Asthma severity: a factor analytic investigation

Asthma Severity: A Factor Analytic Investigation WILLIAMC. BAILEY,M.D., DARLENEM. HIGGINS,BONNIEM. RICHARDS, JAMESM. RICHARDS,Jr., Ph.D., Birmingham, Alabama

PURPOSE: Reviewers of the asthma research litmature have called for improved questionnaires and other measures, particularly for assessing the severity of asthma. To help meet this need, standard multivariate and psychometric techniques were used with data from asthma patients to develop and evaluate a series of scaled questions. Since there is no “gold standard” for -ing asthma severity, we hope this analysis will help improve our ability to more precisely define these important parameters. PATIENTSANDM~WHODSZ~~~ wereco&cted through interviews and review of clinic records for 199 adult patients with asthma from a university clinic population. For evaluating the severity of asthma, eight scales asses4 asthma duration, the incidence of asthma symptoms, the extent to which asthma is an inconvenience to patients, the incidence of respiratory d&eases, medication regimens, medication side effects, and health care utilization. Forced expiratory volume in 1 second as a percentage of predicted normal was included as an objective measure of puhnonary function. A physician rating scale assessed the severity of the underlying disease, not the severity of a particular episode, as either (1) mild (infrequent attacks with interim symptomatic treatment), (2) moderate (more frequent attacks with continuous daily treatment), and (3) severe (continuous symptoms with continuous multiple drug regime- including some systemic steroids). RESULT& In the current aklysis of data from adult asthma patients, the scales correlated positively with a physician judgment scale. Factor analysk with an oblique rotation yielded three factors that provided a concise summary of asthma severity. We have named the factors From the Division of Pulmonary and Critical Care Medicine (WCB), the Lung Health Center (DMH, BMR), and the Dffice of Educational Development (JMR), University of Alabama at Birmingham, Birmingham, Alabama. This study was conducted at the Lung Health Center, University of Alabama at Birmingham, Birmingham, Alabama. This work was supported in part by Grant ROlHL31481-03 from the Division of Lung Diseases, National Heart, Lung, and Blood Institute. Requests for reprints should be addressed to William C. Bailey, M.D., University of Alabama Lung Health Center, 619 South 19th Street, Birmingham, Alabama 35233. Manuscript submitted August 21, 1991, and accepted in revised form April 10. 1992.

September

(A) Symptom Intensity, (B) Airflow Impairment, and (C) Management Intensity. CONCLUSION:hthmaseVerity appearstobe multidimensional rather than unidimensional, including at least three components. The physician rating scale, in combination with measures of the three identified factors, could easily be included in other asthma research protocols to provide a standard, brief assessment of asthma severity and might thus promote greater comparability among studies.

A

sthma is a common disease with great variation in both severity and etiology. This diversity leads to problems for diagnosis and treatment, and more uniform assessments of asthma are needed, particularly for asthma research. Several reviewers have called for improvement in the quality of questionnaires and other measures used in asthma research [l-4]. Sets of questions about respiratory symptoms have been developed by both the British Medical Council [5] and the American Thoracic Society [6], and epidemiologic studies using these and similar measures have shown that respiratory symptoms associated with some respiratory diagnoses [7-lo] are prevalent in numerous countries. A variety of measures are currently used to assess asthma severity; most are logical and seem reasonably effective but have never been scientifically tested in a comprehensive manner. However, simple brief measures are needed for greater specificity in describing variables and in assessing respiratory symptoms and other aspects of asthma [ 11. Uniform assessment of asthma severity is an especially important need, for third-party payers are increasingly requiring documentation of severity to determine eligibility for hospitalization and other forms of health care. In addition, research on outcomes is increasingly important, and such research requires that disease severity be considered when analyzing the impact of any intervention. Since there is no agreed-upon “gold standard” for assessing asthma severity, creative approaches are necessary to more precisely define these important parameters. Several previous researchers have addressed the broad issue of assessing severity. In reviewing the heterogeneity of bronchial asthma, Aas [ll] described five categories of severity using simple clini1992

