National Study of Emergency Department Visits for Acute Exacerbation of Chronic Obstructive Pulmonary Disease, 1993–2005

Authors

  • Chu-Lin Tsai MD, MPH,

    1. From the Department of Emergency Medicine, Massachusetts General Hospital, Harvard Medical School, and Department of Epidemiology, Harvard School of Public Health (CLT, CAC), Boston, MA; and Johns Hopkins University School of Medicine (JAS), Baltimore, MD.
    Search for more papers by this author
  • Justin A. Sobrino BS,

    1. From the Department of Emergency Medicine, Massachusetts General Hospital, Harvard Medical School, and Department of Epidemiology, Harvard School of Public Health (CLT, CAC), Boston, MA; and Johns Hopkins University School of Medicine (JAS), Baltimore, MD.
    Search for more papers by this author
  • Carlos A. Camargo Jr MD, DrPH

    1. From the Department of Emergency Medicine, Massachusetts General Hospital, Harvard Medical School, and Department of Epidemiology, Harvard School of Public Health (CLT, CAC), Boston, MA; and Johns Hopkins University School of Medicine (JAS), Baltimore, MD.
    Search for more papers by this author

  • The project was supported by an unrestricted grant from GlaxoSmithKline (Research Triangle Park, NC).

  • Conflict of interest statement: Dr. Camargo has received financial support (research grants, consulting, lectures) in the past 5 years from Altana, AstraZeneca, Aventis, Aventis Pasteur, Boehringer Ingelheim, Dey, GlaxoSmithKline, MedImmune, Merck, Novartis, Pfizer, Respironics, and Schering Plough. The other authors have no conflicts of interest to disclose.

Dr. Chu-Lin Tsai; e-mail: ctsai2@partners.org. Reprints will not be available.

Abstract

Objectives:  Little is known about recent trends in U.S. emergency department (ED) visits for acute exacerbation of chronic obstructive pulmonary disease (AECOPD) or about ED management of AECOPD. This study aimed to describe the epidemiology of ED visits for AECOPD and to evaluate concordance with guideline-recommended care.

Methods:  Data were obtained from National Hospital Ambulatory Medical Care Survey (NHAMCS). ED visits for AECOPD, during 1993 to 2005, were identified using the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes. Concordance with guideline recommendations was evaluated using process measures.

Results:  Over the 13-year study period, there was an average annual 0.6 million ED visits for AECOPD, and the visit rates for AECOPD were consistently high (3.2 per 1,000 U.S. population; Ptrend = 0.13). The trends in the use of chest radiograph, pulse oximetry, or bronchodilator remained stable (all Ptrend > 0.5). By contrast, the use of systemic corticosteroids increased from 29% in 1993–1994 to 60% in 2005, antibiotics increased from 14% to 42%, and methylxanthines decreased from 15% to <1% (all Ptrend < 0.001). Multivariable analysis showed patients in the South (vs. the Northeast) were less likely to receive systemic corticosteroids (odds ratio [OR] = 0.6; 95% confidence interval [CI] = 0.4 to 0.9).

Conclusions:  The high burden of ED visits for AECOPD persisted. Overall concordance with guideline-recommended care for AECOPD was moderate, and some emergency treatments had improved over time.

Chronic obstructive pulmonary disease (COPD) is a major cause of morbidity and mortality worldwide.1 In the United States, COPD is the fourth leading cause of death, and the rising trend in mortality is unique among the top five causes of death.2 Data from National Health and Nutrition Examination Survey III estimate that approximately 24 million U.S. adults have evidence of impaired lung function.3 The natural history of COPD is punctuated by frequent and recurrent acute exacerbations of COPD (AECOPD), which are associated with a poorer quality of life,4 a faster decline in lung function,5 and increased mortality.6 About 50% of AECOPD require additional medical assistance,7 and U.S. patients often resort to using the emergency department (ED).8 The epidemiology of ED visits for COPD is therefore important for both the understanding of disease burden and the projecting future trends in healthcare utilization. The most recent national report states that ED visits for COPD increased from 1992 to 2000 and that COPD accounted for approximately 1.5 million ED visits in 2000.8

