Association of Asthma Control with Health Care Utilization and Quality ...

Download PDF
6 downloads 27 Views 118KB Size Report
Comment
the last 12 mo (self-perception of asthma control was asked relative to the last 4 wk ... per Asthma Quality of Life Questionnaire; SGRQ St. George's Re- spiratory ...
Association of Asthma Control with Health Care Utilization and Quality of Life WILLIAM M. VOLLMER, LEONA E. MARKSON, ELIZABETH O’CONNOR, LESLY L. SANOCKI, LESLYE FITTERMAN, MARC BERGER, and A. SONIA BUIST Kaiser Permanente Center for Health Research, Portland, Oregon; Merck & Co., Inc., West Point, Pennsylvania; and Pulmonary and Critical Care Division, School of Medicine, Oregon Health Sciences University, Portland, Oregon

Asthma severity and level of asthma control are two related, but conceptually distinct, concepts that are often confused in the literature. We report on an index of asthma control developed for use in population-based disease management. This index was measured on 5,181 adult members of a large health maintenance organization (HMO), as were various self-reported measures of health care utilization (HCU) and quality of life (QOL). A simple index of number of control problems, ranging from none through four, exhibited marked and highly significant cross-sectional associations with selfreported HCU and with both generic and disease-specific QOL instruments, suggesting that each of the four dimensions of asthma control represented by these problems correlates with clinically significant impairment. Qualitatively similar results were found for control problems assessed relative to the past month and relative to the past year. Asthma control is an important “vital sign” that may be useful both for population-based disease management as well as for the management of individual patients. Vollmer WM, Markson LE, O’Connor E, Sanocki LL, Fitterman L, Berger M, Buist AS. Association of asthma control with health care utilization and quality of life. AM J RESPIR CRIT CARE MED 1999;160:1647–1652.

The last 10 yr have seen a major shift in the way physicians in the United States view and treat asthma. Asthma is now seen not as an acute, bronchospastic disease, but rather as a chronic inflammatory disease of the airways. Coincident with this, increasing emphasis has been placed on preventive disease management and on control of the inflammatory process. Current U.S. guidelines (1) recognize four distinct levels of asthma severity: mild intermittent, mild persistent, moderate persistent, and severe persistent, and separate treatment protocols are recommended for each. Although many attempts have been made to operationalize these and previous definitions of asthma severity, all have their shortcomings and no agreement exists on a common standard. We believe the problem is further complicated by a lack of distinction in the literature between underlying disease severity and current level of asthma control. The former is a reasonably stable characteristic of the individual that may change slowly over time, whereas level of control reflects current func-

(Received in original form February 22, 1999 and in revised form May 25, 1999 ) Supported by a grant from Merck & Company, Inc. For copies of the Asthma Therapy Assessment Questionnaire (ATAQ), please contact: Leona E. Markson, Sc.D., Director, Outcomes Research and Management, Merck & Co., Inc., P.O. Box 4, WP39-164, West Point, PA 19486-0004. Correspondence and requests for reprints should be addressed to William M. Vollmer, Ph.D., Center for Health Research, 3800 N. Interstate Ave., Portland, OR 97227-1110. E-mail: [email protected] Am J Respir Crit Care Med Vol 160. pp 1647–1652, 1999 Internet address: www.atsjournals.org

tioning and may change markedly over relatively short time frames (2), although the objective of management is to maintain a good level of control. While measures of both severity and level of control will inevitably be correlated, both concepts are important and conceptually distinct. For example, although mild asthmatics have disease that is, by definition, relatively easy to control (e.g., through “as needed” use of b-agonists) they will occasionally suffer from acute exacerbations during which their level of control may be very poor. Similarly, individuals with moderate to severe asthma require more intensive pharmacotherapy to control their symptoms, and yet with proper therapy and good compliance can experience good symptom control. From the perspective of the practicing clinician, level of control may be the more relevant measure, because the objective of therapy will be to control symptoms and minimize the impact of the disease on patient functioning (i.e., to achieve a good level of control). From the perspective of population-based disease management, asthma control could also serve as a good barometer of the adequacy of health care being provided to a population, as well as serving as an indicator of patients who may benefit from more aggressive management. The Asthma Therapy Assessment Questionnaire (ATAQ) was developed as a disease management (DM) tool to identify individuals whose asthma management may be suboptimal. This brief, self-administered questionnaire generates a fivelevel measure of asthma control (0 5 no control problems to 4 5 four control problems). In addition, the ATAQ is used to identify possible barriers to good disease management. This study describes the distribution and properties of the ATAQ control score among members of a large health maintenance organization (HMO) in the Pacific Northwest who have asthma.