The Amerlcen

Journal

of Medlclne

Volume

93

263

cal parameters and lung functions. Similarly, in examining the issue of case mix, Gonnella and colleagues [12] concluded, “Disease staging is a severity classification system independent of treatment or use of procedures.” In contrast, Markson and associates [13] point out that other classification systems, such as the diagnosis-related groups, which are used as the basis for Medicare hospital payments, allow the use of procedures in distinguishing patient groups. Some classification systems for severity of illness are diagnosis-specific systems, such as the Computerized Severity Index and Patient Management Categories, whereas other classification systems are diagnosis-independent, such as MedisGroups and Acute Physiology and Chronic Health Evaluation [13]. Still other severity classification systems have been reviewed and critiqued in a number of publications [14-171. This severity literature is quite complex, and its complexity suggests a need for further research aimed at defining the severity of asthma more precisely and simply. The current research addresses this issue by considering measures of asthma duration and pulmonary function combined with eight simple questionnaire scales [18] developed by the University of Alabama at Birmingham (UAB) Comprehensive Asthma Program. Although similar to previous questionnaires, the UAB measures evaluated responses on continuous scales with up to five answers, rather than a mere yes or no dichotomy. The scales included measures of a physician judgment scale assessing severity, incidence of respiratory symptoms, recent history of respiratory disease, the extent to which patients are “bothered” by their asthma, intensity of the medication regimen, incidence of medication side effects, and health care utilization. In previous research [18], standard psychometric and correlational procedures for assessing reliability and construct validity [19,20] suggested that these scales could have value as standard assessments in asthma. In the current study, the statistical technique of factor analysis [21,22] was used to search for the general dimensions of severity underlying these specific measures. Factor analysis is a well-accepted technique in psychologic and sociologic research for identifying the dimensions underlying the correlations among a set of variables. Also, it may have considerable usefulness for clinical research, as the current study is intended to illustrate.

PATIENTS AND METHODS Patients Data were collected from 199 adults receiving outpatient treatment for asthma in the UAB Pul264

September 1992 The American Journal of Medicine

monary Medicine Clinic. The subjects appeared typical for the UAB Medical Center but were not necessarily representative of the overall Alabama or Birmingham populations. Thirty-five percent were male; 66% were white. Thirty-one percent had not completed high school; 18% had graduated from college. Fifteen percent were under 30 years old; 10% were 70 or older. Physicians judged 31% to have mild asthma, 48% to have moderate asthma, and 21% to have severe asthma. Forty-seven percent received a diagnosis of asthma for the first time within the past 10 years; 16% received their first diagnosis 30 or more years ago. Fifty-one percent had visited an emergency room and/or had been hospitalized for a respiratory problem within the past year. Scales The data were collected as part of the baseline assessment for the UAB asthma self-management program [23]. To preclude possible distortions due to varying literacy levels, the data were collected through interviews and review of clinic records. However, the scales were designed to be suitable for written questionnaires and are adaptable to a wide variety of research situations. All data other than physicians’ judgments of severity were collected independently by trained interviewers who were members of the project, so there was no artificial overlap between physician judgments and the other variables. Asthma duration was measured by the number of years since the patient first received a physician’s diagnosis of asthma. Pulmonary function was determined by spirometric measurement. The requirements of factor analysis dictated that only one measure of pulmonary function be included, and forced expiratory capacity in 1 second (FEVl) as a percent of predicted normal was judged the best single measure. To accurately assess the patient’s lung function, the best FEVl value from routine visits during the previous 12 months was used so that values were not reduced due to an acute attack. Therefore, low FEVl values would indicate fixed airflow impairment. Sixteen physicians who treat patients in the UAB Pulmonary Clinic were asked to assess the severity of their patients’ underlying disease as either mild, moderate, or severe, using the written guidelines shown in Table I. The physicians then referred the patients to the Asthma Program. Since the physicians were not directly involved in the study, they were not aware of the various instruments and data collection forms used. This avoided possible bias that could occur if clinicians and researchers were using the same criteria. However, to confirm the interobserver reliability of