Another gap in the current literature relates to emergency management of AECOPD. Although several COPD practice guidelines have been published, little is known about the actual ED management of AECOPD in the United States. The main guidelines for the treatment of AECOPD over the past two decades include the first American Thoracic Society (ATS) guidelines published in 1995,9 Global Initiative for Chronic Obstructive Lung Disease (GOLD) guidelines published in 200110 and updated annually,11 the ATS and European Respiratory Society joint guidelines published in 2004,12 and the 2001 American College of Physicians guidelines13,14 with a recent update.15,16 We previously showed that adherence of ED management of AECOPD to current treatment guidelines was low at 29 academic EDs in North America.17 At the national level, the level of adherence remains unknown. Moreover, it is unclear whether the publications of these guidelines have had any influence on ED management over time. Monitoring concordance with guideline recommendations over time helps identify targets for quality improvement efforts.

To address these gaps in current knowledge, we analyzed data from the National Hospital Ambulatory Medical Care Survey (NHAMCS) from 1993 to 2005. The two objectives of this study were 1) to describe the trends in ED visits for AECOPD over the 13-year study period and 2) to evaluate concordance of ED treatment of AECOPD with recommendations in these clinical practice guidelines, as well as the change in concordance over time.

Methods

Study Design

The NHAMCS is a cross-sectional, four-stage probability sample of visits to noninstitutional general and short-stay hospitals, excluding Federal, military, and Veterans Administration hospitals, located in the 50 states and the District of Columbia.18 The NHAMCS is conducted annually by the National Center for Health Statistics (NCHS) and covers geographic primary sampling units, hospitals within primary sampling units, EDs within hospitals, and patients within EDs. The number of EDs sampled is around 400 each year. Trained ED staff collected clinical information during a randomly assigned 4-week data period for each of the sampled EDs, using a structured patient record form. Information included patients’ demographics, reasons for the visit, diagnoses, procedures, and medications given at the visit, as well as data on the physician and hospital. Quality control was performed using two-way independent verification procedure for 10% of the sample records. The nonresponse rate for most items was less than 5%, and error rates were less than 2% for items requiring medical coding.19 Our institutional review board waived review of this analysis.

Study Setting and Population

The NHAMCS data from 1993 to 2005 were combined for this analysis. We further limited ED visits made by adults aged 25 years or older for this analysis, because individuals younger than 25 years are unlikely to have COPD. Adults were defined as persons aged 25 years or older throughout the text for brevity and consistency. The three diagnosis fields in the NHAMCS were coded according to the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM).20 For the current analysis, we used ICD-9-CM codes 491 (chronic bronchitis), 492 (emphysema), and 496 (COPD not otherwise specified) to define COPD. We identified adult patient visits in which any of the three COPD codes was listed in the primary ED diagnosis field as AECOPD ED visits.

Measurements

ED Visit Rates for AECOPD.  To compute ED visit rates for AECOPD, we used the denominator of the estimate of adult resident (civilian, noninstitutionalized) population from the U.S. Census Bureau,21 according to the NCHS recommendations. To reflect the relative burden of AECOPD visits on EDs, we also used the adult ED population (i.e., all adults who attended ED during the study period) as an alternative population denominator. Average annual visit rates for the entire study period were calculated by dividing the estimated cumulative ED visits for AECOPD by the sum of the midyear estimate of denominators for each year. We also report ED visit rates stratified by age, gender, race, ethnicity, insurance, and region. To preserve consistency across years, race was recoded as white, African American, and other; insurance was recoded as private, public, other, self-pay, and unknown. U.S. regions represent standardized geographical divisions defined by the U.S. Census Bureau (Northeast, Midwest, South, and West).21 ED visit rates for AECOPD are reported per 1,000 individuals per year.

