Contemporary Management of Acute Exacerbations of COPD^*: A Systematic Review and Metaanalysis FREE TO VIEW

Acute exacerbations of COPD are associated with significant morbidity, mortality, and health-care expenditures. The in-hospital mortality rate for acute COPD exacerbations is approximately 10%,¹ and approximately 25% for those requiring admission to an ICU.² Hospitalizations for COPD exacerbations have increased significantly over the past 10 years. For instance, the number of hospitalizations for COPD exacerbations in the United States was 463,000 in 1990; whereas by 2000, it was 726,000, representing a 57% increase in just 10 years.³ The total economic costs of COPD in the United States were estimated at $32 billion in 2002, with $18 billion representing direct medical expenditures related largely to in-hospital care.⁴ The Global Initiative for Chronic Obstructive Lung Disease (GOLD) committee⁵ defines COPD exacerbation as a change in a patient’s “baseline dyspnea, cough, and/or sputum that is beyond day-to-day variations, is acute in onset, and may warrant a change in regular medication in a patient with underlying COPD.” Lung inflammation and infection appear to play prominent roles in the pathogenesis of COPD exacerbations, leading to worsening of symptoms and health status and a decline in lung function.⁶ The most common putative precipitants of exacerbations are bacterial or viral infections, and environmental pollution, although in many cases a clear precipitant is not apparent.⁷⁸ As patients frequently experience worsening of lung function and health status, most clinical guidelines recommend the use of bronchodilators (short-acting β₂-agonists and anticholinergics) to reduce dynamic hyperinflation, systemic corticosteroids to reduce lung inflammation, antibiotics to treat potential bacterial pathogens, and occasionally noninvasive positive pressure ventilation (NPPV) in select cases to reduce the work of breathing. Despite these widely promulgated recommendations, very few studies have systematically evaluated the clinical benefits of these interventions in comparison with placebo or standard therapy. The objective of this study was to systematically review and quantitatively synthesize the impact of systemic corticosteroids, antibiotics, and NPPV on treatment failure, need for intubation, in-hospital mortality, and LOS during COPD exacerbations.

We decided a priori to examine the published evidence for systemic corticosteroids, antibiotics, and NPPV in the treatment of acute COPD exacerbations. For each of these therapies, we conducted a literature search by using MEDLINE and EMBASE. We limited the search to English-language articles published from January 1968 to November 2006, conducted in adults (> 19 years of age) using a randomized, controlled trial design. To limit the studies to COPD, we used the following terms: chronic obstructive lung disease, chronic obstructive airway disease, COPD, emphysema, chronic bronchitis, or chronic airflow obstruction. Detailed search terms for each of the therapies and search results are available in Table 1 . To supplement this search, we examined the Cochrane Database of Systematic Reviews as well as bibliographies of retrieved articles.

We included only studies that were conducted during acute COPD exacerbations, as defined by worsening cough or dyspnea or increased sputum production. Studies were excluded when there was clearly an alternative primary diagnosis such as asthma, pneumonia, or cardiogenic pulmonary edema. Most of the included studies had criteria excluding patients with an alternative primary diagnosis based on clinical examination. In all included studies, the treatment group was compared with placebo or with standard medical therapy. Standard medical therapy included supplemental oxygen (if the patients were hypoxemic), bronchodilators, antibiotics, corticosteroids, and diuretics but could not include the treatment being studied. We used the Jadad scale, which is composed of 5 questions about study design, to adjudicate the methodologic quality of the studies (the higher the score, the better the quality of trial design).⁹ For systemic corticosteroids and antibiotics, we restricted the analysis to randomized clinical trials that had a Jadad score ≥ 3. For NPPV, we accepted studies with a score ≥ 2 because study blinding was not possible for practical reasons. All included studies had complete or near-complete follow-up data, and baseline characteristics that were well balanced between the treatment and control groups. Studies in abstract form were included if the methods and results could be adequately analyzed to minimize publication bias.

Data were abstracted from each trial by two authors (B.S.Q, W.Q.G.) independently using a prestandardized data abstraction form. Any discrepancies were resolved by iteration and consensus. Results were analyzed by intention to treat whenever possible. Treatment failures following treatment of COPD exacerbations were defined variably between studies. In most cases, treatment failures were defined as unchanged or deteriorated symptoms requiring additional treatment during the follow-up period. When possible, for each outcome we combined the results from individual studies to produce summary effect estimates. For dichotomous outcomes, relative risks (RRs) and 95% confidence intervals (CIs) were calculated. For continuous variables, weighted mean differences (and 95% CIs) were used to pool the data. Heterogeneity of results across individual studies was examined using the χ² test. If significant heterogeneity was observed (p ≤ 0.10), the DerSimonian and Laird random-effects model was used to pool the results together. In the absence of significant heterogeneity (p > 0.10), a fixed-effects model was used. All analyses were conducted using statistical software (RevMan version 4.2; Cochrane Collaboration; Oxford, England).

