This review updates a report prepared for the Agency for Healthcare Research and Quality (AHRQ) and serves as the background paper for an American College of Physician's Clinical Practice Guideline (4). It addresses the following questions: Which inhaled therapies are effective for treatment and maintenance of stable COPD? When should clinicians consider pulmonary rehabilitation and disease management? When should clinicians prescribe oxygen therapy? Should clinicians base treatment decisions on spirometric results, symptoms, or both?

Detailed information on the use of spirometry for diagnosis and case finding is available in the original AHRQ report at http://www.ahrq.gov/clinic/tp/spirotp.htm. Spirometry for case finding and management would be useful if it identified individuals who were not clinically detected as candidates for COPD treatments, excluded individuals with false-positive clinical presentations for COPD, or independently identified thresholds to guide initiation or modification of therapies. Our previous report identified insufficient evidence to support these conditions.

Methods

Top Methods Results Discussion Author & Article Info References

Data Sources and Selection

For our previous report, we searched PubMed and the Cochrane Library for articles published in English from 1966 through May 2005. The current review extends the search related to COPD therapies through March 2007 by using search terms used in a 2003 review by Sin and colleagues (5) to identify randomized, controlled trials (RCTs), controlled clinical trials, meta-analyses, and reviews published since the completion of their search in 2002. To supplement our search, we examined the Cochrane Database of Systematic Reviews of Effectiveness, examined bibliographies of published articles, and contacted experts. We categorized interventions as 1) inhaled medications (ß₂-agonists, anticholinergics, combination ß₂-agonists and anticholinergics, inhaled corticosteroids, and combination inhaled corticosteroids and long-acting ß₂-agonists or anticholinergics), 2) pulmonary rehabilitation, 3) disease management programs, and 4) oxygen therapy.

Two reviewers used standardized data abstraction sheets to examine titles and abstracts of newly identified references. If both reviewers agreed on eligibility, we included the article. Disagreement among reviewers, although rare, was resolved by discussion, with final decision by the lead author. Trials were eligible if they were randomized; involved persons with COPD that was defined clinically or by spirometry; and measured clinical outcomes, including exacerbations, standardized respiratory health status measures, hospitalizations, and deaths. Studies reporting only spirometry outcomes were ineligible. Inhaled therapy trials had to include 50 or more participants per treatment group and at least 3 months of follow-up. Trials of pulmonary rehabilitation programs had to include at least 6 weeks of follow-up and a usual care comparison group. We excluded studies that compared different types of pulmonary rehabilitation, and we included systematic reviews and meta-analyses of COPD therapies.

Data Extraction

Two individuals extracted data onto standardized forms. The lead author resolved any disagreements. Main outcomes for all interventions were the percentage of participants experiencing at least 1 exacerbation, mean change in respiratory health status, hospitalization, and death. Respiratory health status was assessed by the validated St. George Respiratory Questionnaire (SGRQ) or the Chronic Respiratory Disease Questionnaire (CRDQ). A 4-unit reduction (out of 100) on the SGRQ and a 0.5-unit increase per question on the 7-question CRDQ are defined as clinically noticeable improvements (6). For pulmonary rehabilitation, we collected information on the 6-minute walk test and defined a minimally clinically significant effect size as 53 meters or more.

We collected data on adverse effects of long-acting inhaled therapies (including specifically described adverse effects, "serious adverse effects," treatment adherence, study withdrawals, and withdrawals due to adverse effects) from trials that lasted at least 1 year and from systematic reviews that specifically addressed adverse effects. We assessed whether these studies used placebo or active control run-in periods, as well as the number and reasons for exclusion of potentially eligible patients from randomization during the run-in period.

Study Quality Assessment

We used the methods of Schulz and colleagues (7) to assess the quality of randomized trials on the basis of allocation concealment. We assessed blinding, intention-to-treat analysis, length of follow-up, withdrawals or loss to follow-up, and funding source. We rated the quality of systematic reviews or meta-analysis according to the Strength of Recommendation Taxonomy (8). An RCT was considered high quality if it had allocation concealment, blinding (if possible), intention-to-treat analysis, adequate size, and adequate follow-up (>80%). Systematic reviews or meta-analysis with high-quality studies and consistent findings are indicated as good-quality, patient-oriented evidence.