Clinical Information.  Up to three reasons for each ED visit were coded using Reason for Visit Classification for Ambulatory Care, a standardized sourcebook used in the NCHS studies.22 We evaluated the presence of the following symptoms for each ED visit: shortness of breath (symptom codes 1415–1430), cough (symptom codes 1440), and abnormality of sputum including changes in sputum production or purulence (symptom codes 1470). Disease severity data are limited but include urgency at triage. To be consistent across years, we recoded visits as urgent if “immediacy to be seen” variable was recorded as less than 15 or 15–60 minutes and as nonurgent if recorded as >1–2 hours or longer. We also evaluated the presence of the following concomitant diagnoses for each ED visit: pneumonia (ICD-9-CM 480–486), congestive heart failure (ICD-9-CM 428), and pulmonary embolism (ICD-9-CM 415.1).

The NHAMCS data allow us to examine two diagnostic tests that are relevant to AECOPD (chest radiography and pulse oximetry) and medications for the treatment of AECOPD. From 1993 through 1994, up to five medications were recorded per visit, from 1995 through 2002 up to six medications, and in 2003 through 2004 up to eight medications were recorded per visit. Therapeutic class of medication was based on the National Drug Code Directory (NDCD).23 We identified the following medications for treatment of AECOPD by using the NDCD class codes: 1) bronchodilators (class codes 1940), such as inhaled beta-agonists and anticholinergics and excluding methylxanthines; 2) systemic corticosteroids (class code 1032), excluding inhaled corticosteroids (class code 1947); 3) antibiotics (class code prefix 03), including penicillin, cephalosporins, macrolides, tetracyclines, chloramphenicols, quinolones, and miscellaneous antibacterials; and 4) methylxanthines, which do not have their own class codes and were identified by individual drug codes in the NDCD. If more than one drug in a particular drug category (e.g., antibiotics) was used in a single ED visit, we counted the visit only once as an episode of care in which the drug category was prescribed.

Concordance with Guidelines—Quality Indicators: Process Measures.  On the basis of recommendations contained in the COPD guidelines,12–14,24 we categorized the following diagnostic evaluations and treatments as beneficial quality indicators: chest radiography, pulse oximetry, inhaled bronchodilators, systemic corticosteroids, and antibiotics. The use of bronchodilator and systemic corticosteroids was recently included as quality measures in pharmacotherapy for AECOPD in the Health Care Effectiveness Data and Information Sets in 2008.25 We considered treatment with methylxanthine as nonbeneficial.26

Outcome Measures

ED outcomes included ED intubation rates, hospital admission rates, and intensive care unit (ICU) admission rates. While the follow-up patient outcomes (e.g., in-hospital mortality or recurrent ED visits) are important in evaluating ED quality of care for AECOPD, such data were not available in the NHAMCS.

Data Analysis

Stata v9.0 software (StataCorp, College Station, TX) that adjusts variances to account for the complex designs of the surveys was used to calculate standard errors (SEs) for the NHAMCS estimates. All statistical tests were based on estimates that had a less than 30% relative SE (i.e., the SE divided by the estimate expressed as a percentage of the estimate) and were based on at least 30 cases in the sample data, according to the NCHS recommendations. Descriptive statistics were presented as proportions (with 95% confidence intervals [CI]) or means (with SE). We used the weighted chi-square test for differences between proportions. We used Joinpoint Regression Program version 3.2.0 (Statistical Research and Applications Branch, National Cancer Institute, Bethesda, MD) to test for significant changes in visit rates that may have occurred during the study period. A log-linear regression model was fit with Monte Carlo permutation to assess significant changes in visit rates.27

To test for trends for each quality indicator, logistic regression models were fit in which “calendar year” was a continuous independent variable. Multivariable logistic regression model was performed to assess the independent predictors of treatment with systemic corticosteroids. We specifically modeled treatment with systemic corticosteroids because it represents an emerging treatment option during the study period that is eventually endorsed by several guidelines as an evidence Level A treatment.10,12,14 Multivariable model included age, gender, race, ethnicity, insurance, region, reason for visit, urgency at triage, and calendar year. We tested for two-way interaction between time trend and other significant predictors in the final multivariable model by including an interaction term in the model. Odds ratios (ORs) are presented with 95% CI. All p-values are two-sided, with p < 0.05 considered statistically significant.