Volume 93

their assessmentsof severity, a registered nurse trained for the asthma project reviewed the patients’ charts and in all casescame to the same conclusion regarding severity. The next two variablesmeasuredthe incidenceof various respiratory symptoms and diseasesexperiencedby the subjects.Subjects were asked,“Have you had any (respiratory symptom) in the past 7 days?” The respiratory symptoms that wereexamined included “coughing,” “wheezing,” “shortness of breath,” “ increasedsputum,” “thick sputum,” “green or yellow sputum,” and “decreasedexercise tolerance.” If subjectsanswered“yes” to this question, they were then asked, “Would you describe your (respiratory symptom) as slight, moderate,or severe?”A total scalescorewas computed by summing the scoresfor the individual symptoms. Subjectsnext wereasked,“In the past 12months, haveyou experiencedany (respiratorydisease)that lasted more than 1 day?” If they answered“yes,” they were askedhow many episodeshad occurred. The respiratory diseasesincluded “spells of coughing,” “spells of shortnessof breath,” “colds or upper respiratory infections,” “bronchitis,” and “pneumonia.” Scoresof 0 were assignedto subjectswho did not experiencethe illnessin question,scoresof 1 to subjectswho experiencedoneto threeepisodesof the illness, and scoresof 2 to subjectswho experiencedfour or more episodesor who stated that they suffered from the condition continuously. The scores for the individual illnesses were summed, giving a total symptom scorefor the past 12months. The sixth and seventhvariables were subjective measuresof the extent to which asthma symptoms inconveniencedthe patients and hencewere titled the “bother” scales. Subjects were asked, “How would you rate the severity of your asthma symptoms on the following scaleduring (time period)?“: 5 = severelybothered,unableto function; 4 = considerably bothered,but able to function; 3 = somewhat bothered,severalminor symptoms; 2 = a little bothered,oneor two minor symptoms;1 = not at all bothered, no symptoms. Four time periods pertaining to seasonalvariations included “usually during the spring,” “usually during the summer,” “ usually during the fall,” and “usually during the winter.” Two items pertaining to recentand current episodesincluded “during the past 7 days” and “during the past 24 hours.” Data wereanalyzedin terms of averagescoresfor seasons and for recent and current time periods. The medication regimenwasassessedby indicating the useof five types of medication-an inhaled bronchodilator, continuous theophylline, more than two coursesof steroids in the past year, an inhaled anti-inflammatory agent, and more than 9eptember

TABLE I Physicians’Judgment Scaleto AssessSeverity Mild

Infrequent attacks (< 1 every 2 mo), interim symptomatic treatment

Moderate

More frequent attacks (> 1 every 2 mo), continuous daily treatment

Severe

Continuous symptoms, continuous multiple drug regimens, including some systemic steroids

two coursesof antibiotics in the past year.A scoreof 1 was assignedto eachmedication included in the regimen and a score of 0 to each medication not included.A total scoreevaluatingthe overall intensity of the regimen wascomputed by summing the individual medication scores.Data were obtained through reviews of clinical records and were confirmed or updated by interview questionsabout recent changesin the regimen. The side effects of medication were assessedby asking subjects, “Have you experienced(side effect) from medication prescribedfor asthma within the past 12 months?” The side effects included “pounding heart (tachycardia),”“insomnia,” “nausea,” “bad dreams,” “ white spots in mouth (oral thrush),” “seizures,”and “some other side effect.” Scoresof 0 were assignedto “no” responsesand scoresof 1 to “yes” responses.Total scoreswere computedby summing the scoresfor the individual side effects. Becauseresponsesto theseside-effect scalesmay be influenced both by hypochondriasis and by misattribution of ordinary life events (e.g., bad dreams)to asthmamedication, thesescoresare best interpreted asrelative variation in side effects rather than asthe absolutenumber of such effects. The last variable addressedthe issue of health careutilization. Subjectswereaskedif a respiratory problem had causedthe following eventsduring the pastyear: (1)telephonecall to a physician, (2) office visit to a physician, (3) emergencyroom visit, and (4) hospitalization. The number of contacts were recorded for each event and the scores were summed. Statistical Analysis

The 10variables wereintercorrelated and factor analyzed by the principal-components procedure [21,22].Inspectionof the curveof eigenvalues,using guidelinesspecifiedby Cattell [24], suggestedthat three factorsbe retained.To allow for the possibility of correlationsamongfactors,the first three componentswererotated to a final oblique solution by the Oblimin procedure [21]. Factor scoreson the three rotated components were derived for each subjectandcorrelatedwith demographiccharacter1992