Data Synthesis and Analysis

Intervention effectiveness was described according to baseline respiratory symptom status, spirometrically defined level of airflow obstruction, acute change in spirometry, or spirometric change over time (inhaled medications and use of spirometry to guide therapy). The magnitude of effect across interventions (inhaled therapies and oxygen) was based on relative risks and absolute risk differences, as well as comparison with previously determined, minimally important clinical differences in respiratory health status and exercise capacity. Study results were combined, if appropriate, to produce pooled estimates. We calculated relative risks and 95% CIs for categorical variables and weighted mean differences and 95% CIs for continuous variables. We conducted analyses by using a DerSimonian–Laird random-effects model in Review Manager software, version 4.2 (The Cochrane Collaboration, Oxford, United Kingdom) (9). We assessed heterogeneity by using a chi-square test and the I² test. An I²statistic of 50 or greater indicates substantial heterogeneity (10). If heterogeneity existed, we conducted sensitivity analyses to explore potential causes of heterogeneity.

Role of the Funding Source

This project was funded by the AHRQ, U.S. Department of Health and Human Services. The updated synthesis was conducted in collaboration with the American College of Physicians' Clinical Efficacy Assessment Subcommittee. Panel members assisted in the formulation of questions and reviewing drafts of this report. The funding source had no role in the design, conduct, or reporting of the study or in the decision to submit the manuscript for publication.

Results

Top Methods Results Discussion Author & Article Info References

Yield of the Literature Search

Figure 1 shows that 42 RCTs involving short- or long-acting inhaled monotherapy or combination therapy (ipratropium [11–17], tiotropium [14, 15, 18–25]), long-acting ß₂-agonists (11, 13, 14, 17, 18, 21, 26–41), corticosteroids (28, 29, 32, 33, 38–47), dual D₂ dopamine receptor–ß₂-agonist (sibenadet) (30, 48, 49), short-acting ß₂-agonists, and ipratropium (50–52) versus placebo or active control and 8 meta-analyses of RCTs (5, 53–59) were included for assessment of COPD inhaled therapies. We have identified 10 RCTs and 5 systematic reviews since our AHRQ report. Our updated search yielded an additional 16 RCTs and 2 systematic reviews of nonpharmacologic treatments. Three systematic reviews of 39 unique RCTs and 6 additional RCTs evaluated pulmonary rehabilitation (6 RCTs and 1 systematic review were added for our review) (5, 60–90). Two systematic reviews of 13 unique RCTs and 2 additional trials evaluating disease management, education, and follow-up were eligible (2 RCTs and 1 systematic review were added for our review) (5, 91–106). Supplemental oxygen therapy was not addressed in our original report. We included 8 RCTs and 1 systematic review evaluating 7 of these 8 trials (5, 107–114).

View larger version (67K): [in this window] [in a new window]   Figure 1. Data search and selection. RCT = randomized, controlled trial.

Quality Assessment

Appendix Table 1 and other systematic reviews (5, 60, 61, 91) describe the included randomized trials. We identified no study quality differences according to type of inhaled medication.

View this table: [in this window] [in a new window]   Appendix Table 1. Treatments, Baseline Characteristics, and Study Quality of Individual Trials of Treatments for Chronic Obstructive Pulmonary Disease

Concealment of treatment allocation for inhaled therapies was adequate in 17 studies (12, 22, 25, 26, 29–31, 36–40, 42, 44, 46, 47, 49). All trials were double-blind, and nearly all used intention-to-treat analyses. Several included only participants who were taking at least 1 dose or who had 1 valid postbaseline measurement (17, 23, 30, 32, 38, 42, 44, 48, 49) or excluded participants because of noneligibility after randomization or good practice or ethics violations by individual study sites (26, 39). All but 7 studies were funded by pharmaceutical companies. All trials had adequate participant follow-up (>80%). Six trials lasted 3 years or longer (12, 39, 42, 44, 45, 47).

