NPC Archive Item: Combined LABA/ICS may not offer clinically important benefits over LABA alone in COPD, but is associated with serious adverse effects

NOTE – This is an archive post from the NPC and has not been updated since first publication. Therefore, some hyperlinks may no longer be working.
MeReC Rapid Review NPC Logo

27 October 2009

A large systematic review found that compared with LABA monotherapy, combined LABA/ICS did not statistically significantly reduce severe exacerbations, all-cause mortality, respiratory mortality or CV mortality, in patients with moderate to very severe COPD. Although combined LABA/ICS did provide some statistically significant benefits (e.g. reduction in moderate exacerbations, improvement in FEV1 measures and quality of life scores), the magnitude of these benefits were not considered clinically important. Furthermore, combined LABA/ICS was associated with serious adverse effects, particularly pneumonia.

Level of evidence:
Level 1 (good quality patient-oriented evidence) according to the SORT criteria.

Health professionals should continue to follow NICE guidance on COPD, which recommends that inhaled corticosteroids (ICSs) are only prescribed for specific patients. An ICS should be added in to treatment with a long-acting bronchodilator (salmeterol, formoterol or tiotropium) only in patients with moderate or severe COPD (forced expiratory volume in one second [FEV1] <50% predicted) who have had two or more exacerbations requiring treatment with antibiotics or oral corticosteroids in a 12-month period. Adding an ICS could also be considered in patients with moderate or severe COPD who are still breathless despite monotherapy with a long-acting bronchodilator, but the ICS should be discontinued if there is no benefit after four weeks.

What is the background to this?
As we discussed in a MeReC Bulletin on recent safety issues with inhaled treatments for COPD, there is no good evidence that ICSs improve survival in people with COPD, or reduce the rate of FEV1 decline. However, there is evidence to suggest that ICSs may reduce the frequency of COPD exacerbations, and may improve quality of life.

The potential risks of ICSs range from unpleasant local side effects, such as oral candidiasis and dysphonia, to less common systemic effects, such as adrenal suppression and osteoporosis. Recently, ICS treatment in patients with COPD has also been associated with an increased risk of pneumonia.

Current NICE guidance on COPD recommends that ICSs should be added to long-acting bronchodilators only in patients with moderate to severe COPD, and in specific circumstances as outlined above. The aim of treatment is to reduce exacerbation rates and slow the decline in health status, and not to improve lung function per se. This recent meta-analysis and systematic review of 18 RCTs (n=12,446) aimed to determine the risks of adding in an ICS to a long-acting beta agonist (LABA), compared with LABA monotherapy, and whether combined LABA/ICS provides any significant clinical benefits over LABA monotherapy in COPD.

What does this review claim?
Compared with LABA monotherapy, combination therapy with LABA/ICS did not statistically significantly decrease the risk of severe COPD exacerbations requiring hospitalisation or withdrawal (relative risk (RR) 0.91, 95% confidence interval [CI] 0.82 to 1.01), all-cause mortality (RR 0.90, 95%CI 0.76 to 1.06, P=0.77), respiratory mortality (RR 0.80, 0.61 to 1.05, P=0.58) or cardiovascular (CV) mortality (RR 1.22, 95%CI 0.88 to 1.71, P=0.46). There was a significant reduction in moderate exacerbations requiring oral corticosteroids or antibiotic use (RR 0.84, 95%CI 0.74 to 0.96, P=0.008. For further details of the analysis of primary outcomes, see Table 1 below.

Therapy with a LABA/ICS was also associated with improvements in measures of lung function, St George respiratory questionnaire (SGRQ) scores and end of treatment dyspnoea score, compared with LABA alone. However, LABA/ICS was found to increase the risk of pneumonia, oropharyngeal candidiasis and viral respiratory infections. For more details of secondary outcomes, see Table 2 below.

The authors concluded that compared with LABA monotherapy, the magnitude of the benefits of LABA/ICS therapy did not reach that of the predefined criteria for clinically important effects and were associated with serious adverse effects.

So what?
In this large study, treatment with combined LABA/ICS did not reduce severe COPD exacerbations (requiring hospitalisations or withdrawal), all-cause mortality, respiratory mortality or CV mortality. Only one of the primary outcomes reached statistical significance (moderate COPD exacerbations requiring oral corticosteroids or antibiotic use), although the authors question the clinical significance of this result. The relative risk reduction of moderate exacerbations with combined LABA/ICS was 16%, which is lower than the suggested threshold value of 22% for clinical significance.1 These findings appear to be consistent with other large studies. For example, in a Cochrane review of LABA/ICS combinations, combined LABA/ICS reduced total exacerbations, compared with LABA alone, but there was no reduction in all-cause mortality.

Other benefits were observed with combined LABA/ICS in some secondary outcomes e.g. mean change in pre- and post-bronchodilator FEV1, mean change in SGRQ scores, end of treatment dyspnoea scores, compared with LABA alone. However, the authors of the review also question the relevance of these improvements, as the magnitude of benefit did not reach the suggested clinically important minimal differences (FEV1 0.10 to 0.14L; SGRQ score 4 units).