A total of 10 studies^{10111213141516171819} involving 959 patients (Table 2 ) were identified that examined the effects of systemic corticosteroids during acute COPD exacerbations. The mean age of patients of these studies was 67 years. The patients at the time of presentation to the hospital demonstrated an average arterial pH of 7.40 and Paco₂ of 42 mm Hg. Most of the patients were active smokers with a mean smoking history of 63 pack-years. The treatment was initiated in hospital in eight of the studies,^{1112131415161719} while in two studies¹⁰¹⁸ systemic corticosteroids were administered in an outpatient setting. In half of the studies,^1112141617 investigators used parenteral methylprednisolone or hydrocortisone, while in the other half of the studies,^1013151819 oral prednisone or prednisolone was used.

Six of the 10 studies^101213151618 (involving 742 patients) examined the effects of systemic corticosteroids on treatment failure defined as either clinical deterioration, withdrawal from the study due to unsatisfactory clinical improvement, or relapse of exacerbation symptoms during the follow-up period. The follow-up period varied from 10 to 30 days. Overall, the treatment failure rate was reduced by 46% with the use of systemic corticosteroids compared with placebo during acute exacerbations (RR, 0.54; 95% CI, 0.41 to 0.71; p = 0.27 for heterogeneity) [Fig 1 ]. The study by Emerman et al¹⁴ was excluded from the analysis because patients were administered just one dose of IV methylprednisolone and were followed up for just 48 h. The method of administration (oral or parenteral) did not modify the beneficial effects of systemic corticosteroids on treatment failures.

Systemic corticosteroids also had beneficial effects in reducing the length of hospitalization. Mean LOS was reduced by a weighted mean of 1.42 days with systemic corticosteroids (95% CI, 0.65 to 2.18; p = 0.40 for heterogeneity) [Fig 2 ].

Information on adverse drug effects including heartburn, GI bleeding, hyperglycemia, infection, psychomotor disturbance, and weight gain was variably (and scarcely) reported. The only potential adverse effect that was consistently reported in most of the studies was hyperglycemia. The risk of hyperglycemia was significantly increased with corticosteroid treatment (RR, 5.88; 95% CI, 2.40 to 14.41).¹³¹⁵¹⁶ There was insufficient information on the effect of corticosteroids on mortality.

There have been several placebo-controlled studies^{2021222324252627282930} that have examined the impact of antibiotics on clinically important outcomes during COPD exacerbations (Table 3 ). Of the 11 eligible studies, only 3 placebo-controlled studies²³²⁴²⁷ have been performed since 1987, likely reflecting the acceptance of antibiotics as standard of care in the management of COPD exacerbations over the past 2 decades. Three of the studies²⁰²²²⁴ were conducted in outpatients, seven studies^{21232526282930} were conducted in patients in medical wards, and one study²⁷ was conducted in a medical ICU. The mean age of the patients in these studies was 63 years. Minimum and mean durations of treatment were 7 days and 8.9 days, respectively. The antibiotics most commonly used were oral β-lactams (43%) and tetracycline derivatives (29%).

Compared with placebo, antibiotic use during COPD exacerbations (five studies^2021242730 involving 557 patients) reduced treatment failures, defined as requiring additional antibiotics within the first 7 days or unchanged or deteriorated symptoms within 21 days, by 46% (RR, 0.54; 95% CI, 0.32 to 0.92) [Fig 3 ]. However, there was significant heterogeneity of findings across the studies (p = 0.002). Stratification of studies according to patient type (in-hospital vs an outpatient setting) attenuated the heterogeneity. Antibiotics significantly reduced treatment failures when they were administered to patients who were hospitalized (for three inpatient studies²¹²⁷³⁰: RR, 0.34; 95% CI, 0.20 to 0.56; p = 0.48 for heterogeneity) but not when they were used in ambulatory patients (for two outpatient studies²⁰²⁴: RR, 0.88; 95% CI, 0.56 to 1.39; p = 0.06 for heterogeneity). Three clinical trials²¹²⁷³⁰ involving 181 patients demonstrated that in-hospital mortality can be reduced by 78% with the use of antibiotics during acute COPD hospitalizations (RR, 0.22; 95% CI, 0.08 to 0.62; p = 0.92 for heterogeneity). Some studies²⁵²⁷ reported the effect of antibiotics on short-term lung function, blood gas measurements, or length of stay in hospital, but antibiotics did not materially affect these outcomes compared with placebo.