Concealment of treatment allocation for trials of pulmonary rehabilitation and oxygen therapy was adequate in 1 and 4 studies, respectively (87, 107, 109, 111, 113). One disease management trial adequately randomly assigned practice centers but did not use individual randomization (106). Among nonpharmacologic trials, 7 used intention-to-treat analysis (86, 107–112). Five trials reported double-blinding (87, 89, 111, 112, 114), and 4 trials indicated blinded assessment of outcomes (85, 89, 107, 113). Four oxygen therapy trials lasted 2 years or longer (76–79).

Almost all COPD treatment trials involved participants who were prone to exacerbations, had previous diagnoses of COPD, had disabling respiratory symptoms, had mean FEV₁ less than 50% predicted, and used inhaled therapies. Only 4 RCTs used population-based recruitment and enrolled participants similar to those likely to be identified by spirometric case finding of "at-risk" individuals (11, 42, 44, 46), although some trials provided additional analysis according to spirometric status.

The Table summarizes the strength of the evidence for each question addressed in our review.

Which Inhaled Therapies Are Effective for Treatment and Maintenance of Stable COPD?

Exacerbations

Monotherapies with inhaled long acting ß₂-agonists, a long-acting anticholinergic, or corticosteroids were of similar effectiveness and were superior to placebo or short-acting anticholinergics in reducing exacerbations (Figures 2 and 3). Compared with placebo, inhaled corticosteroids, long-acting bronchodilators (tiotropium, ß₂-agonists), or both reduced the relative risk for having at least 1 exacerbation by 13% to 17% and the absolute risk by 4% to 6%. Ipratropium, a short-acting anticholinergic, was not superior to placebo. In active comparator studies, long-acting ß₂-agonists were of similar effectiveness to corticosteroids or the short- or long-acting anticholinergics, ipratropium, or tiotropium.

View larger version (28K): [in this window] [in a new window]   Figure 2. Number of participants who had at least 1 exacerbation: inhalation treatments versus placebo. LABA = long-acting ß2-agonist; RR = relative risk.

View larger version (37K): [in this window] [in a new window]   Figure 3. Number of participants who had at least 1 exacerbation: inhalation treatments versus active control. LABA = long-acting ß2-agonist; NA = not applicable; RR = relative risk; SABA = short-acting ß2-agonist.

The incremental effect of combination therapy with inhaled corticosteroids and long-acting ß₂-agonists versus monotherapy using these agents was of borderline statistical significance, as assessed in 6 multigroup trials lasting 6 to 36 months (mean baseline FEV₁ <50%) (Figures 2 and 3). The pooled absolute risk differences in the percentage of participants having at least 1 exacerbation for long-acting ß₂-agonists, corticosteroids, and combination therapy were –4% [95% CI, –8% to –1%], –5% [CI, –11% to 1%], and –6% [CI, –12% to –1%], respectively, compared with placebo (28, 29, 34, 41). Combination therapy did not reduce the value compared with monotherapy with either inhaled corticosteroids or long-acting ß₂-agonists (relative risk, 0.88 [CI, 0.75 to 1.17] vs. ß₂-agonists and 0.96 [CI, 0.85 to 1.08] vs. inhaled corticosteroids) (28, 29, 34, 41). A large 3-year RCT (TORCH [Towards a Revolution in COPD Health] [39]) of combination long-acting ß₂-agonist plus inhaled corticosteroid (fluticasone, 500 µg twice daily) versus placebo, long-acting ß₂-agonist, or inhaled corticosteroid monotherapy evaluated the annual rate of moderate to severe exacerbations in symptomatic adults with severe airflow obstruction. Pooling these results was not possible because the study (39) reported only annual rates of exacerbations (rather than proportions). The study investigators observed a statistically significant relative risk reduction of nearly identical magnitude to our pooled findings (relative risk, 0.75 [CI, 0.69 to 0.81] vs. placebo; 0.88 [CI, 0.81 to 0.95] vs. ß₂-agonists; and 0.91 [CI, 0.84 to 0.99] vs. inhaled corticosteroids). However, another trial found no difference in the annual rate of moderate to severe exacerbations or time to first exacerbation (P = 0.15) regardless of baseline FEV₁ among participants randomly assigned to continue combination therapy with salmeterol–fluticasone compared with those in whom fluticasone therapy (500 µg twice daily) was withdrawn (40).