Two sensitivity analyses were performed to test the robustness of our findings. First, the analyses were repeated by restricting the study population to patients aged 45 years or older. Second, in addition to age restriction, we further excluded patient visits for which codiagnoses of pneumonia, congestive heart failure, or pulmonary embolism were listed in the secondary or tertiary diagnosis fields.

Results

ED Visit Rates for AECOPD

Over the 13-year study period, there was an estimated total of 7.3 million (95% CI = 6.5 to 8.0 million) ED visits for AECOPD in the United States, or an average annual 0.6 million ED visits for AECOPD (Table 1). There was a slightly upward trend in the annual number of ED visits for AECOPD from 1993 to 2005, but this trend was not statistically significant (annual percent change = +2%; Ptrend = 0.17; Figure 1). In terms of visit rates, the overall visit rate for AECOPD was 3.2 per 1,000 U.S. population during the study period (Table 1). Similarly, there was a slightly upward trend in the AECOPD visit rates from 1993 to 2005, but this increasing trend was not statistically significant (annual percent change = +1%; Ptrend = 0.13; Figure 1). Using adult ED visits as denominator, the 7.3 million ED visits for AECOPD represented 0.9% of the 824 million adult ED visits from 1993 to 2005. The proportions of AECOPD ED visits among the adult ED visits were essentially unchanged during the study period (Ptrend = 0.55).

Table 1.   ED Visit Rates for AECOPD as the First-listed Diagnosis in the United States, 1993–2005
VariableCumulative Number of ED Visits for AECOPD in Thousands,*N (%)Annual ED Visit Rate for AECOPD per 1,000 US Adult Population (95% CI)†
  1. AECOPD = acute exacerbation of chronic obstructive pulmonary disease; CI = confidence interval; ED = emergency department; nc = not calculable.

  2. *Visit counts were multiplied by sampling weights, which account for the multistage sample design to obtain national estimates.

  3. †Based on the U.S. Census Bureau estimates of the noninstitutionalized civilian population.

  4. ‡Ethnicity missing for 8% of the sample.

  5. §Not calculable due to lack of corresponding denominator in stratum from the U.S. Census Bureau.

Overall7,2693.2 (2.9, 3.5)
Age group, years
 25–44352 (5)0.3 (0.2, 0.4)
 45–54738 (10)1.6 (1.3, 1.9)
 55–641,585 (22)5.1 (4.3, 5.8)
 65–742,345 (32)9.9 (8.6, 11.3)
 75+2,249 (31)11.8 (10.3, 13.3)
Gender
 Female3,728 (51)3.1 (2.8, 3.5)
 Male3,541 (49)3.3 (2.9, 3.7)
Race
 White6,288 (87)3.3 (2.9, 3.7)
 African American859 (12)3.4 (2.7, 4.1)
 Other122 (2)1.1 (0.5, 1.6)
Ethnicity‡
 Hispanic269 (4)1.1 (0.8, 1.5)
 Non-Hispanic6,397 (96)3.1 (2.7, 3.4)
U.S. region
 Northeast1,451 (20)3.2 (2.6, 3.7)
 Midwest1,866 (26)3.5 (2.9, 4.1)
 South2,812 (39)3.4 (2.8, 4.0)
 West1,140 (16)2.2 (1.6, 2.9)
Insurance nc§
 Private1,254 (17) 
 Public5,095 (70) 
 Other264 (4) 
 Self-pay358 (5) 
 Unknown298 (4) 
Figure 1.