1648

AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE

METHODS Population The Northwest Division of Kaiser Permanente (KPNW) is a large, federally qualified, group model HMO located in Portland, Oregon. KPNW has approximately 430,000 members, whose demographic and socioeconomic characteristics are similar to those of the area population as a whole (3). This analysis is based on the results of a survey sent to a subset of members, age 18 and older, having two or more “antiasthma” medication dispensings in 1996 and/or a hospital or emergency department visit for asthma in 1994, 1995, or 1996. In addition, all individuals needed to have current health plan coverage as of June 30, 1997. A total of 13,964 members met these criteria. We surveyed this population in two stages (Figure 1). Between August and September, 1997, everyone received a brief, two-page screening questionnaire (ATAQ). In addition, approximately onequarter of these individuals (n 5 3,490) also received the 36-item Short-Form Health Survey (SF-36) health status questionnaire (4) and the long version of Juniper’s Standardized Asthma Quality-ofLife Questionnaire (AQLQ-S) (5). In December 1997, additional questionnaires were mailed to a random subset (n 5 2,000) of the initial respondents. Those individuals who initially received the SF-36 and AQLQ-S were excluded from this follow-up survey. As with the initial survey, the content of the packets sent out with the follow-up survey differed for different subgroups. The 1,000 individuals younger than age 50 yr received the SF36 and the short version of the AQLQ-S (6). The 1,000 individuals age > 50 yr received the SF-36 and the St. George’s Respiratory Symptom Questionnaire (SGRQ) (7). The collective set of questionnaires from these two surveys provides a broad range of outcomes to assess the ATAQ asthma control index. The analyses in this study are limited to the 5,181 participants who reported, in response to the initial survey, that they had doctor-diagnosed asthma and had taken asthma medications within the past 12 mo. The actual number of participants included in any given analysis varies, however, because not all individuals received the same set of questionnaires.

Survey Instruments ATAQ is a self-administered questionnaire designed to assess level of asthma control and to identify possible disease management problems. It contains questions about whether the respondent has ever been told that he or she had asthma and asks about nocturnal awakenings, missed activities, medication use, relationship with medical

VOL 160

1999

provider, attitudes toward current treatment, perception of control, and health care utilization. ATAQ was developed with input from clinical experts and tested in patient focus groups to make sure that patients understood the intent of the questions. To assess level of asthma control, the questionnaire asks about: (1) self-perception of asthma control; (2) missed work, school, or normal daily activities due to asthma; (3) nighttime waking due to asthma symptoms; and (4) use of “quick relief” inhaler medication. Respondents were graded as either having or not having a control problem in each of these dimensions, and the number of control problems was then summed to provide an index ranging from 0 to 4. Three of the four control problems were assessed relative to both the last 4 wk and the last 12 mo (self-perception of asthma control was asked relative to the last 4 wk only), and separate control summary scores were calculated for each time frame. This analysis focuses primarily on the 4-wk control summary score. Copies of the ATAQ instrument and the scoring algorithm for it are available from the authors upon request. With the exception of quick reliever use, each control dimension is assessed using a single question. The two asthma-specific quality-of-life (QOL) questionnaires that we used, the AQLQ-S and the Mini AQLQ-S, were self-administered. Both questionnaires give rise to an overall summary index and four subindices covering the following domains: activity limitation, symptoms, emotional function, and environmental stimuli. Both differ from Juniper’s original AQLQ in that the latter asks a series of questions about limitations in five activities identified as especially important to the respondent (5, 6, 8) whereas the standardized versions ask about limitations in generic categories of activity (e.g., strenuous, light exercise). The questions all refer to the “last two weeks” and use seven Likert-type response options (e.g., seven response options ranging from “all of the time” to “none of the time”). As with the original AQLQ, the AQLQ-S is composed of 32 questions, while the Mini AQLQ-S is composed of only 15 questions. A difference of 0.5 in any of the scales, which all range from a low of 1 to a high of 7, has been proposed to reflect a clinically meaningful effect (9). The SF-36 is a self-administered, generic, health status instrument containing 36 equally weighted questions covering eight health status domains: physical functioning, role limitations due to physical health, pain, general health, role limitations due to emotional problems, energy/fatigue, emotional well-being, and social functioning (10). The first four of these domains can be grouped into an overall physical health score; the latter four can be combined into an overall mental health score. Each of these two overall scores ranges from 0 to 100, with larger scores indicating better health status. The questionnaire was used with the permission of the RAND Corporation. The SGRQ was designed primarily for use in patients with chronic obstructive pulmonary disease (COPD), although the questions are all asked relative to the patient’s chest problems. It measures three domains of functioning: symptoms, activity, and impacts, and also provides an overall summary score that combines all three domains. Each SGRQ scale ranges from 0 to 100. In contrast to the SF-36 and Juniper scales, however, a lower score indicates better health status while a higher score indicates poorer health status. A difference of 4 on the overall summary score is believed to reflect a clinically significant effect (11).