One 3-group trial lasting for 1 year evaluated combination therapy with all 3 classes of inhalers. The proportion of participants who experienced an exacerbation did not differ among those receiving monotherapy with a long-acting anticholinergic (tiotropium) (62.8%), those receiving combination tiotropium plus a long-acting ß₂-agonist (salmeterol) (64.8%), or those receiving all 3 therapies (tiotropium plus corticosteroid plus a long-acting ß₂-agonist [salmeterol–fluticasone]) (60.0%) (25) (Figure 3). The combination of a short-acting ß₂-agonist (albuterol) plus ipratropium reduced exacerbations compared with albuterol alone (absolute risk difference, –6%) (50–52).

Respiratory Health Status Measures and Hospitalizations

Twenty trials, including the largest (38), reported SGRQ or CRDQ outcomes, but published results often did not permit pooling. Except for 5 trials (11, 25, 28, 33, 36), the average improvement in health status because of monotherapy or combination therapy was not considered clinically significant (6) (Appendix Table 2). In secondary analyses of 2 trials of tiotropium (18, 19), the percentage of individuals achieving a clinically significant difference in the SGRQ was greater with tiotropium than with placebo (49% vs. 35%).

View this table: [in this window] [in a new window]   Appendix Table 2. Inhaled treatments for Chronic Obstructive Pulmonary Disease versus Placebo and Control: Improvement in the St. George Respiratory Questionnaire or Chronic Respiratory Disease Questionnaire at Study End Point

Few RCTs reported hospitalization results. When reported, reductions were not consistently observed and do not permit definitive conclusions on the relative effectiveness of inhaled therapies. Monotherapy with a long-acting ß₂-agonist and combination therapy of long-acting ß₂-agonists and corticosteroids reduced the relative annual rate of severe exacerbations requiring hospitalizations by 17% and 18%, respectively, versus placebo in the TORCH study (39). The 12% relative reduction with inhaled corticosteroids did not achieve statistical significance (rate ratio, 0.88 [CI, 0.74 to 1.03]). Combination therapy was not superior to ß₂-agonists (rate ratio, 1.02 [CI, 0.87 to 1.20]) or inhaled corticosteroids (rate ratio, 0.95 [CI, 0.82 to 1.12]) used as monotherapy. Combination therapy with tiotropium plus salmeterol–fluticasone (but not tiotropium plus salmeterol) reduced hospitalizations for acute COPD exacerbations (rate ratio, 0.53 [CI, 0.33 to 0.86]) and all-cause hospitalizations versus tiotropium alone (25). Three trials lasting 3 to 12 months of long-acting ß₂-agonist therapy in participants with a mean FEV₁ less than 60% predicted demonstrated a 2% reduction (CI, –5% to 1%) compared with placebo that was not statistically significant (11, 18, 34). The Lung Health Study (LHS) I and II enrolled persons with mild to moderate airflow obstruction (mean FEV₁, 75% and 64% predicted, respectively; trial duration, 5 years) (12, 43). The LHS I showed no statistically significant differences in hospitalizations per 100 person-years of exposure between ipratropium and placebo (12). In LHS II, inhaled corticosteroids resulted in a small and nonsignificant decrease in hospitalizations per 100 person-years of exposure for respiratory conditions (P = 0.07) and no difference in nonrespiratory hospitalizations (43). The proportion of participants requiring hospitalization for COPD was lower with tiotropium than with placebo (absolute risk difference, –2% [CI, –4% to –1%]) (18, 19, 22, 24) and with ipratropium (absolute risk difference, –4% [CI, –10% to 1%]) (mean FEV₁ <60%; trial duration, 6 months to 1 year) (14).

Deaths

Death was the primary end point in only 1 trial (39). Mortality rates did not statistically differ in any trial or in pooled analyses of monotherapies (Figure 4). In a retrospective individual-patient data meta-analysis published before the TORCH study (56), inhaled corticosteroids resulted in a 1% absolute reduction in all-cause mortality compared with placebo (hazard ratio, 0.75 [CI, 0.57 to 0.99]). The mortality rate was not reduced among participants with a baseline FEV₁ of 60% predicted or more (hazard ratio, 0.90 [CI, 0.54 to 1.53]). Combination therapy with long-acting ß₂-agonists plus inhaled corticosteroids reduced the relative but not the absolute risk for death compared with placebo (relative risk, 0.82 [CI, 0.69 to 0.98]; absolute risk difference, –0.01 [CI, –0.03 to 0.01]) and inhaled corticosteroids (relative risk, 0.79 [CI, 0.67 to 0.94]; absolute risk difference, –0.01 [CI, –0.03 to 0.02]). Neither the relative nor the absolute risk for death improved with combination long-acting ß₂-agonists plus inhaled corticosteroids compared with long-acting ß₂-agonists (relative risk, 0.90 [CI, 0.76 to 1.08].