 National trends in emergency department visits for acute exacerbation of COPD as the first-listed diagnosis, 1993–2005. The trend lines are for total number of visits (solid line) and annual rate per 1,000 US population (dashed line). The annual numbers and visit rates for AECOPD slightly increase from 1993 to 2005, but this trend is not statistically significant (Ptrend = 0.17 and 0.13, respectively). AECOPD = acute exacerbation of chronic obstructive pulmonary disease; COPD = chronic obstructive pulmonary disease; ED = emergency department.

Table 1 also demonstrates the ED visit rates for AECOPD stratified by demographic factors. The ED visit rates increased with age, but did not differ by sex, race, or region. The visit rates were lower in Hispanics than non-Hispanics.

Characteristics of ED visit for AECOPD

The majority of the ED visits for AECOPD were made by patients aged 65 years or older (Table 1), with a mean age of 67 years (95% CI = 67 to 68 years). The number of visits was approximately equal between men and women. The number of visits was higher in whites, in non-Hispanics, in the South, and in patients with public insurance. With respect to ED presentation, the ED visits more often occurred from December to May (Table 2). The most common reasons for the visit were shortness of breath, which was present at 78% of the visits. Cough and abnormality of sputum were relatively uncommon presenting symptoms. Eighty-four percent of the visits were considered as urgent at triage. A small percentage of the visits had concomitant diagnoses: 8% had pneumonia, 6% had congestive heart failure, and very few of them (<0.1% in the sample) had pulmonary embolism. There were no secular trends in presenting symptoms, urgency at triage, and concomitant medical disorders (Ptrend > 0.2 for all).

Table 2.   ED Presentation and Management among ED Visits for AECOPD in the United States, 1993–2005
VariableProportion of ED Visits for AECOPD, % (95% CI), N = 7,269,000
  1. AECOPD = acute exacerbation of chronic obstructive pulmonary disease; CI = confidence interval; ED = emergency department; ICU = intensive care unit; nc = not calculable.

  2. *Not calculable, because n < 30 or relative standard error >30%.

  3. †Available from 1993 to 2004. Chest X-ray was changed to “any X-ray” in 2005 and 2005 data were not used.

  4. ‡Available from 1995 to 2005.

ED presentation
Season
 December–February29 (26, 32)
 March–May29 (26, 32)
 June–August21 (18, 23)
 September–November22 (19, 24)
Reason for visit
 Shortness of breath78 (76, 81)
 Cough15 (13, 17)
 Abnormality of sputum3 (2, 4)
 Urgent at triage84 (81, 86)
Concomitant diagnosis
 Pneumonia8 (6, 9)
 Congestive heart failure6 (5, 7)
 Pulmonary embolismnc*
ED management
Process measure
 Chest radiography†77 (74, 80)
 Pulse oximetry‡57 (53, 61)
 Bronchodilators60 (57, 63)
 Systemic corticosteroids45 (42, 48)
 Antibiotics28 (25, 31)
 Methylxanthines6 (5, 8)
Outcome measure
 ED intubation1 (0.4, 1.5)
 Hospital admission50 (47, 54)
 ICU admission‡6 (4, 7)

Secular Trends in ED Management of AECOPD

Process Measures.  We examined whether the ED management of AECOPD had changed over the 13-year study period. The use of chest radiography and pulse oximetry remained unchanged from 1993 to 2005 (Ptrend = 0.61 and 0.58, respectively). On average, chest radiography was performed at 77% of the ED visits and pulse oximetry at 57% of the visits.