The American

Journal

of Medicine

Volume

93

265

TABLE II CorrelationsAmong Severity Indices (n = lQQ)* 1 -

1. Years since first asthma diagnosis

2

3

4

5

6

-0.11

0.05

-0.03

-0.07

-0.10

-0.01

-0.35*

-0.01

-0.10

-0.04

-0.21*

-0.20* 0.38t

0.23t

0.33t

2. FEVl% of predicted normal

7

a 0.16+

9

10

0.07

-0.11

-0.04

0.00

3. Physician judgment of severity

0.21*

0.28t

0.193

0.30t

4. Respiratory illnesses past 12 mo

-

0.444

0.41t

0.38t

0.17+

0.27*

0.46t

-

0.29t

0.66$

0.05

0.25t

0.24+

6. Bother scales, seasonal total

0.45t

0.10

0.30t

0.17+

7. Bother scales, current total

-

5. Respirators symptoms past 7 d

-0.204 -

8. No. of medications in regimen

-0.04 0.15

0.00 0.23*

9. Medication side effects past 12 mo

0.35*

10. Health care services utilized past year

-

‘Correlationmatrix is symmetric so values aboveand belowdiagonalare identical. p < 0.05. p
TABLE III FactorAnalysis of Severity IndicesWith Oblique Rotation (n = lQQ)* Factor Structure A Symptom intensity

a Airflow Impairment

C Manap ment Intensity

1. Years since first asthma diagnosis 2. FEVl% of predicted normal

-0.05

0.54 -0.75

-0.02 -0.07

3. Respiratory 4. Physician judgment illnesses ofpast severity 12 mo

0.32 0.61 0.82

5. Respiratory symptoms past 7 d 6. Bother scales-seasonal total 7. Bother scales-current total 8. No. of medications in regimen 9. Side effects in past 12 mo 10. Health care services utilized for respiratory problem last 12 mo

-0.17

E

-0106 0.37 0.28

-0.18 0.53

0.00 0.05

:*:i 0:26 0.29

0.14

0.16

0.49 0.02 -0.19

0.62 0.58 0.80

istics and with scores on the Asthma Symptoms Checklist [25,26]. The Asthma Symptoms Checklist assesses the extent to which various symptoms are experienced during attacks, not the extent to which they are experienced on a day-to-day basis.

RESULTS Table II shows the correlations among the 10 severity indices; significant correlations are identified by asterisks. Physicians’ judgment of severity proved to be a rather general measure with significant correlations at the p
September 1992 The American Journal of Medicine

By convention, the results of a factor analysis are presented in two steps. The first step is to report the factor structure identified through application of the statistical procedures. Table III reports the factor structure identified in the current study. The elements in this table are termed factor “loadings.” The loading of a given variable on a given factor can be interpreted, roughly, as the correlation between that variable and the underlying category, or dimension, represented by the factor. Although factor loadings are roughly comparable to Pearson correlations, no procedure is known for computing the standard error of a factor loading. Therefore, loadings are compared with some conventional rules of thumb rather than being tested for statistical significance. One common rule of thumb is to consider loadings of 0.45 or larger to be “high.” This convention was used in the current study, and variables with loadings of 0.45 or higher are in boldface in Table III. More specifically, Table III summarizes the three oblique rotated factors. An orthogonal rotation by the Varimax procedure [21] yielded a highly similar structure, confirming the stability of these factors. Each factor had several variables with loadings of 0.45 or higher, and each variable had at least one such loading, indicating that the factor analysis was successful in identifying several underlying dimensions of severity. Four of the 10 variables had more than one loading of 0.45 or higher, indicating that those variables were factorially complex. The second step in reporting the results of a factor analysis is to examine the variables with high loadings on each factor in order to identify the element that seems to be common to those variables