View larger version (41K): [in this window] [in a new window]   Figure 4. Mortality: inhalation treatments versus placebo or combination long-acting ß2-agonists and corticosteroid therapy versus monotherapy. LABA = long-acting ß2-agonist; LHS = Lung Health Study; RR = relative risk.

Withdrawals and Adverse Events

Appendix Table 3 shows withdrawals and adverse events with long-acting inhaled therapies compared with placebo. An additional active comparator study evaluated combining the long-acting anticholinergic tiotropium with the ß₂-agonists salmeterol or salmeterol–fluticasone versus tiotropium alone (25). All but the tiotropium combination study (25) used a run-in period before randomization of initially eligible participants. The duration (10 days to 3 months), interventions allowed or provided (placebo, study drug, nonstudy chronic COPD medications, or rescue therapies), and reasons for exclusion (adherence, adverse events, and additional eligibility criteria) varied across studies. The mean percentage of persons who were enrolled in the run-in period but were not subsequently randomly assigned was 23% and ranged from 10% to 29% in the 12 trials that reported this information (25, 28, 29, 35, 36, 39–43, 45, 47). In the 7 trials that reported reasons for exclusions, 19% were mainly due to adverse events, followed by inadequate adherence to run-in medications (28, 36, 39–42, 47). None of the trials adequately described how the cause, severity, or duration of an adverse event was assessed, with the exception of fractures. Inconsistencies in adverse events reporting limited quantitative synthesis.

View this table: [in this window] [in a new window]   Appendix Table 3. Study Withdrawals and Adverse Effects for Trials Lasting 1 Year or More: Inhaled Treatment versus Placebo

"All study withdrawals" occurred less frequently among persons randomly assigned to tiotropium (21%) (19, 24), long-acting ß₂-agonists (33%) (28, 29, 35, 36, 39, 41), corticosteroids (31%) (28, 29, 35, 36, 39, 41), or combination long-acting ß₂-agonists plus corticosteroids (32%) compared with placebo (28% to 44%). All study withdrawals were less likely to occur with combination therapy than with long-acting ß₂-agonist monotherapy (32% vs. 37%; relative risk, 0.82 [CI, 0.71 to 0.96]) or corticosteroid monotherapy (32% vs. 37%; relative risk, 0.87 [CI, 0.80 to 0.94]) (28, 29, 39, 41). Fewer withdrawals occurred with the combination of all 3 classes of long-acting inhaled agents (anticholinergics, ß₂-agonists, and corticosteroids) versus long-acting anticholinergic monotherapy (relative risk, 0.54 [CI, 0.30 to 0.96]) (25). "Withdrawals due to adverse effects" were similar or lower with inhaled therapies than with placebo. About 50% of enrollees remained adherent to therapy as prescribed. Adverse events during follow-up were usually minor and were seldom more than with placebo. "Serious adverse events" did not statistically significantly differ with inhaled treatment used as monotherapy or in combination therapy versus placebo. "Serious adverse events" occurred in 10% of participants receiving inhaled corticosteroids as monotherapy or combination therapy in the TORCH trial compared with 6% of participants receiving placebo or long-acting ß₂-agonists (39). Compared with placebo, adverse events that were considered to be related to treatment were more common with tiotropium and corticosteroids but not with long-acting ß₂-agonists. The frequencies of serious adverse events did not differ between combination therapy and long-acting ß₂-agonists or corticosteroids used as monotherapy (28, 39, 40).