With respect to medications, bronchodilator use (either beta-agonists or anticholinergics) was relatively stable from 1993 to 2005 (Ptrend = 0.92). On average, bronchodilators were given at 60% of the ED visits. By contrast, the use of systemic corticosteroids increased significantly from 29% in 1993 through 1994 to 60% in 2005 (Ptrend < 0.001; Figure 2). The most commonly used antibiotics were cephalosporins (39%) and macrolides (29%). The use of antibiotics increased from 14% in 1993 through 1994 to 42% in 2005 (Ptrend < 0.001), whereas the use of methylxanthines decreased from 15% in 1993 through 1994 to <1% in 2005 (Ptrend < 0.001).

Figure 2.

 National trends in ED treatment of acute exacerbation of COPD as the first-listed diagnosis, 1993–2005. Antibiotics include penicillin, cephalosporins, macrolides, tetracyclines, chloraphenicols, quinolones, and miscellaneous anitibacterials. The P-value for trend for all three medications is <0.001. AECOPD = acute exacerbation of chronic obstructive pulmonary disease; COPD = chronic obstructive pulmonary disease; ED = emergency department.

Outcome Measures.  The trends in ED intubation, hospital admission, and ICU admission were relatively stable from 1993 to 2005 (Ptrend = 0.87, 0.16, and 0.89, respectively). On average, intubation was performed in only 1% of the ED visits. With regard to ED disposition, 50% of the visits resulted in hospitalization with 6% to the ICU.

Treatment with Systemic Corticosteroids and Overall Concordance

In the univariable analyses, the likelihood of receiving systemic corticosteroids in the ED did not vary by age, gender, race, or ethnicity (Table 3). Patients with public insurance were less likely to receive systemic corticosteroids, compared with the privately insured. Geographically, systemic corticosteroids were less likely to be given at ED visits in the South (38%). Systemic corticosteroids were more likely to be given at visits presenting with shortness of breath and at visits that were deemed urgent at triage.

Table 3.   Unadjusted and Multivariable Analysis of Factors Associated Systemic Corticosteroid Treatment among ED Visits for AECOPD in the United States, 1993–2005
VariableProportion of AECOPD Visits at which Systemic Corticosteroids Were Administered, %* Unadjusted OR (95% CI) Adjusted OR (95% CI)†
  1. AECOPD = acute exacerbation of chronic obstructive pulmonary disease; CI = confidence interval; ED = emergency department; OR = odds ratio.

  2. *Numbers in parentheses in rows of dichotomous variables indicate the percentage of the reference category receiving systemic corticosteroids.

  3. †Multivariable model adjusting for all variables shown in the table.

Age group, years
 25–44431.01.0
 45–54481.2 (0.6, 2.4)0.9 (0.5, 2.0)
 55–64471.2 (0.6, 2.3)0.8 (0.4, 1.5)
 65–74471.2 (0.6, 2.3)0.9 (0.5, 1.8)
 75+421.0 (0.5, 1.9)0.8 (0.4, 1.6)
Gender
 Female441.01.0
 Male461.1 (0.9, 1.3)1.2 (0.9, 1.6)
Race
 White451.01.0
 African American410.9 (0.6, 1.3)1.0 (0.7, 1.5)
 Other461.0 (0.4, 2.6)0.7 (0.3, 1.8)
Ethnicity
 Hispanic51 (vs. 45)1.3 (0.7, 2.3)1.2 (0.6, 2.2)
Insurance
 Private531.01.0
 Public430.7 (0.5, 0.9)0.8 (0.5, 1.2)
 Other430.7 (0.3, 1.2)0.9 (0.4, 1.8)
 Self-pay400.6 (0.3, 1.1)0.6 (0.3, 1.3)
 Unknown520.9 (0.5, 1.9)0.9 (0.4, 2.1)
U.S. region
 Northeast491.01.0
 Midwest480.9 (0.7, 1.3)0.9 (0.6, 1.3)
 South380.6 (0.5, 0.9)0.6 (0.4, 0.9)
 West531.2 (0.8, 1.6)1.2 (0.9, 1.8)
Reason for visit
 Shortness of breath50 (vs. 29)2.5 (1.8, 3.4)2.3 (1.7, 3.3)
 Cough47 (vs. 45)1.1 (0.8, 1.5)1.3 (0.9, 1.9)
 Abnormality of sputum48 (vs. 45)1.1 (0.6, 2.1)1.4 (0.7, 2.8)
 Urgent at triage49 (vs. 33)2.0 (1.4, 2.8)1.7 (1.2, 2.5)
 Calendar year, per 1-year increase1.04 (1.04, 1.05)1.10 (1.07, 1.14)