Volume 93

ASTHMA

and to assign a title to the factor that appears to summarize that common element. The variables with high loadings on Factor A and their associated loadings included the following: 7. Bother scalescurrent total, 0.88; 5. Respiratory symptoms past 7 days, 0.82; 6. Bother scales-seasonal total, 0.64; 4. Respiratory illnesses past 12 months, 0.61. The common element in these variables appears to be the frequency with which patients experience respiratory symptoms and the extent to which they are distressed by those symptoms. Therefore, a descriptive name for Factor A might be Symptom Intensity. The variables with high loadings on Factor B included the following: 2. FEVl% of predicted normal, -0.75; 1. Years since first asthma diagnosis, 0.54; 3. Physician rating of severity, 0.53; 8. Number of medications in regimen, 0.49. The common element in these variables is less immediately obvious. It should be remembered, however, that the FEVi value was the highest value that had been obtained over a l-year period. Therefore, this factor appears to include a permanent, long-term reduction in air flow, and Air Flow Impairment might be an appropriate title. The high loading variables for Factor C included the following: 10. Health care services utilized for a respiratory problem in the last 12 months, 0.80; 8. Number of medications in regimen, 0.62; 3. Physician rating of severity, 0.59; 4. Respiratory illnesses past 12 months, 0.58; 9. Side effects in past 12 months, 0.58. The common element in these variables appears to be the extent to which management of the patient’s asthma requires active treatment. Therefore, Factor C was given the title Management Intensity. Table IV shows the correlations among the three rotated factors, the correlations of the factors with demographic data, and the correlations of the factors with scales of the Asthma Symptoms Checklist [25,26]. Asterisks again identify the statistically significant correlations. These results indicate that the Symptom Intensity and Treatment Intensity factors are related but that the Airflow Impairment factor is independent of these dimensions. Symptom Intensity correlated with being younger, being female, and with the airway obstruction, fatigue, and irritability scales of the Asthma Symptom Checklist. Airflow Impairment correlated positively with being a male, being older, and the presence of other co-morbid conditions, and negatively with symptoms of airflow impairment on the Asthma Symptom Checklist. This negative correlation is plausible because the checklist questions measure reversible symptoms as experienced and reported by patients, whereas the factor from this analysis September

SEVERITY

/ BAILEY ET AL

TABLE IV FactorsCorrelated With Factors,DemographicVariables, and Asthma SymptomChecklist ,. B hdageSymptom Airflow merit Intensity Impairment Intensity A

Factorscorrelatedwith factors Symptom Intensity (n = 199) Airflow Impairment (n = 199) Management Intensity (n = 199)

0.30*

0.00 o>o

0.30*

0.09

059

-

o.zo* 0.06 0.21* -0.04 0.03

-0.06 -0.09* -0.19*

Factorscorrelatedwith demo. graphicvariables Male sex fn = 100) White race fn = 199) Age fn = 199) Education level fn = 188) Smoking status (n = 172) Total co-morbidities present (n = 199)

-0.17*

-0.11 -K’ 0:13

-0.07

0.16+

-0.01 0.08

-0.04

Factorscorrelatedwith Asthma SymptomChecklist Asthm_alSiyptom Checklist n Airwav obstruction Fatigue Irritability Panic-fear Hyperventilation

0.18+

-0.17+

0.04 0.09 0.08

-0.01 -0.04

0.11 0.12

0.15

0.14 0.01

-0.08


involves irreversible Treatment Intensity younger.

reduction in airflow. The factor correlated with being

COMMENTS Asthma is treated by a variety of physicians, including general internists, pediatricians, family practitioners, and other primary care providers. Many questions remain unanswered about the current status of diagnostic accuracy, especially with respect to severity. This factor analysis is an important step in identifying the descriptions that best define severity. Specifically, results from this factor analysis indicate that a description of asthma severity includes at least three components-Symptom Intensity, Airflow Impairment, and Management Intensity. These three factors provided a concise summary of the information measured by the individual variables and clarified the important elements to be considered in diagnosing asthma severity. The physician judgement scale to assess severity significantly correlated with a broad variety of asthma characteristics. Most of the variables measuring asthma characteristics were based on patient selfreports and, therefore, are not unimpeachable. However, the significant correlation between the physician judgement scale and the measure of pulmonary function, FEVl% of predicted normal, pro1992

