Inhaled corticosteroid doses in asthma: an evidence-based approach

In asthma, taking inhaled corticosteroids (ICS) substantially improves morbidity and mortality,1 and is the basis of pharmacotherapy for disease control.2,3 However, there is great variability in the doses of ICS prescribed for asthma. While the Australian guidelines recommend a specific dose for adults (eg, 500 μg of fluticasone per day), the recent Global Initiative for Asthma (GINA) guidelines recommend a wide dose range from 200 μg to 1000 μg of beclomethasone per day.3 There is now evidence that most (> 90%) of the benefit from ICS is achieved at relatively low doses, equivalent to 250 μg of fluticasone per day.4 Despite this, very high doses of ICS continue to be used, particularly in Australia and New Zealand,5 and there are emerging reports of significant side effects occurring with high dose ICS use.6

We sought to evaluate the balance between the efficacy and safety of different doses of ICS for asthma, and to communicate this to prescribers in an efficient way using the evidence-based measures of number needed to treat (NNT — the number of patients required to receive an intervention for an additional patient to benefit) and number needed to harm (NNH — the number of patients needed to receive an intervention for an additional patient to develop a side effect).7

Level 1 evidence8 of the efficacy and safety of different doses of ICS in asthma was identified by searching the Cochrane Database of Systematic Reviews9 using the search terms "asthma or wheez* and inhaled corticoster* or beclometh* or triamcin* or flutic* or budes* or betameth* or flunis* or cicles* or momet*". Six completed systematic reviews fitted the criteria (ie, compared ICS with either placebo or different doses of ICS for chronic asthma).10-15 Data were extracted from the results of meta-analyses included in the reviews.

Estimates of efficacy of ICS based on several parameters (FEV1, peak expiratory flow, night waking and rescue β2-agonist use) were calculated using the weighted mean difference or standardised mean difference from the meta-analyses for each dose of ICS compared with placebo.10 Insufficient data were available for efficacy estimates of fluticasone 2000 μg.

NNT and NNH were calculated for each dose of ICS using the odds ratios and control-group event rates from the meta-analyses. The method of calculating NNT is given in Box 1. The randomised controlled trials used standard predefined criteria to report the number of subjects withdrawn because of poor asthma control or worsening asthma and development of side effects.

Beclomethasone and budesonide were found to be superior to placebo in efficacy in terms of symptoms, lung function and exacerbations.12,14 However, dose–response effects could not be adequately evaluated, as the relatively small number of trials assessing a wide range of doses limited the ability to aggregate results.13,15 Adequate data were available for evaluation of the dose–response effects for fluticasone in asthma.10,11 These results are displayed graphically in Box 2 A–D. Clinically small, but occasionally statistically significant, dose–response effects were present for high versus low dose ICS for several outcomes (Box 2 A-D).

The NNT to prevent withdrawal because of worsening asthma decreased with increasing doses of fluticasone (Box 3). Treatment of three people with fluticasone 100 μg daily, or treatment of two people with 500 μg of fluticasone daily, was able to prevent one person developing a significant deterioration in asthma control. These data indicate a relatively flat dose–response curve for use of ICS in asthma.

ICS therapy also led to a significant increase in the side effects of hoarseness/dysphonia and oral candidiasis for all doses up to 500 μg daily. There was a trend for more sore throats to be reported for doses < 200 μg/day and a significant increase in sore throats for doses of 500 μg/day. The NNH for hoarseness/dysphonia at 200 μg fluticasone daily was 131, whereas the NNH reduced markedly to 23 (15–52) for a daily dose of 500 μg of fluticasone (Box 3) (ie, for every 23 people treated with fluticasone 500 μg daily, one person developed clinically significant hoarseness/dysphonia). Similarly, the NNH for oral candidiasis was 61 at 200 μg fluticasone daily, and this reduced to 21 (14–46) for fluticasone 500 μg/day (Box 3) (ie, for every 21 people treated with fluticasone 500 μg daily, one person developed oral candidiasis).

No differences were reported in plasma cortisol levels in doses up to 500 μg per day. In one study (good quality) fluticasone 1000 μg/day was associated with a significantly lower plasma cortisol level, but no difference was reported in the other studies.

Box 4 shows the benefits and harm with increasing fluticasone doses. The efficacy curve for different doses of fluticasone is relatively flat. As fluticasone doses increased, there was little gain in terms of efficacy; however, the NNH for oral candidiasis reduced from every 90th patient suffering oral candidiasis with fluticasone 100 μg to every sixth patient with fluticasone 2000 μg.

In contrast, the dose–response curve for side effects is steep. Side effects are relatively rare at low doses of ICS, with an NNH for fluticasone 100 μg per day of 131. With increasing doses, the side effect rate progressively increases. At 2000 μg fluticasone daily, the NNH drops to 6. This contrasts with an NNT of 2 indicating a very narrow margin of safety. When these data are displayed graphically (Box 4), with the relevant portion of the 95% CI for each parameter, the gap (unshaded area) represents the safety margin associated with ICS use in asthma. This clearly shows that the main effect of increasing ICS dose in asthma is to increase side effects, with little additional benefit to the patient.

Based on the available evidence, the use of lower doses of ICS would be associated with fewer side effects without loss of efficacy. The results question the current practice in Australia — the widespread use of high-dose ICS.

2: Efficacy of different doses of fluticasone in asthma

Results are weighted mean difference or standardised mean difference with 95% CI, obtained from meta-analyses of randomised controlled trials.10 * P < 0.05 v 50 μg. † P < 0.05 v 100 μg. ‡ P < 0.05 v 200 μg.11

4: Comparison of the relative effects of increasing doses of fluticasone in asthma

Results for fluticasone are displayed in terms of benefit (number needed to treat; NNT) and harm (number needed to harm; NNH) on a logarithmic scale, with the relevant portion of the 95% CI (shaded area). * Significant heterogeneity present (P < 0.05).