Statistical Methods

Figure 1. Summary of survey design. KPNW 5 Kaiser Permanente, Northwest Division; ATAQ 5 Merck’s Asthma Therapy Assessment Questionnaire; SF-36 5 36-item Short-Form Health Survey health status questionnaire; AQLQ-S 5 standardized version of the Juniper Asthma Quality of Life Questionnaire; SGRQ 5 St. George’s Respiratory Questionnaire.

We used the Pearson chi-square statistic and analysis of variance to compare proportions and scaled data (e.g., QOL indices), respectively, across the different levels of the control index. All analyses were performed using the SAS statistical software package (SAS Institute, Cary, NC). Unless otherwise stated, all p values are two-sided and the term “significant” implies a p value , 0.05. To determine the relative importance of the four control dimensions, we also fit a series of (reverse) stepwise regression models to predict the health care utilization and QOL outcomes from the four component items that comprise the control index. General linear model regressions were run for the QOL scores and logistic regressions were run for the HCU outcomes (12). In all cases, the four dummy variables were entered as predictors, representing: whether the participant felt that his or her asthma was not controlled, whether the participant had missed normal activities, whether the participant

1649

Vollmer, Markson, O’Connor, et al.: Asthma Control, QOL, and HCU had been woken up by his or her asthma, and whether the participant was overusing his or her reliever medications. Most analyses were run twice: once assessing asthma control over the past 4 wk, and once assessing asthma control over the past year. Results of the two analyses were very similar, and only those based on the 4-wk control score are reported in detail.

RESULTS Sixty-two percent of individuals responded to the initial survey. This figure varied markedly by age, however, ranging from 34% for those age 18 to 25 to 78% for those older than age 65. The overall response rate to the second survey was 65%. This also varied by age; 53% for the younger cohort and 77% for the older cohort. Table 1 presents data on the age–gender distribution of the 5,181 respondents to the initial survey who indicated that they had asthma and also reported taking asthma medications within the past year. Self-reported information on coexisting COPD, defined as having ever been diagnosed as having chronic bronchitis, emphysema, or COPD, is also shown. Overall, 67% of sample participants were women, although this declined slightly with age. One-third of study participants also reported having COPD. As expected, the prevalence of self-reported COPD increased markedly with age. Figure 2 shows the distribution of number of control problems in the past 4 wk for both males and females. Fifty-two percent reported at least one control problem, and 13% reported problems in at least three of the dimensions of control. Men and women did not differ significantly on the number of control problems they reported, nor did the distribution of control problems vary by age. Figure 3 shows the proportion of respondents who reported having each of the control problems in the past month. Forty percent reported night waking, 30% poor perceived control, 22% missed activities, and 8% reliever overuse. When looked at over the past year, night waking increased to 66%, missed activities to 47%, and reliever overuse to 15%. Poor perceived control was only assessed relative to the past month. Association with Health Care Utilization