Multivariable analysis confirmed that patients in the South (vs. the Northeast) were less likely to receive systemic corticosteroid treatment (OR = 0.6; 95% CI = 0.4 to 0.9), independent of baseline patient characteristics and severity at ED presentation. Presentation to the ED with shortness of breath (OR = 2.3; 95% CI = 1.7 to 3.3) and urgency at triage (OR = 1.7; 95% CI = 1.2 to 2.5) increased the likelihood of receiving systemic corticosteroids. During the study period, the likelihood of systemic corticosteroid treatment increased over time (OR = 1.10; 95% CI = 1.07 to 1.14 per 1-year increase). There were no significant interactions in the final model.

Sensitivity Analyses

When restricting the study population to patients aged 45 years or older, the cumulative number of ED visits decreased to 6.9 million (5% reduction). With this small change in patient population, the overall trend in ED visit rates was preserved. With respect to ED management, the trends in process and outcome measures were quite similar in this subset analysis, i.e., the significant trends in the use of systemic corticosteroids, antibiotics, and methylxanthines (Ptrend < 0.001 for all) and the nonsignificant trends in others. Likewise, the significant predictors of corticosteroid use were similar to those in the primary analysis (i.e., calendar year, southern region, shortness of breath, and urgency). Further exclusion of visits with concomitant medical diagnoses did not materially change the main results (data not shown).

Discussion

These nationally representative data demonstrate a persistently high disease burden of ED visit rates for AECOPD from 1993 to 2005. With respect to actual ED management of AECOPD, the overall concordance with guideline recommendations was moderate. On an encouraging note, we found improvements in some of the guideline-concordant care: significant increases in the use of systemic corticosteroids and antibiotics and a significant decrease in the use of methylxanthines during the study period. None of the quality measures showed significant deterioration.

The upward trend in ED visit rates for AECOPD prior to 2000 is consistent with a previous study by Mannino and colleagues8 using these same NHAMCS data. We have extended their earlier finding by reporting that the high ED visit rates for AECOPD persisted in recent years. Although many new pharmacologic interventions have become available that can prevent COPD exacerbations,28 this does not necessarily translate into reduced ED visits. We have shown that the frequency of ED visits for AECOPD is associated with not only disease factors, but also healthcare-related factors.29 The consistently high ED visits for AECOPD might be related to the remaining challenges in our healthcare system, such as shortage of primary care and/or lack of universal insurance coverage.30

The significant improvement in some ED treatments of AECOPD nicely parallels the advance in our understanding of AECOPD over the past decades. National guidelines provide important treatment information and may influence clinical practice, if not directly from reading the guidelines, then through a diffusion of knowledge through EM residency training, journals, and other sources of physician education. For instance, the decreasing trend in the use of methylxanthines was concordant with changes in recent guidelines that consider methylxanthines a second-line therapy10,12 and improved knowledge about its use in emergency the treatment of AECOPD.26 With respect to systemic corticosteroids, the 1995 ATS guidelines reported “limited information supporting the use of intravenous or oral steroids in the management of COPD exacerbations.”9 From 1995 to 2001, a number of important studies demonstrated that the use of systemic corticosteroids may reduce treatment failure and improve dyspnea in AECOPD.31-33 As a result, the GOLD guidelines in 2001 clearly recommend the use of systemic corticosteroids in AECOPD (Evidence A).10 Moreover, a 2003 ED-based study showed that prescription of systemic corticosteroids reduced the post-ED relapse rate in AECOPD.34 These new data may have contributed to the increasing ED use of systemic corticosteroids over the 13-year study period.

