Published ahead of print on November 29, 2007, doi:10.1164/rccm.200708-1290OC
© 2008 American Thoracic Society doi: 10.1164/rccm.200708-1290OC
Underreporting Exacerbation of Chronic Obstructive Pulmonary Disease in a Longitudinal Cohort1 Respiratory Epidemiology and Clinical Research Unit, Montreal Chest Institute, and 2 Department of Epidemiology, Biostatistics, and Occupational Health, McGill University; and 3 Department of Medicine, McGill University Health Center, Montreal, Quebec, Canada Correspondence and requests for reprints should be addressed to Jean Bourbeau, M.D., Respiratory Epidemiology and Clinical Research Unit, Montreal Chest Institute, 3650 St. Urbain, Montreal, PQ, Canada H2X 2P4. E-mail: jean.bourbeau{at}mcgill.ca
Rationale: Unreported exacerbations and failure to seek medical attention may have consequences on the health of patients with chronic obstructive pulmonary disease. Objectives: This study aims to determine the incidence of reported and unreported exacerbations, to identify predictors of reporting, and to compare the impact of reported and unreported exacerbations on health status. Methods: The study is based on a multicenter Canadian cohort of patients with chronic obstructive pulmonary disease. Measurements and Main Results: Patients completed a daily diary from which exacerbations were defined as a worsening of at least one key symptom (dyspnea, sputum amount, sputum color) recorded on at least 2 consecutive days. Patients were asked to contact the study center if there was a sustained worsening of symptom. Reported exacerbations were events that led to contacting study center or health care visit. The study enrolled 421 patients. The overall incidence of diary exacerbations was 2.7 per person per year, but only 0.8 per person per year was reported. Predictors of reporting included age (HR [hazard ratio], 0.90; 95% confidence interval [CI], 0.81–0.98 per 5-yr increase), FEV1% predicted (HR, 0.84; 95% CI, 0.70–0.99 per 10% increase), number of symptoms at onset (HR, 1.59; 95% CI, 1.37–1.84 per additional symptom), and time of the week (HR, 0.35; 95% CI, 0.22–0.56 weekend vs. weekday). There was a clinically important decline in health status for 52% of patients with reported exacerbation and 43% with unreported exacerbations. Conclusions: This study has shown that less than one-third of the exacerbations were reported. The number of symptoms at onset was the most important predictor of reporting an exacerbation, and both reported and unreported exacerbations had an impact on health status.
Key Words: chronic obstructive pulmonary disease • exacerbation • health status • predictor
Chronic obstructive pulmonary disease (COPD) is a common condition that results in substantial economic and social burdens to society (1–3). The natural course of COPD includes a progressive decline in patient functional capacity and health-related quality of life (4). Acute exacerbation of COPD is one factor related to both transient and long-term decline of functional capacity and health-related quality of life (5–7). The issue of underreporting of COPD exacerbations has been addressed by a series of articles, all of which are based on a cohort of patients with COPD living in East London (5, 8–10). Results from this cohort suggest that close to half of all exacerbations remain unreported. A priori, one would expect that reporting is likely to be influenced by severity and impact of symptoms. Surprisingly, reported and unreported exacerbations identified in the London study had very similar characteristics, including type of symptoms, change in lung function, duration of symptoms, and duration of lung function change (5, 8–10). A high incidence of unreported exacerbations with severity similar to reported and treated exacerbations would indicate an unmet health care need. Self-management education programs with standing prescriptions for antibiotics and/or prednisone in the event of an exacerbation have been proposed as a strategy for early recognition and treatment of exacerbations (11, 12). Such a strategy has been shown to be effective in reduction of hospital admissions (11, 13), and may be one way of addressing underreporting of exacerbations. The mechanisms and consequences of underreporting sustained changes in symptoms are not well known. The failure to seek medical attention may have consequences for both the patient and the health care system. Additional research is needed to determine how patterns of symptom worsening, patient behavior in reporting or not reporting exacerbations, and impact on health status are related among patients with COPD. The present study took advantage of a large multicenter Canadian cohort of patients with a history of COPD exacerbations (14, 15). More specifically, the objectives of the study were as follows: (1) to determine the relative incidence of reported and unreported exacerbations, (2) to identify predictors of delay and failure to report an exacerbation, and (3) to compare the impact of reported and unreported exacerbations on patients' health.