The most common specific adverse effects of tiotropium were dry mouth, occurring in 10.3% of participants (relative risk, 4.4 [CI, 2.2 to 8.8] vs. placebo) (19, 24), and urine retention (odds ratio, 2.5 [CI, 0.5 to 14] vs. placebo) (53). Respiratory infections and pneumonia were similar with long-acting ß₂-agonists and with placebo (28, 35, 36, 38). A meta-analysis of 20 RCTs assessed the cardiovascular effects of inhaled ß₂-agonists (primarily salmeterol and formoterol) in patients with asthma or COPD. ß₂-Agonists were associated with an increase in cardiovascular events compared with placebo (2.7% vs. 0.7%) (56). Of these events, 87% were due to sinus tachycardia. Major cardiovascular events were higher compared with placebo, although they did not statistically differ (relative risk, 1.66 [CI, 0.76 to 3.60]). Another pooled analysis concluded that respiratory deaths increased with long-acting ß₂-agonists and decreased with anticholinergics (59). However, their conclusions were based on very few events; were not verified in our review of the published primary literature; included findings from duplicate publications; and did not incorporate the TORCH study, which found no difference in deaths due to pulmonary causes between placebo and salmeterol (5% in each group) (39).

Three trials provided information about the risk for pneumonia with inhaled corticosteroid use lasting up to 3 years. Pooled analysis showed significant heterogeneity (P = 0.02; I² = 74%), which disappeared (P = 0.56; I² = 0%) when the smallest trial that enrolled younger patients with mild airflow obstruction was excluded. In 2 trials, inhaled corticosteroids were associated with an increased risk for pneumonia compared with placebo (relative risk, 1.55 [CI, 1.33 to 1.80]). Inhaled corticosteroids were associated with an increased frequency of oropharyngeal candidiasis (28, 29, 42, 43, 45), throat irritation (28, 29, 42, 45), and a moderate to severe degree of easy bruising (29, 42, 45). After 3 years, lumbar spine and femur bone mineral density were lower in the LHS II triamcinolone group (43), but not in a small subset evaluated in TORCH (39). Pooled results from 3 RCTs indicated that fracture incidence was similar for inhaled corticosteroids used alone or in combination with long-acting ß₂-agonists for up to 3 years versus placebo (pooled relative risk, 0.96 [CI, 0.55 to 1.68]) (39, 42, 45). In the trial evaluating all 3 classes of long-acting inhaled therapies, 47% of patients in the tiotropium plus placebo group discontinued study medications compared with 43% in the tiotropium plus salmeterol group and 26% in the tiotropium plus salmeterol–fluticasone group (P < 0.001) (25). Serious adverse events were similar across the 3 treatment groups.

When Should Clinicians Consider Pulmonary Rehabilitation and Disease Management?

Pulmonary rehabilitation but not disease management may improve health status and exercise capacity during the program in symptomatic adults with severe airflow obstruction. Conclusions based on published findings are problematic because exacerbations, hospitalizations, standardized health status measures, and exercise capacity were infrequently reported (Appendix Tables 4 and 5) (60–90). Most pulmonary rehabilitation programs contained 4 major components: endurance or exercise training, education, behavioral modification, and outcome assessment. Programs primarily emphasized endurance training and enrolled patients with severe to very severe COPD (mean FEV₁, 31% to 54% predicted). Only 6 trials identified in the systematic review by Sin and colleagues (57) reported mean differences in SGRQ scores versus controls (pooled difference, 4.4 [CI, 0.3 to 8.4]), and 3 studies observed the average improvement between control and intervention greater than the 4-point minimally important difference (62, 67, 72). The average effect for the CRDQ dyspnea subscale was clinically significant (mean difference [vs. control] ranged from 0.2 to 14), but the increase in exercise tolerance measured by distance walked in 6 minutes was less than the 53-meter threshold that was determined to be clinically significant. Pulmonary rehabilitation did not reduce deaths, although sample size and study duration were insufficient to adequately evaluate this end point (5). A review of 6 small RCTs (n = 230) found that respiratory rehabilitation after acute COPD exacerbations in patients with severe airflow obstruction (baseline FEV₁ <40% predicted) reduced hospitalizations (relative risk, 0.26 [CI, 0.12 to 0.54]; 3 trials reporting) and produced a clinically significant improvement in exercise capacity, as measured by the increased distance walked during the 6-minute walk test (64 to 215 meters; 4 trials reporting) and the SGRQ and CRDQ dyspnea subscales compared with usual care (3 trials reporting) (61).

