, 18) have proven to be accurate reflectors of asthma severity. Auscultation of wheezes is not a sensitive or specific predictor of asthma severity (19).

Methods for developing clinical prediction rules (20, 21) use mathematical techniques to determine the independent contribution of individual variables to the ability to predict a clinical outcome or to guide a clinical course of action. These methods might have been useful in examining the value of interactions among the individual variables proposed by the Expert Panel II for scoring asthma severity, as well as other possible variables. The requirement to use the worst grade from an individual variable to determine overall asthma severity categorization minimized the contribution of most variables in this analysis and allowed one variable, nocturnal asthma awakenings, to categorize most patients as severe, persistent. FEV₁ may have had a small effect in shifting patients with mild symptoms into the moderate, persistent category.

Other methodological issues relate to scoring of each categorization variable. The Expert Panel II methods include the option of using multiple parameters for assessing individual variables. For instance, either A.M. PEF, PEF variability, or FEV₁ could be used as a physiologic measure of lung function (1). Reddel and coworkers (22) found A.M. PEF to be a more useful reflection of asthma severity than PEF variability. Initial analysis in this study showed that asthma severity was categorized similarly by FEV₁ and A.M. PEF. This is an important observation because it allows flexibility in evaluating physiologic lung function. Spirometry is performed surprisingly infrequently by physicians (especially primary care physicians) managing patients with asthma (23, 24). The Expert Panel II refers to various daytime asthma symptoms, but does not specify whether a single symptom should be graded or all possible symptoms. If only a single symptom were used, as wheeze was in this analysis, it added little to overall categorization. Using the worst category from all four daytime asthma symptoms resulted in most patients being classified as severe, persistent. Furthermore, methods for collecting data pertaining to daytime asthma symptoms and nocturnal symptoms have not been described by the Expert Panel II (1). It is highly likely that using a structured diary card approach will result in capturing symptom occurrence more frequently than clinicians recognize through the usual clinical interview. This may be a consequence of either the patient failing to remember the symptoms or the clinician not having adequate time to fully explore the history. The advantages of a structured daily diary card may be especially relevant to nocturnal symptoms, which patients may not associate with asthma (25).

The NAEPP panel used consensus, rather than an evidence-based approach, in formulating recommendations for categorizing asthma severity (26). Other methods have been proposed for severity categorization (15, 27), including a method proposed by Ellman and colleagues which has been validated against asthma deterioration over time (15). An accepted, validated asthma severity categorization method should allow appropriate balancing of risks and benefits for various possible pharmacotherapeutic approaches and improve compliance with NAEPP guidelines on asthma care in health care systems (32). It would be an important step toward the development of consistent international asthma management plans (33) and would facilitate clinical research on emerging new pharmacotherapeutic approaches to asthma. Of fundamental importance would be learning how treatment will alter the natural history of asthma, e.g., resolution of the disease, fatal attacks, severe disability. Categorizing asthma severity will also provide a basis for interpreting changes in asthma morbidity and mortality in different populations and for normalizing patient populations so that outcomes and costs of therapy could be compared. Asthma management varies around the world (34) and there is still substantial morbidity and undertreatment of asthmatics (35, 36).

A strength of this analysis was the consistency of data collection. Methods for assessing daytime and nocturnal asthma symptoms and recording -agonist use for symptom relief were consistent across the five clinical trials. Only patients compliant with diary card completion and attendance at visits were entered into the clinical trials. Lung function was measured using standard clinical techniques. The diagnosis of asthma in all patients was established through a clinical evaluation and was supported by a history of reversibility to inhaled -agonist. The 744 inhaled steroid nonusers and the 1,429 patients in the combined data set make this the largest study of this issue presently available. The demographic characteristics of this population are similar to those described in a large health maintenance organization, making this study population especially relevant (37). However, complete historical information on asthma control, specifically on life-threatening episodes of asthma, previous hospitalizations for asthma, and need for systemic or oral steroids, was not obtained. Furthermore, the vast majority of patients were white; studies on this issue are needed in patients of different ethnic backgrounds.

A potential weakness of this analysis was that the clinical trials were intended to support regulatory submissions for a CFC-free inhaled steroid. Protocols specified patient populations with persistent symptoms; improvements in efficacy would be most clearly seen in these groups. Inclusion criteria for these protocols were specifically designed to capture patients with mild, moderate, and severe disease. Consequently, the results of this analysis should be applicable to the general population of patients with asthma.

The Expert Panel II specified that asthma severity should be determined before treatment is started (1). This rule limits applicability of this categorization method to only a small portion of asthma management. It also obscures a critical issue in asthma management; there is a distinction between asthma severity and control of asthma symptoms and lung function (38). Patients may have severe disease, but, with the benefits of pharmacotherapy, may have well-controlled asthma with minimal symptoms and relatively normal lung function. It would be worthwhile if methods were available for categorizing both asthma severity and control. Of note, in this analysis patterns of categorization were similar for the combined data set (which included inhaled steroid users and nonusers) to those found in the inhaled steroid nonusers. This demonstrates both severe asthma and poor asthma control. As might have been expected from protocol requirements, patients entering these clinical trials were receiving much lower doses of inhaled steroids than recommended by the Expert Panel II (1). A critical objective of all asthma management guidelines is to ensure optimal anti-inflammatory therapy for all patients with asthma.

The stepped care approach is intuitively appealing because it balances the benefits of therapy against the risks. This benefit/risk assessment particularly applies to dosing inhaled steroids. Of fundamental importance in using a stepped care approach to asthma pharmacotherapy is establishing an objective means of categorizing asthma severity. Unfortunately, methodological issues suggest that asthma severity categorization methods, such as is recommended by the NAEPP Expert Panel II, should be used with caution. A validated method for categorizing asthma severity would be of considerable importance to researchers, health care providers, and patients.