The relative benefits of treatment with combined LABA/ICS must also be weighed against the risks. This review found a relative risk increase of 63% for pneumonia and 59% for oropharyngeal candidiasis for LABA/ICS compared with LABA. As we have previously blogged, other large RCTs and meta-analyses have demonstrated an increased risk of pneumonia with ICSs in people with COPD. A recent meta-analysis suggests that budesonide does not increase the risk of pneumonia, although further research is necessary to establish if budesonide does offer any long-term safety advantage over other ICSs (e.g. fluticasone) in COPD.

There are also other potential risks which are important and were not considered in this review e.g. osteoporosis (especially in the presence of other risk factors). When considering adding an ICS to treatment regimens for people with COPD, prescribers should consider, and discuss with patients, the potential increased risk of pneumonia, as well as osteoporosis and other side effects.

This review was well conducted but it is important to consider some potential limitations, which are particularly relevant to all COPD studies. The included studies did not consistently use similar definitions for COPD exacerbations or pneumonia, which may have affected the results. No studies were specifically designed or powered to detect differences between the two treatment groups for all-cause, respiratory or CV mortality. Furthermore, allocation concealment was unclear in 13 studies.

Further information on the management of COPD can be found on the COPD section of NPC

1. Cazzola M, MacNee W, Martinez FJ, et al. Outcomes for COPD pharmacological trials: from lung function to biomarkers. Eur Respir J 2008;31:416–68

Study details
Rodrigo GJ, Castro-Rodriguez JA, Plaza V. Safety and efficacy of combined long-acting β-agonists and inhaled corticosteroids vs long-acting β-agonists monotherapy for stable COPD. Chest 2009;136:1029–38

Design: Meta-analysis and systematic review of 18 RCTs (n=12,446). Fourteen studies used salmeterol/fluticasone and four studies used formoterol/budesonide. Eleven of the studies were long term (>52 weeks). Allocation concealment was adequate in only 5 studies, and was unclear in the remaining 13 studies.

Patients: Stable adults (>40 years) with moderate to very severe COPD (<80% predicted FEV1), according to the GOLD classification. Mean age of patients was 64 years with an average baseline FEV1 of 40% of predicted normal values. Eleven studies reported that a mean of 31% (range 0–55%) of patients had received ICSs before enrollment in the study.

Intervention & comparison: Inhaled combined LABA/ICS as the intervention arm compared with LABA monotherapy for more than 1 month.

Outcomes: Primary outcomes were severe COPD exacerbations (requiring hospitalisation or withdrawals), moderate COPD exacerbations (requiring oral corticosteroids or antibiotic use), all-cause mortality, respiratory death (due to a respiratory event e.g. COPD exacerbation, pneumonia) and cardiovascular mortality, during the treatment period. Secondary outcomes included measures of lung function, health-related quality of life (using SGRQ) and adverse effects.


Table 1: Primary outcomes ― LABA/ICS vs. LABA


LABA/ICS No./total no. (%)

LABA alone No./total no. (%)

RR (95%CI)

I2 a

Severe COPD exacerbations



0.91 (0.82 to 1.01)


Moderate COPD exacerbations



0.84 (0.74 to 0.96)


All-cause mortality



0.90 (0.76 to 1.06)


Respiratory deaths



0.80 (0.61 to 1.05)


CV mortality



1.22 (0.88 to 1.71)


a The I2 statistic measures heterogeneity in a meta-analysis. A value of 0% indicates no observed heterogeneity, and larger values show increasing heterogeneity.

Table 2: Secondary outcomes ― LABA/ICS vs. LABA


LABA/ICS No./total no. (%)

LABA alone No./total no. (%)

Measure (95%CI)

I2 a

Mean change in pre-bronchodilator FEV1



Weighted mean difference (WMD) 0.06 (0.04 to 0.07)


Mean change in post-bronchodilator FEV1



WMD 0.04
(0.02 to 0.05)


Mean change in SGRQ



WMD –1.88
(–2.44 to –1.33)


End of treatment dyspnoea score



Standardised mean difference –0.20
(–0.25 to –0.15) P=0.0001



263/5,212 (5.0)

153/4,520 (3.4)

RR 1.63 (1.35 to 1.98) P=0.0001


Oropharyngeal candidiasis

292/3,521 (8.4)

200/2,741 (7.2)

RR 1.59 (1.07 to 2.37) P=0.002


Viral respiratory infections

441/4,844 (9.1)

342/4,362 (7.8)

RR 1.22 (1.07 to 1.39) P=0.004


b The number of participants in the study arm

Post publication amendment (4/11/09)

In the first publication of this blog (27 October 2009), we included the numbers needed to treat (NNTs) and numbers needed to harm (NNHs), which had been calculated by the authors of the meta-analysis from pooled effect estimates. Following feedback and further consideration, these have now been removed due to the heterogeneity of different durations of treatment in the included studies.

Sponsorship: No sponsorship from institutions or pharmaceutical industry was provided.

Please comment on this blog in the NPC discussion rooms, or using our feedback form.

Make sure you are signed up to NPC Email updates — the free email alerting system that keeps you up to date with the NPC news and outputs relevant to you