Table 2 shows the association between number of control problems in the past 4 wk and three different measures of selfreported acute HCU for asthma in the past year: doctor visits, emergency department care, and hospitalizations. The proportion of individuals reporting each measure increased significantly as the number of control problems increased. For example, 2% of those with zero control problems in the past month reported having been hospitalized for asthma in the past year, versus 24% for those with four control problems (p , 0.001). Similarly, the proportion reporting two or more outpatient visits for acute attacks increased from 19% to 72% (p , 0.001). Results were comparable when we looked at the number of control problems reported over the last year. Not surprisingly, however, those reporting none or one control problem in the

Figure 2. Number of asthma control problems in the past 4 wk by gender, based on 1,696 men and 3,485 women reporting doctordiagnosed asthma and taking medications for asthma within the past year.

past 4 wk had higher HCU than those reporting none or one control problem in the past year—many of those with no difficulties in the past 4 wk probably had some control problems at other times in the past year, and the attendant HCU. More detailed analyses indicated that those with control problems in the past year, but not the past month, exhibited patterns of HCU that were intermediate to those seen for people with recent control problems and those seen for people with no control problems in the past year. Association with QOL

Table 3 shows the association between number of control problems in the past 4 wk and general health status (SF-36), asthma-specific QOL (AQLQ-S), and QOL as measured by the SGRQ. The SF-36 data are based on all subjects who received the questionnaires, whether as part of the first or the second survey. The AQLQ-S data are based only on the long version administered in the first survey, although similar results were seen for the mini AQLQ-S administered in the sec-

TABLE 1 CHARACTERISTICS OF SAMPLE Age

n Female, % COPD,* %

18–45

46–65

661

Total

1,869 71 21

2,071 70 38

1,241 58 52

5,181 67 35

* Self-report of doctor-diagnosed chronic bronchitis, emphysema, or COPD.

Figure 3. Frequency of individual asthma control problems in the past 4 wk, based on 5,181 respondents reporting doctor-diagnosed asthma and taking medications for asthma within the past year.

1650

AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE

VOL 160

1999

TABLE 2 DISTRIBUTION OF SELF-REPORTED HCU BY NUMBER OF CONTROL PROBLEMS Number of Control Problems in Past 4 wk None (n 5 2,499)

One (n 5 1,216)

Two (n 5 795)

Three (n 5 520)

Four (n 5 142)

p Value†

19%

26%

39%

56%

72%

, 0.001

15%

22%

30%

41%

58%

, 0.001

2%

5%

8%

13%

24%

, 0.001

> 2 visits to doctor or medical provider for worsening asthma in past year > 1 emergency department or urgent care visit for asthma in past year > 1 inpatient stay for asthma in past year

* Source: ATAQ. † Two-sided p value based on Pearson chi-square test wtih four degrees of freedom.

ond survey. The SGRQ data are based on the 707 individuals age 50 and older who completed this as part of the second survey. For both the SF-36 and the AQLQ-S, higher numbers represent better health status/QOL, whereas the reverse is true for the SGRQ. Each of the QOL indices is highly and significantly associated with the control index. The associations between level of control, HCU, and QOL were qualitatively similar for both participants younger than 50 yr of age and age > 50 yr, and formal tests of statistical interaction between age and level of control were generally not statistically significant. Relative Influence of Four Control Constructs

To assess the relative influence of the four dimensions of our control index on these results, we fit reverse stepwise regression models to each of the 14 outcomes summarized in Tables 2 and 3, starting in each case with a model that included dummy indicator variables for each of the four control dimensions as assessed relative to the last month. The indicator of “missed normal activity” was retained in all 14 models and had the largest parameter estimate of all retained variables in 11 of the models (including all of the SF-36, SGRQ, and HCU models) (Table 4). “Poor perceived control” and “night waking” were the next most influential. The dimension indicating reliever medicine overuse was dropped out of eight of the models and had the smallest parameter estimate in three others.