This is a secondary analysis of data collected during a drug study. The design and methods of this study have been described in more detail elsewhere (14), and are summarized below.
Study Patients
Study Design
Data Collection To assess longitudinal change in health status, patients were asked to complete the SGRQ and the CCQ at each scheduled visit. Additional information gathered at each visit included changes in medication, health care resource utilization, and current smoking. Patients were asked to contact the study center if there was a sustained (at least 24 h) worsening of any symptom. The first day of change in symptoms was taken as the day of onset of exacerbation. In addition to the first patient-initiated call, there were two follow-up calls: one 5 to 7 days after the first call and a second 12 to 14 days after the first call. Each telephone interview included the completion of the SGRQ, excluding the symptoms domain, and a reminder to the subject to complete the exacerbation notebook (self-administered CCQ, modified MRC, and activity of daily living). Patients were asked to report information on a daily basis in a diary. Patients were reminded at monthly intervals to complete diary records (either at scheduled visits or by phone). The information recorded in the daily diary included symptom changes, peak expiratory flow (PEF), use of rescue medication, and use of health care resources. The following symptoms were included in the daily diary: dyspnea, sputum color and amount, wheeze, cough, colds, or sore throat. The symptom questions had dichotomous response options, with a positive response indicating that the symptom was worse than at baseline.
Exacerbations
Statistical Methods Statistical analysis was performed using Stata version 8 (Stata Corp., College Station, TX).
Patients and Diary Recording Characteristics of the entire study population are given in Table 1. There were 421 patients enrolled in the cohort; 225 patients had a daily diary exacerbation (i.e., increase in at least one key symptom for at least 2 d) and 196 patients had no daily diary exacerbation. Patients were more frequently men (57%) and had moderate to severe disease, and 40% were occasional or current smokers. The distributions of baseline variables in the two subgroups were similar except that there were slightly more women, with higher education level, and a higher number of prior exacerbations in those with at least one diary exacerbation. Of the patients with at least one daily diary exacerbation, 62% reported an exacerbation. An exacerbation was also reported in 19% of those with no daily diary exacerbation.
There were only 131 individuals (31%) with no missing data, but 325 individuals (77%) out of the entire study population had better than a 95% completion rate of the daily diary.
Exacerbations The overall estimated rate of diary exacerbations was 2.7 per person per year. The estimated rate of reported exacerbations was 0.8 per person per year and 1.9 per person per year for unreported exacerbations. Table 2 presents the characteristics of reported and unreported exacerbations. Reported exacerbations were in general longer, had more symptoms, and required more rescue medication. Change in PEF was similar for both groups of events. Among those exacerbations with a single symptom present at onset, only 14% were reported, whereas 59% of those with four or more symptoms were reported. Reporting was also related to symptom type, with 34% of those with dyspnea reported versus 53% of those with a change in sputum color being reported.
Predictors of Delay or Failure to Report Exacerbations Table 3 presents the relationship between subject characteristics and the likelihood of reporting an exacerbation. Analysis was performed on 213 first events to assess the predictors of reporting, and 69 of these events were reported (i.e., had the specified outcome before recovery). Two subject-level variables were found to be significantly associated with reporting: age and baseline FEV1 (% predicted normal value). Those who were older and who had less severe disease were less likely to report an exacerbation. The best model including both age and sex was a model with age, sex, and baseline FEV1 (% predicted normal). Crude and adjusted results were very similar, indicating no strong confounding of the main associations between age, sex, and FEV1 (% predicted normal), and reporting an exacerbation.