Fourteen studies compared the use of NPPV to standard therapy in the management of acute COPD exacerbations (Table 4 ). All of the studies were performed since 1993.^{3132333435363738394041424344} Mean age of the patients in these studies was 67 years. Arterial blood gas measurements on study entry for included patients revealed a mean pH of 7.31 and Paco₂ of 68 mm Hg. In most cases, investigators used bilevel positive airway pressure (BPAP) administered through a ventilator via nasal or face mask. NPPV was initiated as early as possible following hospitalization on a general medical or respiratory ward, or an ICU. The mean duration of NPPV use was 8.5 h/d, with a range of 6 to 14 h/d for a mean duration of 4.3 days (range, 3 to 10 days). NPPV was continued as long as necessary to avoid endotracheal intubation. Most studies had prespecified clinical criteria for endotracheal intubation based on a set of clinical parameters and arterial blood gas measurements obtained serially during follow-up.

The totality of randomized controlled trials demonstrates that NPPV reduced the need for intubation, improved the risk of in-hospital mortality, and shortened hospital stays during acute COPD exacerbations. In the 12 controlled randomized trials (959 patients), NPPV reduced the need for intubation by 65% (RR, 0.35; 95% CI, 0.26 to 0.47; p = 0.82 for heterogeneity) [Fig 4 ].^{31323435363738394041424344} The benefits were modified by the average pH of the study participants. The beneficial effects of NPPV increased as the baseline pH decreased (p = 0.047) [Fig 5 ].

We found 11 studies^{313234353638394041424344} involving 940 patients that evaluated the effects of NPPV on in-hospital mortality. Overall, compared with placebo, the in-hospital mortality rate was reduced by 55% (RR, 0.45; 95% CI, 0.30 to 0.66; p = 0.99 for heterogeneity) [Fig 6 ]. Although there was significant heterogeneity in the data (p = 0.005 for heterogeneity), the length of hospital stay was significantly shortened by a weighted mean of 1.94 days with the use of NPPV (95% CI, 0.01 to 3.87) [Fig 7 ].

Acute exacerbations of COPD are an increasing cause of morbidity, mortality, and economic burden in the United States and elsewhere. Despite the widespread promulgation of clinical guidelines by various expert committees and professional societies over the past decade, there continues to be considerable heterogeneity in the way in which acute exacerbations are managed between physicians.³¹ For instance, a large survey³¹ found that < 3% of patients with acute COPD exacerbations were treated with NPPV and < 85% were treated with antibiotics and systemic corticosteroids, which are thought to be cornerstones of inpatient management. The primary objective of this study³¹ was to determine the strength of clinical data that support the use of these therapies on clinically relevant health outcomes such as treatment failures, the need for intubation, in-hospital mortality, and LOS. The present study builds on previous systematic reviews and metaanalysis by adding data from most recently published trials by integrating the findings on these three therapies in one succinct report (as opposed to previous published analyses, which have evaluated these interventions separately, which we believe will enhance the translation of these findings into clinical practice),³² and by presenting “new” analyses to address clinical issues that have not been previously evaluated.

GOLD guidelines recommend systemic corticosteroids for in-hospital management of COPD exacerbations, while antibiotics are recommended for any exacerbations (regardless of severity) that lead to increased dyspnea, sputum volume, and sputum purulence. NPPV is believed to be beneficial only in exacerbations that are associated with moderate-to-severe respiratory acidosis.⁵ The findings of this review largely support these recommendations with some notable exceptions. Firstly, we found that systemic corticosteroids reduced treatment failure by 46% during both inpatient and outpatient management of COPD exacerbations, which are similar to the findings by Wood-Baker and colleagues.³³ Secondly, we found that antibiotics reduced treatment failures by 46% and improved survival of hospitalized patients. The survival effect was observed only among hospitalized patients. It should be noted, however, that although the combined analysis showed a survival benefit, each individual study was relatively small in size and was underpowered to answer this critical question. Even in the treatment failure data, there was significant heterogeneity in the findings across the studies indicating lack of consistency and robustness in the results. In the future, large clinical trials are needed to clearly determine the role of antibiotics in the management of COPD exacerbations. Until then, the current evidence suggests that antibiotics might be a reasonable choice for severe exacerbations requiring hospitalization or exacerbations that fail to improve despite systemic corticosteroid therapy.³⁴ Thirdly, we found that NPPV reduced the need for endotracheal intubation in patients with severe exacerbations and demonstrated arterial (respiratory) acidosis. The beneficial effect of NPPV was modified by the arterial pH of the study patients. At pH of 7.26, for instance, the risk of intubation in the standard group was nearly 60%; whereas in the NPPV group, the risk was only 20%. At higher pH values, the risk differential was much smaller. Thus, our findings support the recommendations of the GOLD committee that NPPV may be used for severe exacerbations when the pH is < 7.35. Our findings are also consistent with three systematic reviews³⁵³⁶³⁷ that evaluated the role of NPPV for acute exacerbations of COPD.