View this table: [in this window] [in a new window]   Appendix Table 4. Summary of Outcomes for Clinical Trials of Pulmonary Rehabilitation

View this table: [in this window] [in a new window]   Appendix Table 5. Summary of Outcomes for Clinical Trials of Disease Management

Studies evaluating disease management used patient education, self-management with development of a treatment action plan, or enhanced follow-up with a respiratory health worker or pharmaceutical care coordinator (91–106). Appendix Tables 4 and 5 shows details of these studies. A total of 2911 patients with COPD were enrolled in 15 studies that lasted from 3 months to 1 year (5, 91, 105, 106). Average baseline FEV₁ was less than 50% predicted, and all patients were taking inhaled bronchodilators. The only trial reporting exacerbations noted fewer episodes in the self-management plus telephone follow-up group (92). Pooled mortality rates from trials lasting at least 9 months and providing results did not differ between intervention and control (relative risk, 0.88 [CI, 0.66 to 1.18]) (Table 4). The RCTs of brief interventions found no evidence for a reduction in all-cause readmissions, and data from long-term or more intensive intervention RCTs were equivocal about health care utilization outcomes (91). The pooled difference in SGRQ health status scores versus usual care was less than clinically noticeable (weighted mean difference, –2.5 [CI, –4.8 to –0.1]). The relative risk and number of hospital readmissions did not differ (relative risk, 0.86 [CI, 0.68 to 1.08]).

When Should Clinicians Prescribe Oxygen Therapy?

Supplemental oxygen used during most of the daytime each day reduced deaths in patients with very severe airflow obstruction and daytime hypoxemia (107–114). Four trials had follow-up of 2 to 5 years (107–110). Baseline Pao₂ ranged from 51 to 75 mm Hg. Interventions included using fixed doses of supplemental nocturnal oxygen for resting hypoxemia, titrating supplemental oxygen to maintain daytime arterial Pao₂ between 60 and 80 mm Hg, using as-needed ambulatory oxygen in addition to home oxygen, and using short-burst oxygen therapy for activity-limiting dyspnea among patients with COPD who were not hypoxemic at rest.

Exacerbations or hospitalizations were rarely reported. Supplemental oxygen used for 15 or more hours daily to maintain a Pao₂ greater than 60 mm Hg reduced deaths in 2 studies (n = 290) that enrolled persons with mean baseline FEV₁ less than 30% and mean resting Pao₂ of 55 mm Hg or less (relative risk, 0.61 [CI, 0.46 to 0.82]) (108, 109). In 2 additional trials (n = 211), supplemental oxygen (mean use, 9 to 13 hours per day) did not reduce deaths among individuals with similar spirometric values but daytime Pao₂ greater than 60 mm Hg (relative risk, 1.16 [CI, 0.85 to 1.58]) (109, 110).

Three small short-term studies assessed the effect of ambulatory oxygen on respiratory health status (111–113). Mean changes in CRDQ scores and exercise tolerance did not achieve clinically detectable improvement. The number of hospitalizations over 6 to 12 months and urgent care visits did not differ among cylinder oxygen (mean, 2.2 hospitalizations [SD, 2.4]), cylinder air (mean, 1.8 hospitalizations [SD, 1.5]), and usual care (mean, 1.4 hospitalizations [SD, 1.0]) (112).

Should Clinicians Base Treatment Decisions on Spirometric Results, Symptoms, or Both?

Evidence of intervention effectiveness was limited to individuals with both bothersome respiratory symptoms (especially dyspnea and frequent exacerbations) and an FEV₁ less than 60% predicted. Almost all treatment trials enrolled participants with symptomatic COPD who were prone to exacerbations and had a mean FEV₁ less than 50% predicted. No data were available to determine whether long-acting ß₂-agonists were effective in symptomatic individuals with FEV₁ greater than 60% or prevented symptoms among asymptomatic individuals.