Table 4 presents the relationship between characteristics of the event and the likelihood of reporting an exacerbation. The average number of symptoms present during the first 3 days was strongly associated with reporting. Each additional daily symptom was associated with an estimated 58% (95% confidence interval, 37–81%) increase in the likelihood of reporting. Time of week was another variable that was strongly associated with reporting; weekend days (i.e., Friday, Saturday, and Sunday) were associated with a lower likelihood of reporting. Other event characteristics that were significantly associated with reporting include change in rescue medication use from baseline and type of symptoms. The average number of symptoms was associated with PEF change, rescue medication use, and the presence of any particular symptom. Adjusted hazard ratios refer to estimates from a model including age, sex, FEV1 (% predicted normal), and average number of symptoms at onset. The adjustment due to age, sex, and FEV1 was very minor and the observed changes in estimated parameters were essentially due to the adjustment for average number of symptoms at onset. After taking into account the overall number of symptoms, there was heterogeneity by symptom type in whether a symptom was positively or negatively associated with reporting, with cough and sputum quantity being in the former category and cold and wheeze in the latter category.
Impact of Exacerbation Reporting on Health Status Table 5 presents the distribution of SGRQ (total and subscore) changes between the 1- and 3-month visits stratified by presence and type of exacerbation. Despite substantial variation, this deterioration was highest in patients who had a reported exacerbation, particularly when considering the symptom and impact components of the SGRQ. There was also a difference between those with unreported exacerbation and those with no exacerbation. The two exacerbation groups had statistically worse P < 0.05) deterioration in SGRQ total score than those without exacerbation. The deterioration in SGRQ total score was clinically significant (increase of 4 or more) in 52.3% of patients who reported their exacerbations as compared with 43.2% of those with unreported exacerbations and 18.9% of those with no exacerbations.
This is the first study to evaluate the predictors of reporting exacerbations and to study the consequences that reported and unreported exacerbations have on health status. The results show that characteristics of the event were the strongest predictors of reporting. The patient characteristics most associated with reporting an exacerbation were age and baseline spirometry. Patients who reported their exacerbation were more likely to have worsening of their health status compared with patients not reporting or those with stable disease. The overall results of our study are not comparable to the London study because our operational definition was slightly different. For comparison, we note that 60% of the events with at least two symptoms and one key symptom (dyspnea, sputum quantity, sputum color) recorded in a daily diary card were not reported. This result is similar to results of the earlier London studies (9, 10), but higher than the results from later studies after substantially improved compliance (5, 20). This high rate of underreporting in a separate population makes it likely that underreporting is a widespread phenomenon. Although there was no direct measure of exacerbation severity, the total number of symptoms at onset was a remarkably consistent predictor of reporting. Reporting was also related to need for rescue medication. This is likely a proxy of severity of the exacerbation. The symptoms with the strongest association to reporting an exacerbation in this study were cough and sputum quantity, and the symptoms least associated with reporting were colds and wheeze. This relative ranking contrasts with the consensus definition of an exacerbation as involving increased dyspnea, increased sputum quantity, and change in sputum color, and requires further investigation. Sputum color was identified as one of the key determinants of health care utilization in a small study looking at patient perspectives on exacerbations (21), but it was not the strongest determinant of reporting in this study. Need or perceived need seems to be the most important predictor of health care utilization (22, 23). Reporting an exacerbation is a proxy for health care utilization and therefore the results may be considered in light of previous research on health care utilization. Patient characteristics, age and baseline spirometry were also associated with reporting an exacerbation. The finding that patients with a lower FEV1 were more likely to seek medical attention was also noted in another study (24). This is also consistent with the observation showing that physician-rated exacerbation severity is correlated to the severity of the underlying disease (25). Age appears to be independently associated with reporting: those who are older are less likely to report an exacerbation. A negative association between age and reporting could also be explained by increasing feelings of helplessness, frustration, or futility. The results for other patient-level characteristics were inconclusive. Other studies have shown that socioeconomic status is related to health care utilization. Socioeconomic status affects both health status and access to health care, and consequently the increased need of those with lower socioeconomic status might be offset by lower access to health care. Finally, work of Adams and colleagues (21) indicates one main limitation of our analysis—namely, that psychological factors and impact on activities of daily living were not considered as possible predictors and may in fact be as important as disease status and symptoms. Anxiety and depression are