Bronchodilators are the primary agents used to manage chronic obstructive pulmonary disease (COPD). They modestly improve forced expiratory volume in one second (FEV1) and reduce dynamic hyperinflation; breathlessness may lessen and exercise tolerance increase despite little improvement in spirometric measurements 1. The bronchodilators currently available for COPD include β2 agonists (e.g., salbutamol and salmeterol), anticholinergics (e.g., ipratropium bromide) and methylxanthines (e.g., theophylline).

According to the Canadian guidelines for the treatment of stable COPD 2 first line treatment consists of ipratropium, two to four doses three to four times daily, plus a short acting β2 agonist administered on an "as needed" basis. If the patient uses substantial amounts of short acting β2 agonists, or if the symptoms are greater at night than in the early morning, a long acting β2 agonist (salmeterol or formoterol) is added twice daily. However, recently some have recommended the latter as first line agents for stable COPD 3, 4567 potentially replacing the less expensive ipratropium 4, 5678.

Several trials have demonstrated the usefulness of salmeterol and formoterol for the management of COPD8, 91011. According to a 1998 meta-analysis 12, in patients with non-reversible COPD these agents produce small increases in FEV1; however, these changes alone may not correlate highly with symptom relief 13. The authors of the meta-analysis suggested that these drugs be prescribed only for patients who find they provide definite clinical improvement: reduced breathlessness or better exercise capacity. All three trials in the meta-analysis 10141516. compared long acting β2 agonists with placebo. Since then, other studies of these agents in COPD, including comparisons with anticholinergics 811 have appeared in the literature.

Canadian provincial drug plan managers have noted a substantial increase in the use of salmeterol and formoterol in recent years, an observation supported by data from International Medical Services Canada, which collects information on Canadian patterns of drug prescribing and estimates use: between 1997 and 2001, the use of salmeterol and formoterol for COPD increased 1,150% and 1,975%, respectively, whereas the use of ipratropium for COPD decreased by 37% 17.

In light of the new trials and the recent changes in prescribing practices, we undertook a systematic review to evaluate the efficacy and safety of long acting β2 agonists when compared with placebo or anticholinergic agents in patients with stable, poorly reversible, COPD.

We identified thirteen reports of twelve randomized controlled trials describing the effect of administering the long acting β2-agonists, salmeterol and formoterol, to patients with poorly reversible COPD.

It is not clear from the reports whether the twelve selected trials had sufficient power to detect significant differences between treatment and control groups in the various subjective and objective outcome measures. Since data were not pooled for meta-analysis, we were not able to conduct a sensitivity analysis based on the quality of trial reporting. Accordingly, we cannot comment on the possible influence of quality on the effect size of the outcome measures. Clinical heterogeneity among the trials limited assessment of the overall effect of the interventions. Since we did not perform a meta-analysis, statistical heterogeneity was not an issue.

We selected reports that met our inclusion criteria, regardless of publication status, language and trial quality using a systematic research methodology; this approach has been shown to minimize potential selection and publication bias and lead to more reliable conclusions36 We made every effort to conduct our review and report its results with the highest rigour.

A potential limitation of our research is that we did not seek trials comparing long acting β2-agonists marketed outside of Canada (e.g., bambuterol) or those trials comparing long acting β2-agonists to agents other than ipratropium and placebo (e.g., short-acting β2-agonists, methylxanthines). Similarly, we excluded those trials in which the FEV1 response to a bronchodilator was not reported or greater than 15%. Thus, our results may not be generalisable to the greater population of patients who can be currently defined as having COPD18. We plan to include a greater number of comparators and a broader population in an upcoming analysis37

Our results are similar to those of an earlier review 12 that identified three placebo controlled trials included in our review, but there are two important differences. In the earlier review FEV1 endpoint data from the placebo and salmeterol groups in two crossover trials 1516 were pooled; the weighted mean differences were not significant. We preferred to analyse net improvement in FEV1 (the difference from baseline to endpoint), as we felt that it more accurately reflected the impact of maintenance therapy. In addition, no trials comparing long acting β2 agonists and ipratropium were available at the time of the earlier review.

Another review has recently been published.38 However, these authors restricted their search to MEDLINE and failed to identify a clinical trial comparing formoterol with ipratropium.26 As a consequence, the evidence describing the use of formoterol versus ipratropium is limited to a single trial.39 In contrast, we opted to exclude this trial after identifying two trials because roughly 40% of patients exhibited partial reversibility of FEV1 (15% to 80%) to an inhaled dose of 200 mcg salbutamol at baseline. We are in agreement with the authors' summary of the evidence surrounding salmeterol versus ipratropium.

We believe our findings are in accord with current guidelines, such as the GOLD guidelines, that suggest bronchodilators should be prescribed according to individual patient responses. However, policymakers with limited health service resources need to be aware of an identifiable sub-population of patients with poorly reversible COPD for which long acting β2-agonists may result in reduced efficiency (cost-effectiveness).

Our research also suggests clinical investigators of COPD trials should stratify trial participants into groups for which outcomes may consistently differ. Of the trials identified, four23303140 used this approach. Outcome information from patients with poor reversibility was also analyzed in an abstract41 of an excluded trial but not in the published report.39 We were unable to ascertain sufficient details surrounding this analysis to add it to our findings.

In terms of clinical outcomes and safety, we could not find convincing evidence that salmeterol and formoterol have demonstrated advantages to ipratropium, a less expensive drug, for patients with stable COPD and poor reversibility. Compared to placebo, we found evidence of reduced rescue inhaler usage and improved spirometric outcomes without a significant impact on quality of life or exercise capacity.

Donald Husereau, Vijay Shukla, Michel Boucher, and Shaila Mensinkai have no competing interests to declare. CCOHTA is an independent, nonprofit health research agency funded by the federal, provincial, and territorial governments of Canada. Robert Dales sits on the advisory committees for GlaxoSmithKline (makers of the long acting β2 agonist, salmeterol) and Boeheringer Ingelheim (makers of the anticholinergic agent, ipratropium bromide).

DH edited and prepared the final manuscript for publication. VS led development of the research protocol, supervised the literature review, and summarized results. DH and VS were responsible for reviewing articles, judging their relevance, assessing their quality, and extracting data. MB assisted in developing the research protocol and in conflict resolution during study selection. RD assisted in developing the research protocol and provided clinical expertise. SM designed and conducted the electronic searches and provided expertise in the area of information science. All authors either wrote sections or critically reviewed drafts of this article.