| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 St. George's, University of London, London, United Kingdom; and 2 AstraZeneca R&D, Charnwood, Loughborough, United Kingdom
Correspondence and requests for reprints should be addressed to Paul Jones, Ph.D., F.R.C.P., St. George's, University of London, Cranmer Terrace, London, SW17 0RE, UK. E-mail: pjones{at}sgul.ac.uk
ABSTRACT
Exacerbations of chronic obstructive pulmonary disease (COPD) offer a considerable clinical challenge for clinical practice and drug development. The underlying pathobiology of these characteristic events is unclear. We are far behind other disease areas, such as cardiology, where there are effective approaches for diagnosing and managing acute and potentially fatal events. A joint initiative has begun between the pharmaceutical industry, academia, and the Food and Drug Administration to produce a valid diary card measure of exacerbation severity: the EXAcerbations of Chronic pulmonary disease Test–Patient-Reported Outcome (EXACT-PRO). This article describes the background to this initiative, using the consensus definition of a COPD exacerbation as the starting point from which to propose a new method for quantifying the severity of the event. This approach takes into account the relationships between the symptoms that make up an exacerbation and shows that they can be used in a single continuous severity scale. Methods for defining exacerbation onset and cessation are also proposed.
Key Words: exacerbations • diary card • symptoms • patient-reported outcomes
Quantification of the severity exacerbations in chronic obstructive pulmonary disease (COPD) is important in medical management when making choices in treatment (1–3). It is also especially important in determining whether a new therapy is effective in limiting the consequences of exacerbations. Such quantification is usually made clinically and is based on the symptoms, physical signs, and response to therapy. There is a growing interest in the development of a biomarker, equivalent to the use of troponin estimations for the diagnosis of myocardial infarction (4). This remains an area of active research, however, and is yet to provide a validated measure.
At present, there is no method for quantifying the severity of exacerbations. As a result, we have limited ability to accurately gauge the effectiveness of new developments in medical care or therapy. In drug development, the current convention is to estimate the rate of exacerbations, usually measured over at least 1 year. This requires a clear definition of what constitutes an exacerbation. As will be discussed later, this is currently based on a change in symptoms that is judged by a clinician to warrant an increase in therapy—usually an increase in short-acting bronchodilator, addition of antibiotics and/or corticosteroids, or hospital admission. Although this can record the intensity of treatment, it tells us little about the severity of the exacerbation, or how the event has impacted on the patient. This is an important shortcoming, because there is good evidence that COPD exacerbations can have a large and sustained impact on patient's symptoms (5) and health status (6), even in the absence of hospital admission.
The magnitude of the change in symptoms and their duration have been captured in semiquantitative terms, most notably in Seemungal and colleagues' studies of exacerbations (Figure 1) (5), and have led to important insights. In particular, they have shown that measurement of these variables should enable us to better define and measure exacerbations; however, there is a need for an instrument with better-characterized measurement properties. This article introduces an initiative designed to meet this need and will consider how best to combine the distinct clinical elements of exacerbations to provide a single overall measurement of severity.
|
A joint initiative led by United BioSource Corporation (UBC) has undertaken the EXACT-PRO (EXAcerbations of Chronic pulmonary disease Test–Patient-Reported Outcome) project to develop a patient-reported outcome (PRO) measure to evaluate the effects of treatment on acute exacerbations of chronic bronchitis and COPD. This is a multisponsor initiative, with most major pharmaceutical companies engaged, and chaired by Paul Jones and Sanjay Sethi. The development team is made up of clinical experts, UBC staff, and representatives from the U.S. Food and Drug Administration (FDA). The goal is to develop a single measure for use in clinical trials to evaluate outcomes of therapy. Because exacerbations are symptom-related patient experiences and the decision to seek treatment is patient based, the instrument will be a PRO. The objective is to develop an instrument that uses a small number of daily screening questions coupled with an intelligent system that enables a switch to a larger number of questions when a threshold has been crossed. For this reason, the ideal platform for the instrument will be an electronic diary, although the feasibility of a paper-based version will be explored.
The work is prospective and sets out to define the validity, sensitivity, and specificity of potential symptoms for identifying and quantifying the change that occurs with an exacerbation. The final instrument will be able to detect and measure the severity of exacerbations and identify exacerbation characteristics that are associated with patients seeking medical attention, physician's choice of therapy or hospital admission, as well as duration of recovery. This article describes the background to this initiative.