In view of the high mortality risk associated with COPD exacerbations, mortality reduction is one of the major aims of hospital-based management of acute COPD. We found that both antibiotics and NPPV reduced in-hospital mortality in a selected group of patients but the mechanisms responsible for their survival benefits were uncertain. It is likely that antibiotics reduced mortality by reducing treatment failure and by possibly preventing nosocomial infections during hospitalizations especially in patients who require endotracheal intubation. The risk of developing ventilator-associated pneumonia can be as high as 30%, and the case fatality rate associated with such infections can be > 50%.³⁸ A metaanalysis³⁹ involving 36 trials and 6,922 patients demonstrated a 22% reduction in mortality with the administration of antibiotic prophylaxis in patients receiving ventilation. The survival benefit from NPPV, however, may be related in part to a 65% reduction in the need for endotracheal intubation and invasive mechanical ventilation and its associated morbidity and mortality.⁴⁰ The impact of systemic corticosteroids on mortality is uncertain. Only four small studies^10111213 (involving 360 patients) reported on mortality, and the event rates were too small (total of 10 deaths) to draw any meaningful conclusions.

Both systemic corticosteroids and NPPV can reduce LOS by a weighted mean of 1.42 days and 1.94 days, respectively. Only two studies²⁵²⁷ analyzed the impact of antibiotics on LOS, and the results were conflicting. More studies on antibiotics are needed to determine their effect on LOS. Reducing LOS is an important goal of acute COPD management because hospital stays are associated with morbidity and with increased costs. An average hospital day costs the system approximately $600.⁴¹ Although we found that antibiotics improved clinical outcomes in patients who require hospital-based care, we did not observe a significant difference in the clinical benefits among different classes of antibiotics. In the study by Anthonisen et al,²⁰ for instance, there were no material differences in clinical efficacy between trimethoprim-sulfamethoxazole, amoxicillin, or doxycycline. Several other studies⁴² comparing the various antibiotic agents head-to-head have also failed to demonstrate clear superiority of newer agents over older classes of respiratory antibiotics. Antibiotic selection should therefore be guided by recent history of antibiotic use and local microbial resistance patterns.

Another controversial area is the optimal dose and duration of systemic corticosteroid treatment. In the studies^101213151618 showing clinical benefits, the corticosteroid dose (expressed in oral prednisone equivalents) varied from 0.5 to 1.0 mg/kg/d, and the treatment duration varied from 8 to 15 days. Extending the duration of therapy beyond 2 weeks does not appear to confer additional benefits. Niewoehner et al,¹⁶ for instance, failed to demonstrate additional benefits from extending systemic corticosteroid therapy from 2 to 8 weeks. Despite the numerous adverse drug effects associated with corticosteroid use, the only short-term adverse effect reported from the included studies was a sixfold increase in the risk of hyperglycemia.

There were several important limitations to this systematic review. Firstly, the definition of an acute exacerbation was based on clinical criteria and not based on more objective measures such as lung function, which may have resulted in some diagnostic misclassification. This may have in certain circumstances (eg, misclassification with asthma) led to a slight overestimation of the benefit of systemic corticosteroids. Secondly, not all studies included in the analysis of NPPV had objective criteria for intubation. Since it was not possible to blind the treating physicians and other health-care professionals, patients in the standard medical therapy group may have been intubated sooner than the NPPV group, resulting in a higher “need for intubation” rate. However, the overall mortality benefit of NPPV would unlikely to have been influenced by this bias. Thirdly, although antibiotics resulted in a larger reduction in treatment failure rates among hospitalized compared with ambulatory patients, the exact severity or clinical characteristics of patients who benefited from antibiotic therapy is uncertain since there was a scarcity of reported data on lung function and blood gases in these studies. Fourthly, although there were many short-term clinical benefits of systemic corticosteroids, antibiotics and NPPV, the long-term effects of these interventions remain uncertain. Fifthly, publication bias is of concern. To mitigate this bias, we carefully searched the bibliographies of salient articles and included studies that were published only in the abstract form. Finally, owing to a variety of metholdogic and clinical issues including heterogeneity of trial design, definitions of exacerbations and underlying clinical characteristics of study patients, the risk estimates of the interventions are not directly comparable.

In summary, acute COPD exacerbations may be treated effectively with systemic corticosteroids, antibiotics, and NPPV. Systemic corticosteroids reduce treatment failures in both in-hospital and outpatient settings, while antibiotics appear more effective in patients requiring hospitalization. NPPV should be considered for carefully selected patients who demonstrate arterial respiratory acidosis.