TABLE 3 MEAN QOL SCORES BY NUMBER OF ASTHMA CONTROL PROBLEMS* Number of Control Problems in Past 4 wk

SF-36 Scales Physical Mental Juniper scales Overall Activity Symptom Emotional Environmental SGRQ Overall Symptom Activity Impact

None

One

Two

Three

Four

(1,045)‡ 66 71 (459) 5.7 5.8 5.8 5.7 5.3 (347) 30 42 41 19

(556) 58 65 (250) 4.9 5.2 4.7 4.8 4.4 (175) 40 51 53 29

(359) 50 59 (170) 4.3 4.6 4.1 4.2 3.9 (109) 46 59 57 36

(206) 36 47 (103) 3.1 3.6 2.9 2.9 3.1 (56) 56 62 68 46

(66) 35 45 (38) 2.9 3.3 2.7 2.5 3.0 (19) 63 68 74 54

p Value† , 0.001 , 0.001 , 0.001 , 0.001 , 0.001 , 0.001 , 0.001 , 0.001 , 0.001 , 0.001 , 0.001

* Source: ATAQ. † Two-sided p values based on one-way analysis of variance. ‡ Approximate sample sizes shown in parentheses. Actual numbers vary due to missing values.

For the 1-yr control indicators, all four control dimensions were retained in all but three models. Missed activities again had the largest parameter estimates for the SF-36, SGRQ, and HCU models, poor perceived control (still measured relative to the last month) was most predictive for the Juniper scales.

DISCUSSION We have demonstrated that a relatively simple index of asthma control correlates strongly with both generic and disease-specific indices of QOL, as well as with self-reported HCU in the previous year, in a large population of adult members of a HMO. The results held equally for individuals age 18 to 49 yr and for those > 50 yr of age. Among the four dimensions of control making up the index, missed activities was most strongly related to both HCU and QOL outcomes. Although a number of indices of asthma control, or of various dimensions of control, exist (13–18), the concept of level of control is often confused with underlying severity of disease (2). Severity represents a relatively stable characteristic of the individual that reflects the underlying pathophysiology of the disease. Level of control, by contrast, refers to current asthma status (e.g., symptoms, activity restriction), which is capable of varying markedly over very short time frames; it is thus not a fundamental characteristic of the individual. In clinical trials or outcomes research, for example, level of control may be viewed as a logical outcome variable, reflecting the outcome of treatment, whereas severity is typically viewed as a potential confounding factor for which adjustment must be made to provide valid inference. Unfortunately, no gold standard exists for measuring severity. In an ideal world, everyone is optimally managed and fully compliant, and severity can be unambiguously assessed by the resulting medication regimen (1–2). In the real world, life is not so simple. As the 1997 National Asthma Expert Panel report notes, severity in the less-than-optimally managed patient will tend to correlate with numerous dimensions of control, including current symptoms, activity limitation, and rescue medication use, in addition to clinical parameters such as methacholine reactivity and responsiveness to a bronchodilator (1). The problem is that, if severity is measured using many of these latter factors, it becomes inextricably confounded with level of control. By contrast, level of control is the net result of severity, medical management, and patient adherence. In theory, it is a directly measurable phenomenon, although no consensus exists as to which, if any, of the existing measures of control are “optimal.” Part of the reason for this lack of consensus is the multidimensional nature of “control.” While several symptom

1651

Vollmer, Markson, O’Connor, et al.: Asthma Control, QOL, and HCU TABLE 4 SUMMARY OF BEST FITTING REGRESSION MODELS FOR INDIVIDUAL DIMENSIONS OF ASTHMA CONTROL AS MEASURED IN THE PAST MONTH*

Linear regression coefficients SF-36 physical health composite† SF-36 mental health composite† Juniper summary scale‡ Juniper activity limitation scale‡ Juniper symptom scale‡ Juniper emotional function scale‡ Juniper environmental exposure scale‡ St. George’s symptom scale§ St. George’s activity scale§ St. George’s impact scale§ St. George’s total scale§ Logistic regression coefficientsi > 2 doctor visits in past year for asthma > 1 Emergency department/urgency care visit in past year for asthma > 1 inpatient stay in past year for asthma

Poor Perceived Control

Missed Activity

Night Waking

Reliever Overuse

27.6 (1.3) 26.8 (1.2) 20.9 (0.1) 20.7 (0.1) 21.0 (0.1) 21.1 (0.1) 20.7 (0.1) 5.5 (2.2) 7.5 (2.4) 8.7 (1.8) 7.6 (1.8)