DEFINITION OF A COPD EXACERBATION
There has been much discussion concerning the definition of a COPD exacerbation, and a consensus definition has evolved that brings together patients' experience and years of clinical scientific work (1, 7–10). It includes the following description: "A sustained worsening of the patient's condition, from the stable state and beyond normal day-to-day variations, that is acute in onset and necessitates a change in regular medication in a patient with underlying COPD" (11). This has influenced the event-driven definitions of exacerbations that have been used extensively to determine the frequency rate of exacerbations in clinical trials (12, 13). In such studies, a moderate exacerbation is defined as an increase in symptoms that requires treatment with antibiotics and/or corticosteroids and a severe exacerbation is one that requires hospitalization. These definitions have become validated by consistently discriminating between effective and ineffective therapies in clinical trials; however. the criteria used by clinicians when making such treatment decisions are unknown. There is a need to develop, from this clinical descriptive definition, an operational tool that allows objective (as opposed to behavioral) quantification of severity.
ADAPTATIONS TO THE DEFINITION TO ALLOW QUANTIFICATION
There are several adaptations that will aid this process. The wording "patient's condition" can be translated to patient-reported symptoms. Clinicians use symptoms to gauge severity and to provide a diagnosis. In clinical practice, signs are also important, but they are difficult to standardize for scientific study and it is not possible for the patients to elicit them themselves at home, apart from respiratory rate. Regulatory authorities for drug approval, such as the FDA, are increasingly concerned to capture the patients' perception of their clinical status, and extensive work has been done to establish the criteria for developing and using such PROs (14). PROs can assess any aspect of a patient's health status that comes directly from the patient (i.e., without the interpretation of the patient's responses by a physician or anyone else) and can be used to measure the impact of an intervention on one or more aspects of patient's health status. In COPD, they could quantify symptoms recorded during an exacerbation, whether or not the patient seeks medical attention.
The next term to consider within the consensus definition is a sustained "worsening" of the patient's symptoms. We must consider whether this represents worsening of existing chronic symptoms or whether the exacerbation is characterized by the emergence of a new symptom emerging as the acute state develops—for example, sputum purulence. There is good evidence that an exacerbation is associated with worsening of chronic symptoms. For example, in an analysis of event-based exacerbations, taken from a trial of Symbicort (AstraZeneca, London, United Kingdom) in COPD (15), breathlessness, cough, chest tightness, and nighttime waking during exacerbations increased with the event to a peak, then returned toward the baseline state (Figure 2). Leidy and colleagues have also reported a similar picture (16). Each symptom has a different stable baseline from which it rises as the exacerbation developed. This argues in favor of the exacerbation being a worsening of an existing disease state.
|
It is probable that there are multiple components to an exacerbation, but the changes in pattern of inflammation produce physiologic consequences that are manifested as a worsening of the respiratory symptoms that are present in the basal COPD state: for example, cough, sputum production, breathlessness, and exercise limitation.
IDENTIFYING ITEMS SUITABLE FOR INCLUSION IN AN EXACERBATION PRO
There is longstanding understanding of the clinical characteristics of exacerbations, and the core features have been incorporated into patient self-report diaries (16, 19, 20). However, the FDA's guidance for the development of PROs shows that the items should be derived from the patients themselves. This is usually done through qualitative research in the form of patient interviews and focus groups. The items that are identified by this process then need to be tested for respondent bias (e.g., sex, race, and age) and general understandability—a process known as "cognitive debriefing." At that stage, the items have validity in that they have been reported by patients and are free from obvious demographic biases, but that does not mean that they are suitable for use in a measurement instrument.
MEASUREMENT OF OVERALL EXACERBATION SEVERITY
Although it is clear that specific symptoms rise with an exacerbation and fall with recovery, for most purposes it is necessary to measure overall exacerbation severity. To do this, the severity of each exacerbation symptom must be aggregated to produce an overall severity score. To ensure that this estimate is robust, the PRO must have a number of important measurement properties: (1) it must truly measure exacerbation severity; (2) it should be reliable and perform the same way in every patient and every setting; (3) it should have "interval scaling properties," which means that any given distance between two points along its scale should indicate the same difference in severity, regardless of where the points lie. In other words, the PRO should perform like a ruler. Indeed, the items in a symptom scale are analogous to graduations along a ruler (Figure 3). To ensure that the new PRO has the properties just described, all of its items must, when combined, form a unidimensional construct—in this case, overall exacerbation severity. It cannot be assumed that this fundamental criterion will be met if plausible items are combined without rigorous testing.
|
RELIABILITY OF A PRO
Reliability is an important property of a PRO. A reliable questionnaire is one whose measurement characteristics do not change between patients, between settings or between severity states. The issue of whether an exacerbation is a quantitative or qualitative change from stable to acute state becomes important here. A change in the type (as opposed to the severity) of the state may lead to the emergence of different symptoms, as discussed in the case of sputum purulence. Or the change may be characterized by a change in the severity of one symptom relative to another. If this occurs, the measurement properties of the PRO will alter between states (effectively, the intervals on the "ruler" will change).
