Article Text


Ethics of placebo controlled studies of inhaled steroids for COPD
  1. D L Hahn1
  1. 1Dean Medical Center, East Clinic, 1821 S Stoughton, Road, Madison, WI 53716, USA;
  1. P S Burge2,
  2. S A Lewis3
  1. 2Birmingham Heartlands Hospital, Birmingham, B9 5SS, UK;
  2. 3Division of Respiratory Medicine, City Hospital, Nottingham, UK

Statistics from

The meta-analysis by Sutherland et al1 of the effect of inhaled corticosteroids on the progression of airflow limitation in patients with chronic obstructive pulmonary disease (COPD) found a small improvement in forced expiratory volume in 1 second (FEV1) of −7.7 ml/year (95% CI −14.2 to −1.3) which is similar to the results of the meta-analysis performed by Highland et al2 (−5 ml/year (95% CI −11.2 to 1.2)) using a very similar data set. The meta-analyses employed slightly different study selection criteria and analytical techniques, and questioned the clinical significance of such small differences in FEV1. The selected primary studies suffered from potential drop-out bias and significant selection bias, in that almost all of the studies subjected to these meta-analyses excluded patients with a bronchodilator response.2 Studies of asthma and COPD as separate entities are limited because asthma and COPD (observed in cross section) represent a continuum,3 and the small number of available prospective observations indicates that asthma and COPD are sometimes different clinical manifestations of the same underlying aetiology evolving over time.4 Given all the uncertainties, questions and limitations, Highland et al2 concluded (correctly in my opinion) that “additional studies are needed to evaluate the effects on quality of life, risk for systemic side effects, dose-response relationships in corticosteroid-responsive patients, and the economic effect of inhaled corticosteroids”. On the other hand, in an editorial accompanying the paper by Sutherland et al, Burge and Lewis5 state: “It is no longer ethical to do more long term placebo controlled studies [of inhaled corticosteroids in COPD]”. Given the uncertainties, questions and limitations which Burge and Lewis acknowledged, I was puzzled by this statement and would like to ask them why they reached this conclusion.


Authors’ reply

To make randomised controlled studies ethical, the investigator must believe that neither treatment is known to be superior to the other. If the trial is to be placebo controlled, the investigator must believe that no non-allowed treatment is known to be of benefit. Both Drs Duerden and Hahn want more placebo controlled trials of inhaled corticosteroids (ICS) in COPD before they recommend their use. Our editorial pointed out that ICS were of established benefit in reducing exacerbations of COPD,1 so any future placebo controlled study would need to be in patients without a history of exacerbations. As exacerbations are associated with disease severity, and as about 80% of patients with an FEV1 <50% predicted have exacerbations over a 3 year period,2 any trial would need to be in those with early disease. The Copenhagen City Lung Study found that inhaled budesonide 800 μg/day had no benefit in a population of smokers selected with a reduced FEV1/FVC ratio, the majority of whom had an FEV1 above 80% predicted.3 This leaves the group with an FEV1 between 50% and 80% predicted, many of whom have not been identified by their medical practitioners. This was the group included in the EUROSCOP and Lung Health 2 studies, where the results included in the meta-analysis were the most divergent, probably because of the relatively low dose of ICS used in the Lung Health study.4,5 In the symptomatic patient with more severe disease, the combination of a long acting β agonist and an ICS has been shown to be superior to either alone and is now the treatment of choice.6,7 This leaves the presymptomatic population in whom a decline in FEV1 is the only practical outcome measure. No randomised study using an intention to treat analysis has shown a reduction in FEV1 decline with any treatment in any disease, including ICS in asthma, although several studies have shown an improvement in FEV1 with ICS in COPD.2,4,6 Our editorial tried to explain why changes in FEV1 decline were difficult to show in patients with COPD. Patients with progressive disease are likely to be given ICS by their clinicians outside any trial, reducing the power of any study.

Any treatment should weigh the potential risks against any benefit. Dr Duerdin wants a better analysis of the risks of ICS in patients with COPD, particularly related to bone loss, and points out the reduced bone mineral density in the triamcinolone group in the Lung Health 2 study.5 The reported results are in 359/412 of a convenience sample who had three measurements of bone density. After 3 years the lumbar spine density reduced from 0.988 to 0.985 g/cm2 and the femoral neck from 0.762 to 0.747 g/cm2 in the triamcinolone group. EUROSCOP studied bone density measurements in 194 subjects4 and showed very small changes which were significantly less at the femoral trochanter in the budesonide group (0.04%/year v 0.36%/year in the placebo group). Randomised controlled studies are probably not the best method for assessing the extent of long term adverse effects, but the evidence from the randomised studies to date shows that the risks are relatively minor compared with the risks of death from the natural progression of the disease. Introducing ICS at an earlier stage may alter the risk/benefit ratios. The adverse effects on the bones are probably best studied in asthmatics of whom many are already taking long term ICS in equivalent doses.

There is a striking difference between the way that cardiac and respiratory physicians greet new treatments whose individual effects are present but relatively minor. There has been a meaningful reduction in cardiac deaths attributed to the combination of several treatments with modest individual effects. This has resulted in more smokers living to develop significant COPD. It is likely that improvements in the quality and quantity of life in patients with COPD will come from a combination of treatments, among which ICS have a place. The main unanswered questions are—at which stage to start and what dose to use? Randomised trials in these areas are badly needed. They will require large numbers, enthusiasm from respiratory clinicians, and are likely to need public rather than pharmaceutical industry funding.


View Abstract

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.