The American

Journal

of Medlclne

Volume

93

267

vided indirect evidence that the variables were reliably measuring severity. As shown from the factor analysis in Table III, physicians’ judgements were more strongly associated with Airflow Impairment and Management Intensity than with Symptom Intensity. These results were to be expected in view of the instructions to physicians to rate the severity of the underlying disease, rather than the severity of a particular episode. The results are encouraging as they confirm that the physicians, using a simple scale of mild, moderate, or severe, made reliable assessments of the severity of asthma. The correlations between the factors and patient demographic characteristics were interesting but not very surprising. None of the factors was related to race, however, and one might have expected some relationship in view of the higher asthma death rate in the black population [27]. Symptom Intensity and Management Intensity scores were higher in younger patients. Several explanations of this finding could be suggested. It might be that older patients have learned to control their asthma better or have become more resigned to more limitations and, therefore, report fewer symptoms. It also might be that younger patients are more active, which could lead to acute asthma attacks, stimulating more symptoms. Finally, it is possible that health care providers are more likely to consider younger patients to have reversible disease and, therefore, treat them more aggressively with more intense regimens. The correlation of Symptom Intensity with airflow impairment, fatigue, and irritability seems consistent with the construct validity of this factor. The Airflow Impairment factor correlated with age, a finding that may reflect the simple fact that older patients have had asthma for a longer period of time and the long-standing asthma has resulted in fixed obstruction. Certainly, the correlation of Airflow Impairment with an increased number of co-morbid conditions would be expected in an older population. The negative correlation between Airflow Impairment and symptoms of airflow obstruction may again reflect a tendency for patients with more chronic problems to become resigned to their condition and report symptoms of attacks less frequently. Such a pattern might put them at risk for undertreatment. Airflow Impairment also correlated with being male, a finding for which we have no explanation. Numerous gender differences in airway reactivity have been reported in the literature. Most investigations in children show an excess of responsiveness in males over females [28]. Of three studies [29-311 that have reported a sex-specific prevalence of increased airway responsiveness in adults, two reported a slightly higher prevalence in females than in 299

September 1992 The American Journal of Mediclne

males. In the Lung Health Study, which is directed at smokers with mild airflow obstruction, nonspecific airways hyperresponsiveness was noted in a significantly higher percentage of women (85.1%) than men (58.9%) [32]. These differences are poorly understood, and this subject needs to be examined more fully in research that directly addresses the gender issue. In conclusion, the factor analysis and correlations show that these simple, scaled questions, particularly the physician judgement scale, provided a useful summary of asthma severity. Further, the three factors that emerged from the analysis showed that severity of asthma is at least a three-dimensional entity. Thus, when physicians and other medical professionals make judgements about the severity of asthma cases, information that reflects these three dimensions-Symptom Intensity, Airflow Impairment, and Management Intensity-should be considered. We believe that the physician judgement scale and some measures of these three identified factors could easily be included in research protocols and would thus promote meaningful comparisons among asthma studies by providing a standard, brief measure of asthma severity without affecting or limiting other data collection. For example, including four variables, the physician judgement scale, an FEVi value not obtained during an acute attack, a “bother” scale, and a question concerning health care services utilized for respiratory problems during the previous year, might give a more uniform overall assessment of asthma severity. Treatment recommendations, such as those made in the recently published Expert Panel Report [33], must be based on severity levels; simple and well-documented measures of severity are essential. It seems likely that other diseases would also have complex and multidimensional aspects of severity as well, and similar studies analyzing the various components of severity might be useful.

REFERENCES 1. Burney P, Detels R, Peckham C, Samet JM, Tager IR. Recommendations for research in the epidemiology of asthma. Chest 1987; 91: 194-5. 2. Burney P. Chin S. Developing a new questionnaire for measuring the prevalence and distribution of asthma. Chest 1987; 91: 79-83. 3. Ingram RH. Research needs: unsolved questions about the etiology of asthma. Chest 1987; 91: 193-4. 4. Samet JM. Epidemiologic approaches for the identification of asthma. Chest

1987; 91: 74-8. 5. Medical Research Council. Respiratory Medical Research Council, 1979. 6. Ferris BG. Epidemiology standardization

symptoms project.

questionnaire.

London:

Am Rev Respir Dis 1978;

118 (6, pt 2): l-120. 7. Buist AS, learned from Chest 1987; 8. Carrasco

Vollmer WM. Prospective investigations in asthma: what have we longitudinal studies about lunggrowth and senescence in asthma?

91: 119-26. E. Epidemiologic

aspects of asthma

in Latin America.