prevalent in patients with COPD (26). Anxiety and depression may play a role both in symptom development and patient behavior and should be considered as potentially important predictors. Psychological factors have been found to be related to the reporting of respiratory symptoms (cough, wheeze, sputum, and dyspnea) (27). In this study, we have also evaluated whether exacerbations that are reported compared with unreported exacerbations have different consequences on patient health. As expected, there were also observed differences between the health status change among the different groups, with those with a reported exacerbation having on average the worst outcomes and those with no exacerbation having on average the best outcomes. However, there was substantial variation in the change in health status in all groups, indicating that there are other important factors affecting individual patient health status. One of the strengths of our study is that our population was recruited from 59 centers, including nonacademic centers. The different types of clinical practice make the study population naturally closer to patients with COPD in the general population. The relatively brief period of the study sample reduces possible biases that might be introduced by training effects and selective retention. However, drug trial populations have well-known limitations that relate to inclusion/exclusion criteria (28). The exclusion criteria have an effect on the underlying population parameters, including exacerbation rate, and may well have an effect on the likelihood of underreporting. The study can identify factors that might be associated with reporting in the target population, but access to health care is likely to be an independent risk factor for underreporting in the general population. This study was a secondary analysis of existing data. The particular diary used offered the possibility of assessing symptoms most often associated with exacerbation. The inclusion of other symptoms, such as cough, wheeze, sore throat, and cold, enabled further exploration. However, fever, fatigue, and night awakenings, which might provide additional insight, were not included. The main limitation of the diary used in this study is that because the diary responses were dichotomous, there were substantial floor and ceiling effects. Missing data in the daily diary were also related to the outcome of interest. The combination of both missing data and ceiling effects resulted in failure to identify some of the reported exacerbations in the daily diary. It is not possible to correct for ceiling effects, because once a symptom is present, no further change will be recorded. There were other potential limitations to the current analysis. There were only 69 failures (i.e., reported exacerbations) in the Cox model used, giving limited power to assess predictors and possible interactions. It was also not possible to validate the model on an independent sample. Furthermore, the analysis used ignored possible differences in symptom trajectory (e.g., early recovery of some symptoms), and these differences might be related to reporting. Finally, the number of possible combinations of symptom variables posed additional complications. Although there were 112 possible combinations, the exacerbations were not evenly distributed among these combinations and therefore inferences cannot be made for the unobserved combinations of multiple symptoms. In summary, this study has shown that event-based definitions of exacerbations almost certainly fail to capture all exacerbations. Reporting is related to subjects and event characteristics—in particular, the total number of symptoms. Unreported exacerbations may not be serious enough to warrant an emergency visit or hospitalization, but we have shown that they may still have an important impact on health status for a given patient. Using a symptom-based definition for an exacerbation provides different and likely complementary information on the health status of patients. The clinical implication of these results is important. Treatment might be appropriate for many of these events and so the main issues are to use strategies that will increase recognition of significant changes of symptoms and increase treatment access for patients. The means of an action plan to help patients recognize symptom changes, to implement self-care, and to self-initiate prescription in the event of an exacerbation has been suggested as a promising strategy (12, 29). Research is necessary to establish the potential role of this strategy and early treatment of COPD exacerbation. Furthermore, the long-term impact of unreported exacerbations on health status, lung function decline, hospitalizations, and mortality is unknown. Whether delay or failure to seek treatment during symptom change is causally related to time to next exacerbation and long-term outcome is unknown.
Supported by AstraZeneca Canada, Inc. Originally Published in Press as DOI: 10.1164/rccm.200708-1290OC on November 29, 2007 Conflict of Interest Statement: L.L.'s institution receives funding from Merck Frosst Canada Ltd., Eli Lilly Canada, Inc., Novartis Pharmaceuticals, Inc., and The Alliance: Sanofi-Aventis and Procter and Gamble Pharmaceuticals Canada, Inc. R.W.P. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. P.E. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. J.B. has received fees for speaking at conferences, and for serving as an expert on advisory boards for GlaxoSmithKline, Boehringer Ingelheim, Pfizer, and AstraZeneca. J.B.'s institution, MUHC Research Institute, received research grants from GlaxoSmithKline ($300,000), Boehringer Ingelheim ($50,000), Pfizer ($50,000), and AstraZeneca. Received in original form August 31, 2007; accepted in final form November 27, 2007
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
AT A GLANCE COMMENTARY