220.6 (1.4) 216.1 (1.3) 20.9 (0.1) 21.1 (0.1) 20.7 (0.1) 20.9 (0.1) 20.9 (0.1) 9.1 (2.4) 21.5 (2.7) 15.0 (2.1) 15.7 (2.1)

25.1 (1.2) 22.9 (1.1) 20.8 (0.1) 20.5 (0.1) 21.1 (0.1) 20.8 (0.1) 20.7 (0.1) 8.8 (2.0) — 6.6 (1.7) 6.0 (1.7)

— — 20.4 (0.1) — 20.5 (0.1) 20.7 (0.2) — — — — —

0.4 (0.1)

0.9 (0.1)

0.4 (0.1)

0.5 (0.1)

0.3 (0.1) —

0.8 (0.1) 1.1 (0.2)

0.4 (0.1) 0.5 (0.2)

0.5 (0.1) 0.9 (0.2)

* Data expressed as mean (SE) of coefficients from best fitting model based on stepwise regressions. † SF-36 scales range from 0 (poorest health status) to 100 (best health status). ‡ Juniper scales range from 1 (worst quality of life) to 7 (best quality of life). § St. George’s scales range from 0 (best quality of life) to 100 (worst quality of life). i Logistic regression coefficients have the interpretation of ln (odds ratios) for given HCU outcome associated with control problem.

scales have been proposed, for example, most tap only one dimension of control (16). The ATAQ control index represents an attempt to capture the multidimensional nature of control. It was designed to identify levels of suboptimal asthma control that may warrant a reexamination of a patient’s treatment program. Each of the four control dimensions assessed by the ATAQ identifies a situation that should trigger patients and clinicians to more closely examine the patient’s health state. These “red flags” include patient perception that the asthma is not well-controlled, missed activities due to asthma, nocturnal awakening due to asthma, and high use of “quick relief” medication. Because the control index was designed primarily for populationbased assessments of asthma management, objective measures of lung function were not included in the index. Measures of resource use (e.g., ambulatory visit rates) or type of asthma regimen (e.g., on two controller agents) were purposely excluded from the control index, because patients may need provider visits or complex regimens to maintain good control. The independent contribution of the four dimensions of the ATAQ control index is evident in the consistent and marked gradation in HCU patterns and QOL scores associated with each increase in the number of control problems. The clinical relevance of the HCU differences (Table 2) is selfevident. For the QOL scores (Table 3), which tap a patient’s sense of well-being and function (10), these associations also appear to represent what have been proposed to be clinically important differences. For the AQOL-S scales, for example, the drops in QOL associated with each succeeding control problem were typically greater than or equal to 0.5, which Juniper equates to a clinically significant change from baseline when measuring a group of asthmatics who report to have changed (improved or worsened) in their global evaluation of their asthma (19, 20). Similarly, the changes in the overall SGRQ score associated with each succeeding control problem were also greater than Jones’ stated clinically significant difference of four (11). Admittedly, assessing the clinical significance of QOL measures, or changes in such measures, is con-

troversial, and as with disease severity no consensus exists on how to go about it (21–23). Nevertheless, the multivariate analyses looking at the separate dimensions of the control index, particularly those based on experience in the last year, further support the independent contribution of these control dimensions. Despite the encouraging results of this study, it is important to recognize the limitations inherent in this analysis. First, the study population, though large, may not be representative of the general population of adults with asthma. The response rate to the first survey was approximately 62%, and to the second survey approximately 65%, and in both cases responses were much higher among older members of the health plan. This in part helps to explain the high percentage of respondents who also reported doctor-diagnosed chronic bronchitis, emphysema, or COPD. Nonetheless we did attempt to survey the full spectrum of the population with asthma, from the very mild to the very severe, in an effort to enhance the generalizability of our findings. Second, our data on HCU is based on self-report, and so the associations of our control index with HCU may be influenced by recall bias in that those with worse control may be more likely to recall their HCU history. Unfortunately we were not able to independently validate this information. Previous research has shown that recall of HCU data is fairly reliable for hospital care and visits to respiratory specialists, but may be unreliable for other types of HCU (24, 25). However although the true magnitude of the associations with HCU may have been overstated, the magnitude of the observed associations is such that we feel confident that our bottom line message is accurate. Ultimately, the clinical significance of the ATAQ control index will be determined by whether or not it is useful for clinical decision-making or population-based disease management, and in particular whether it is predictive of future HCU. Third, our measure of control is not independent of severity. While to some extent this derives from the limitations of our measures (for instance, the National Asthma Education