Chest 1987;

91: 93-7. 9. Crookston

Volume 93

JB. Prevalence

rates of asthma in developing countries

and their

ASTHMA

comparison with those in Europe and Latin America. Chest 1987; 91: 97-103. 10. Dodge R. Burrows B. The prevalence and incidence of asthma and asthmalike symptoms in a general population sample. Am Rev Respir Dis 1980: 122:

567-75.

11. Aas

K. Heterogeneity of asthma. Allergy 1981; 36: 3-14. 12. Gonnella JS, Hornbrook C. Louis 2. Staging of disease: a casemix measurement. JAMA 1984; 251: 637-44. 13. Markson LE, Nash DB, Louis DZ. Gonnella JS. Clinical outcomes management and disease staging. Evaluation and the health professions 1991; 14:

201-27. 14. Cretin S, Worthman LG. Alternative systems for case-mix classification in health care financing. Santa Monica, CA: Rand, 1986. 15. Geehr EC. Selecting a proprietary severity-of-illness system. Tampa, FL: American College of Physician Executives, 1989. 16. Hornbrook MC. Hospital case mix: its definition, measurement and use. Part II. Review of alternative measures. Medical Care Review 1982; 39: 73-123. 17. lezzono LI. Measuring the severity of illness and case mix. In: Goldfield N. Nash DB. editors. Providing quality of care: the challenge to clinicians. Philadelphia: American College of Physicians, 1989: 70-105. 19. Richards JM Jr, Bailey WC, Windsor RA. Martin B. Soong S. Some simple scales for use in asthma research. J Asthma 1988; 25: 363-71. 19. Gulliksen H. Theory of mental tests. New York: Wiley, 1950. 20. Cronbach LJ, Gleser GC. Nanda H, Rajaratnam N. The dependability of

SEVERITY

/ BAILEY ET AL

23. Bailey WC, Richards JM Jr, Manzella BA, Windsor RA. Brooks CM, Soong SJ. Promoting self-management in adults with asthma: an overview of the UAB program. Health Educ Q 1987; 14: 345-55. 24. Cattell RB. The Scree test for the number of factors. Multivariate Behavioral Research 1966; 1: 245-76. 25. Brooks CM. Richards JM Jr, Bailey WC, Martin B, Windsor RA. Soong S. Subjective symptomatology of asthma in an outpatient population. Psychosom Med 1989; 51: 102-8. 25. Kinsman RA, Luparello T. Banion R. Sheldon S. Multidimensional analysis of the subjective symptomology of asthma. Psychosom Med 1989; 35: 25O-fj7. 27. Sly M. Mortality from asthma, 1979-1984. J Allergy Clin lmmunol1988; 82:

705-17. 28. Weiss ST, O’Connor GT, Sparrow D. The role of allergy and airway responsiveness in the natural history of chronic airflow obstruction (CAO). In: Weiss ST, Sparrow D, editors. Airway responsiveness and atopy in the development of chronic lung disease. New York: Raven Press: 1989: 181-240. 29. Bardana EJ, Sobti KL. Cianculli FD, eta/. Aspergilusantibody in patients with cystic fibrosis. Clin Aller 1975: 129: 1164-7. 30. McFarlane H. Allan JD. van der Zeil P. Passive cutaneous anaphylaxis (PCA) and specific IgE in cystic fibrosis and their heterozygotes. Clin Aller 1977; 7:

279-84. 31. Ormerod LP, Thomson RA, Anderson CM, et a/. Reversible airway obstruction in cystic fibrosis. Thorax 1980; 35: 76S-72.

behavioral measurements: theory of generalizability for scores and profiles. New York: Wiley, 1972. 21. Harman HH. Modern factor analysis, 3rd ed. Chicago: University of Chicago

32. Tashkin DP, Alto% MD, Bleeker ER, et al. The lung health study: airway responsiveness to inhaled methacholine in smokers with mild to moderate airilow limitation. Am Rev Respir Dis 1992; 145: 301-10.

Press, 1976. 22. McDonald baum, 1985.

33. Expert Panel on the Management of Asthma, National Asthma Education Program, National Heart, Lung, and Blood Institute. Guidelines for the diagnosis and management of asthma. 1991: 8-13.

RP. Factor analysis and related techniques.

Hillsdale, NJ: L. Erl-

September

1992

The American

Journal

of Medicine

Volume

93

269