No treatment trial evaluated modifying therapy, instituting combination inhaled therapy, or monitoring disease status according to spirometric results. However, these are unlikely to be beneficial because earlier findings (http://www.ahrq.gov/clinic/tp/spirotp.htm) demonstrated that 1) clinical improvement is not closely associated with an individual's spirometric response to therapy; 2) treatments other than smoking cessation provide only a small change in long-term decline in lung function; 3) wide intraindividual variation exists in spirometric decline; 4) higher doses of inhaled therapies have not been shown to provide clinically significant improvement compared with lower doses; 5) combination therapy provided little to no benefit compared with monotherapy; and 6) interventions were not effective in asymptomatic persons.

Discussion

Top Methods Results Discussion Author & Article Info References

Current evidence suggests that COPD treatment benefits are primarily related to reduced exacerbations among exacerbation-prone adults with activity-limiting dyspnea and FEV₁ less than 60% predicted. Inhaled corticosteroids and long-acting bronchodilators seem to be of similar effectiveness in reducing exacerbations compared with short-acting bronchodilators, but they differ in their adverse effects. Evidence indicates that average improvement in respiratory health status is clinically insignificant, but some individuals achieve a noticeable improvement. Mortality reduction occurs with long-term supplemental oxygen in symptomatic patients with severe airflow obstruction and resting hypoxemia. Studies of oxygen inconsistently reported other outcomes. When reported, treatment-related improvements were typically small. Studies of pulmonary rehabilitation showed improvements in health status and dyspnea but not in walking distance during the program. Neither disease management nor ambulatory oxygen seem to have benefits.

Combination therapy with inhaled corticosteroids and long-acting ß₂-agonists was of borderline statistical significance in reducing exacerbations and improving health status compared with monotherapy. Compared with long-acting ß₂-agonists alone, combination therapy did not reduce mortality. Compared with corticosteroids alone, combination therapy produced a 1% to 2% absolute mortality benefit that was of borderline statistical significance. Reductions in hospitalizations versus long-acting monotherapies were generally small and were not consistently observed. Health status improvements were generally not clinically significant. Tiotropium, added to a long-acting ß₂-agonist or corticosteroid plus long-acting ß₂-agonist, did not reduce exacerbations or improve dyspnea versus tiotropium monotherapy (25).

Adverse effects of long-acting inhaled therapies were usually mild, although pneumonia may be more common with inhaled corticosteroids. There was no association with fractures, but trials were short in duration. Most trials used a treatment run-in period and enrolled exacerbation-prone persons who were previously receiving and tolerating long-acting inhaled therapy. Consequently, adverse effects, treatment adherence, and effectiveness may be different in clinical practice than in published trials. All-cause withdrawals and withdrawals due to adverse effects were fewer with long-acting inhaled therapies and combination therapies than with placebo and monotherapies, respectively, suggesting that the perceived benefits of long-acting inhalers outweigh harms.

In adults with mild to moderate airflow obstruction who did not report respiratory symptoms, treatment with ipratropium did not prevent symptom development. No studies evaluated treatment of asymptomatic individuals with severe airflow obstruction. Among symptomatic participants with FEV₁ greater than 50% but less than 80% or those with normal airflow but having chronic sputum production, 7 large studies of inhaled corticosteroids or anticholinergics that lasted at least 1 year found little to no improvement in exacerbations, health status, hospitalizations, or deaths (12, 29, 40, 43, 45, 47; http://www.ahrq.gov/clinic/tp/spirotp.htm).

Respiratory symptoms are common, clinical examination has poor accuracy for determining airflow obstruction severity (http://www.ahrq.gov/clinic/tp/spirotp.htm), and few adults have airflow obstruction severe enough that treatments have demonstrated effectiveness. Therefore, adopting a strategy that targets use of long-acting inhaled corticosteroids or bronchodilators as monotherapy to individuals reporting activity-limiting respiratory symptoms (especially dyspnea) and having an FEV₁ less than 60% would maintain benefits and minimize unnecessary testing or ineffective treatment. Pulmonary rehabilitation in these individuals may be beneficial, and long-term nocturnal supplemental oxygen in the presence of resting hypoxemia can reduce mortality. Spirometry to monitor disease status or modify therapy has not been evaluated in randomized trials. Studies are required to determine whether the relative effectiveness among therapies varies according to an individual's baseline or follow-up spirometry findings.