1652

AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE

and Prevention Program [NAEP] Expert Panel Report [1] defines severity in terms of one’s level of control in the absence of medication), at another level we expect and hope for these measures to be correlated. More severe patients should tend to have worse control! The real challenge, toward which this is a first step, is to develop simple-to-use, standardized tools that allow us, to the greatest extent possible, to independently assess both severity and level of control. In part, we can accomplish this by using different metrics, but also we can accomplish this by using different time frames for asking our questions. For instance, although the developers of the questionnaire originally intended the control index to focus just on the past month, we added the parallel questions relative to the past year specifically for this study in order to determine what additional information was gained by doing this. Our findings suggest that, whereas we achieved qualitatively similar results when assessing control either relative to the past month or to the past year, those patients reporting control problems in the past year, but not the past month, had a profile that was distinct from those with recent control problems. We believe that the shorter one’s time frame, the more likely it is that you tap a concept, current level of control, that is distinct from underlying severity of disease. A fourth potential limitation of this study is that some of the dimensions of our control index are closely related to the QOL indices we used to help validate them. Although it is true that some of the ATAQ questions are similar to some of the questions in the various QOL scales, this does not explain either the magnitude of the associations with any specific scale or the persistence of the association across the various QOL scales. It also is of interest that each of the four dimensions of control appeared to be related to QOL, though admittedly in the multivariate analyses the missed activity and poor perceived control were the most significant predictors. In addition, the control index reported here is a much simpler and easy-to-compute measure than are the QOL measures we have compared them to. Thus, it has the advantage of being a simple measure that nonetheless has profound discriminative properties. In summary, our findings suggest that some standardized measure of asthma control, whether it be this one or another similarly validated measure, might play a useful role in population based disease management programs, especially when linked with data on medication use and/or HCU. Such a measure, if assessed via a survey to a random sample of patients with asthma, could provide a useful index of the health of a population, and more particularly of the success of one’s population based disease management efforts. These and similar outcomes can also be used, for example, to compare outcomes across administrative units within a single managed care organization, or can be used by purchasers to compare patient outcomes for asthma across health plans. Ideally an attempt should be made to adjust any such analysis for severity case mix, although as noted previously this may be difficult. Another potential role for a control index is in the management of individual patients. Level of control could be routinely assessed at each visit as another “vital sign,” together with blood pressure and peak flow for example, and used to monitor patient progress over time. Acknowledgment : The authors are indebted to Pam Algatt-Bergstrom, a collaborator on the development of the Asthma Therapy Assessment Questionnaire.