Rasch modeling provides a very elegant method of testing for this. It can analyze whether an item behaves in the same way under different conditions. This depends on an important aspect of the methodology—the item characteristic curve (ICC)—which shows the goodness of fit of an item to the model. A perfect fit is characterized by patients' responses to the item forming an S-shaped curve (Figure 4). If an item behaves the same way in different states, the ICC will remain the same, but if its measurement characteristics change, the ICC will change (Figure 4). We have demonstrated recently that six of the items in the seven-item Acute Exacerbation of Chronic Bronchitis Symptom Scale (AECB-SS)—a diary card designed to quantify exacerbations—remained stable between stable and acute states. By contrast, the ICC for the item concerned with sputum color changed between states (21). This suggests that, although sputum purulence may be a marker for the change between stable and acute states, it is unlikely to be a reliable item for quantifying exacerbation severity.
|
As already described, the best measurement properties of a questionnaire are obtained with a questionnaire that has unidimensional properties, but this raises an important question: Is an exacerbation unidimensional or does it have different aspects? Terminology becomes important here. In the context of Rasch modeling, "unidimensional" means a set of items that change in the same way in response to changes in severity. But that does not mean that all the items must measure the same aspect of the exacerbation. There is evidence that there are at least two different aspects to a COPD exacerbation. Principal components analysis of the items in the AECB-SS identified two components: one is composed of items concerned with cough and sputum; the other is made up of dyspnea, daily activity, and sleep disturbance (Spencer S, Sagnier P-P, Wilson R, Jones PW, unpublished data). However, it is possible for a PRO to address a unidimensional construct even though there may be different aspects to the construct. This is illustrated in Figure 5, which is a theoretical Rasch item map for a PRO that has two domains, but all of the items still fit the unidimensional model and form a single severity continuum. This is not just a theoretical possibility; it has been demonstrated with the St. George's Respiratory Questionnaire (21).
|
The next challenge will be to determine what constitutes a change in the symptoms from "the stable state beyond the normal day to day variations." This can be approached in the individual patient by setting a percentage change limit on what is the normal day-to-day variation. However, although such an approach might suit an individual patient's treatment plan, a different approach may be needed for a population of patients with COPD. A PRO for exacerbation will have been developed in populations of patients and, strictly speaking, will not be valid for an individual but will be valid for groups of individuals. This point the direction for establishing the threshold change "beyond normal daily variation." Prospective studies with the developed instrument can establish the normal variation and the sensitivity and specificity for different sizes of change in score for detecting the onset of an exacerbation. It should be possible to set an absolute value to the change in symptoms, which in clinical trials would be the method of choice. The use of Rasch methodology in the development of the instrument would make this task easier because interval-scaling properties mean that a given change in score will mean the same change in severity, regardless of the baseline state.
When discussing changes in score, it is important to distinguish between the size of change and the implications of the change. This can be illustrated best by an analogy with temperature: a rise of 2°C causes ice to melt if the baseline temperature is –1°C and water to boil if the temperature is 99°C. Similarly a 1-unit rise in score may characterize an exacerbation, but its impact will depend on the patient's clinical state when stable (Figure 6). The event-based outcome, decided by the treatment of the exacerbation, can still be retained, but, as can be appreciated, it will be closely linked to the baseline status of the patients.
|
The initial questions for the EXACT-PRO have been derived from patient interviews. Duplicates and unworkable items are removed and the items formatted to enable creation of an e-diary. This first phase, overseen by experts in the United States and Europe and members of the FDA, has been completed. Phase 2 of the project involves data collection in a large clinical study followed by empirically based item reduction and an evaluation of the reliability and validity of the measure. Phase 3 will be a prospective study of the sensitivity and specificity of the scores for detecting the onset and resolution of exacerbations.
CONCLUSIONS
There is no simple tool to record the onset and severity of the event. By returning to the clinical condition and considering the constituent symptoms, it will be possible to develop a single score that can capture both the magnitude and duration of the event. The process of evaluation of this approach has begun. The new instrument will provide new insights in exacerbations of COPD and assessments of preventative and acute therapy. It will also provide a most important clinical criterion for validation of potential biomarkers.
FOOTNOTES
Conflict of Interest Statement: P.J. has received from Bayer Pharma, over the period 2003–2005; consultancy fees totaling $15,000; and advisory board fees totaling $6,500; in 2004, a lecture fee of $1,500 and, over 2003–2004, a research grant of $25,000 were received. He has also received funds for a clinical trial over the period 2005–2006 totaling $150,000. From United BioSource in 2006, he received £6,000 in consultancy fees. T.H. is a full-time employee of AstraZeneca.
(Received in original form July 31, 2007; accepted in final form September 17, 2007)
REFERENCES
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