Although the use of systemic corticosteroids has increased, it remained low (60%) in 2005, the most recent year available. The finding that shortness of breath and urgency at triage predicts the use of systemic corticosteroids seems reasonable, given its likely relation to exacerbation severity. However, it is less apparent why ED visits in the South were associated with a lower likelihood of prescription of systemic corticosteroids compared with the Northeast. Prior studies have demonstrated regional variations in the quality of care for stable COPD35 and in hospital care for AECOPD;36 we extend this finding to a previously unstudied ED population. Future studies should assess reasons for this disparity, particularly at the ED level, and identify the barriers to the delivery of high-quality emergency care.

In the management of AECOPD, chest radiographs are useful in identifying mimics of exacerbation, and pulse oximetry can be used to evaluate the need for supplemental oxygen therapy.10,12 Although both are recommended in the guidelines,10,12 the NHAMCS data suggest that both were underused over the 13-year study period (77 and 57%, respectively). In our previous multicenter study of AECOPD,17 which involved 29 academic EDs in 2000 to 2001, all available quality measures, including chest radiography (87%), pulse oximetry (96%), systemic corticosteroids (62%), and bronchodilators (91%), were better than their corresponding national averages. Taken together, these two studies provide preliminary benchmarking data for ED quality of care for AECOPD. As stated in the Institute of Medicine report,37 continuous monitoring and tracking processes of care is needed to reduce practice variation and to shift the average performance toward excellence.

Limitations

Many sources of variation can affect national estimates of ED visit rates for AECOPD, including case definitions, ICD-9 codes, and population denominators. We have attempted to capture, in our opinion, the more precise patient population by using primary diagnosis of AECOPD. However, this approach might exclude a small number of true AECOPD cases whose codes were listed in the secondary or tertiary diagnosis field.

NHAMCS data did not include the information on out-of-hospital use of medications, i.e., self-medications or medications given by emergency medical services. For example, the low use of bronchodilators may be due to their use at home or en route to the hospital,17 which then led to nonuse of bronchodilators in the ED.

Some of the quality shortfall may be due to underdocumentation. Such underdocumentation, however, cannot explain both the increasing and the decreasing trends we identified over the 13-year period. Finally, the ideal outcome measures for the processes of care should have included the study endpoints that were reported in the randomized clinical trials (e.g., relapse); however, such follow-up data are not available in the NHAMCS. The outcome measures we used did have limitations. Intubation events were rare, and thus may not be powered to detect the impact of improved ED care. In addition, the decision to admit may be influenced by nuanced physician judgments, and this might explain why admission rates did not decrease in response to improved ED care over time.

Conclusions

We provide updated information on the epidemiology of ED visits for AECOPD and demonstrate persistently high ED visits rates from 1993 to 2005. We also provide information on the national trends in ED management of AECOPD by showing that some, but not all, emergency treatments have improved over the 13-year study period. Our study has important implications for clinicians and policy makers. First, given the large burden of disease, much work remains to be done to reduce the AECOPD ED visit rates, both at the individual level (i.e., secondary prevention of exacerbations) and at the system level (e.g., improving access to primary care). Second, although the overall quality of care delivered to ED patients with AECOPD has improved, there is still room for improvement, particularly in southern EDs. Given the positive experiences in the management of other diseases such as acute myocardial infarction and pneumonia,38 continuous quality improvement efforts in refinement, dissemination, and implementation of clinical practice guidelines may further improve ED quality of care for AECOPD.

Acknowledgments

The authors thank Jennifer A. Emond, Sunghye Kim, and Andrea J. Pelletier for their help with the statistical analysis of earlier data sets.

Ancillary