VOL 160

1999

References 1. National Asthma Education and Prevention Program, National Heart, Lung, and Blood Institute, and National Institute of Health Guidelines for the Diagnosis and Management of Asthma. 1997. Asthma Expert Panel Report 2. National Institutes of Health, Bethesda, MD. NIH 1:97–4051. 2. Cockcroft, D. W., and V. A. Sywstun 1996. Asthma control vs. asthma severity. J. Allergy Clin. Immunol. 98(6, Pt. 1):1016–1018. 3. Greenlick, M. R., D. K. Freeborn, and C. R. Pope. 1988. Health Care Research in an HMO: Two Decades of Discovery. John Hopkins University Press, Baltimore. 10–16. 4. Ware, J. E., and C. D. Sherburne. 1992. The MOS 36-item Short-Form Health Survey (SF-36): conceptual framework and item selection. Med. Care 30:473–483. 5. Juniper, E. F., A. S. Buist, F. M. Cox, P. J. Ferrie, and D. R. King. 1999. Validation of a standardized version of the Asthma Quality of Life Questionnaire. Chest (In press) 6. Juniper, E. F., G. M. Guyatt, F. M. Cox, P. J. Ferrie, and D. R. King. 1999. Development and validation of the Mini Asthma Quality of Life Questionnaire. Eur. Respir. J. (In press) 7. Jones, P. W., F. H. Quirk, C. M. Baveystock, and P. Littlejohns. 1992. A self-complete measure of health status for chronic airflow limitation: the St. George’s Respiratory Questionnaire. Am. Rev. Respir. Dis. 145: 1321–1327. 8. Juniper, E. F., G. H. Guyatt, P. J. Ferrie, and L. E. Griffith. 1993. Measuring quality of life in asthma. Am. Rev. Respir. Dis. 147:832–838. 9. Juniper, E. F., G. H. Guyatt, A. Willan, and L. E. Griffith. 1994. Determining a minimal important change in a disease-specific quality of life questionnaire. J. Clin. Epidemiol. 47:81–87. 10. RAND. 1992. RAND 36-item health survey. 1.0. RAND Health Sciences Program, Santa Monica, CA. 11. Jones, P. W, F. H. Quirk, and C. M. Baveystock. 1991. The St. George’s Respiratory Questionnaire. Respir. Med. 85(Suppl. B):25–31. 12. McCullagh, P., and J. A. Nelder. 1983. Generalized Linear Models. Chapman and Hall, New York. 13. Apter, A. J., R. L. ZuWallack, and J. Clive. 1994. Common measures of asthma severity lack association for describing its clinical course. J. Allergy Clin. Immunol. 94:732–737. 14. Chan-Yeung, M, J. Manfreda, H. Dimich-Ward, J. Lam, A. Ferguson, P. Warren, E. Simons, I. Broder, M. Chapman, T. A. E. Platts-Mills, and A. Becker. 1995. Mite and cat allergen levels in homes and the severity of asthma. Am. J. Respir. Crit. Care Med. 152:1805–1811. 15. Katz, P. P., E. H. Yelin, S. Smith, and P. D. Blanc. 1997. Perceived control of asthma: development and validation of a questionnaire. Am. J. Respir. Crit. Care Med. 155:577–582. 16. O’Connor, G. T., and S. T. Weiss. 1994. Clinical and symptom measures. Am. J. Respir. Crit. Care Med. 149(2, Pt. 2):S21–S28. 17. Wahlgren. D. R., M. F. Hovell, G. E. Matt, S. B. Meltzer, J. M. Zakarian, and E. O. Meltzer. 1997. Toward a simplified measure of asthma severity for applied research. J. Asthma 34:291–303. 18. Juniper, E. F., P. M. O’Byrne, G. H. Guyatt, P. J. Ferrie, and D. R. King. 1999. Development and validation of a questionnaire to measure asthma control. Eur. Respir. J. (In press) 19. Juniper, E. F., G. H. Guyatt, R. S. Epstein, P. J. Ferrie, R. Jaeschke, and T. K. Hiller. 1992. Evaluation of impairment of health-related quality of life in asthma: development of a questionnaire for use in clinical trials. Thorax 47:76–83. 20. Juniper, E. F., G. H. Guyatt, D. H. Feeny, P. J. Ferrie, L. E. Griffith, and M. Townsend. 1996. Measuring quality of life in children with asthma. Quality Life Res. 5:35–46. 21. Guyatt, G. H., E. F. Juniper, S. D. Walter, L. E. Griffith, and R. S. Goldstein. 1998. Interpreting treatment effects in randomized trials. B.M.J. 316:690–693. 22. Norman, G. R, P. Stratford, and G. Regehr. 1997. Methodological problems in the retrospective computation of responsiveness to change: the lesson of Cronbach. J. Clin. Epidemiol. 50:869–879. 23. Wright, J. G. 1996.The minimal important difference: who’s to say what is important? J. Clin. Epidemiol. 49:1221–1222. 24. Vollmer W. M., M. L. Osborne, and A. S. Buist. 1994. Uses and limitations of mortality and health care utilization statistics in asthma research. Am. J. Respir. Crit. Care Med. 149:S79–S87. 25. Ungar, W. J., P. C. Coyte, and the Pharmacy Medication Monitoring Program Advisory Board. 1998. J. Clin. Epidemiol. 51:1335–1342.