On 2017 Oct 06, Peter Gøtzsche commented:

The authors compared inhalers with combination drugs (a steroid plus a long-acting beta-2 agonist), with placebo and write in their abstract that the number needed to treat to prevent one death with fluticasone/salmeterol was 42 (1). They explain, just before Objectives, that the largest randomised trial of combination therapy (TORCH) demonstrated a significant reduction in mortality versus placebo (P = 0.052) and that they wished to see whether other combined inhalers had a similar effect. Just above “Implications for research,” we are told that “whether a combination is better than the two components taken separately was not addressed in this review,” and under “Authors’ Conclusions” the authors advocate that the combination should be compared with its two components.

This information is misleading. Firstly, the review authors overlook that the TORCH trial (and several other trials) was designed to answer what they call for in future research, namely whether the combination was better than any of its components.

Secondly, the authors give the readers the impression that the combination reduces mortality. However, the fact is that the steroid contributes absolutely nothing to the mortality benefit. The primary outcome in the TORCH trial was total mortality (2). GlaxoSmithKline randomised 6184 patients to four groups: placebo; salmeterol; fluticasone; and both drugs together. By definition, this design is factorial. It is powerful, as it allows the investigators to study three research questions with a sample size that would usually only allow one question to be answered. Such a trial can tell us whether the two drugs are effective, and whether the combination is better than any of its components. However, the analysis in the TORCH trial included only half of the patients, thereby spoiling the advantage of the factorial design, although the published trial protocol stated that a factorial analysis was to be performed (3).

Nowhere in the 15-page trial report is the factorial analysis to be found, and the abstract of the TORCH trial gives the readers the clear impression that the combination was better than any of its components, which is the result the authors of the Cochrane review quoted.

The authors of a letter to the editor used a factorial analysis and showed that the effect of the combination was entirely due to salmeterol (4); the hazard ratio for fluticasone was 1.00 (0.87 to 1.15), p = 0.99. In other similar trials, both GlaxoSmithKline and AstraZeneca did not perform a factorial analysis (5).

1 Nannini LJ, Poole P, Milan SJ, et al. Combined corticosteroid and long-acting beta2-agonist in one inhaler versus placebo for chronic obstructive pulmonary disease. Cochrane Database Syst Rev 2013;11:CD003794.

2 Calverley PM, Anderson JA, Celli B, et al. Salmeterol and fluticasone propionate and survival in chronic obstructive pulmonary disease. N Engl J Med 2007;356:775–89.

3 Gøtzsche PC. Questionable research and marketing of a combination drug for smoker’s lungs. J R Soc Med 2014;107:256-7.

4 La Vecchia C and Fabbri LM. Prevention of death in COPD. N Engl J Med 2007;356:2211–2.

5 Suissa S, Ernst P, Vandemheen KL, et al. Methodological issues in therapeutic trials of COPD. Eur Respir J 2008;31:927–33.

On 2017 Aug 24, Peter Gøtzsche commented:

None

On 2017 Aug 24, Peter Gøtzsche commented:

None

On 2017 Oct 06, Peter Gøtzsche commented:

The authors compared inhalers with combination drugs (a steroid plus a long-acting beta-2 agonist), with placebo and write in their abstract that the number needed to treat to prevent one death with fluticasone/salmeterol was 42 (1). They explain, just before Objectives, that the largest randomised trial of combination therapy (TORCH) demonstrated a significant reduction in mortality versus placebo (P = 0.052) and that they wished to see whether other combined inhalers had a similar effect. Just above “Implications for research,” we are told that “whether a combination is better than the two components taken separately was not addressed in this review,” and under “Authors’ Conclusions” the authors advocate that the combination should be compared with its two components.

This information is misleading. Firstly, the review authors overlook that the TORCH trial (and several other trials) was designed to answer what they call for in future research, namely whether the combination was better than any of its components.

Secondly, the authors give the readers the impression that the combination reduces mortality. However, the fact is that the steroid contributes absolutely nothing to the mortality benefit. The primary outcome in the TORCH trial was total mortality (2). GlaxoSmithKline randomised 6184 patients to four groups: placebo; salmeterol; fluticasone; and both drugs together. By definition, this design is factorial. It is powerful, as it allows the investigators to study three research questions with a sample size that would usually only allow one question to be answered. Such a trial can tell us whether the two drugs are effective, and whether the combination is better than any of its components. However, the analysis in the TORCH trial included only half of the patients, thereby spoiling the advantage of the factorial design, although the published trial protocol stated that a factorial analysis was to be performed (3).

Nowhere in the 15-page trial report is the factorial analysis to be found, and the abstract of the TORCH trial gives the readers the clear impression that the combination was better than any of its components, which is the result the authors of the Cochrane review quoted.

The authors of a letter to the editor used a factorial analysis and showed that the effect of the combination was entirely due to salmeterol (4); the hazard ratio for fluticasone was 1.00 (0.87 to 1.15), p = 0.99. In other similar trials, both GlaxoSmithKline and AstraZeneca did not perform a factorial analysis (5).

1 Nannini LJ, Poole P, Milan SJ, et al. Combined corticosteroid and long-acting beta2-agonist in one inhaler versus placebo for chronic obstructive pulmonary disease. Cochrane Database Syst Rev 2013;11:CD003794.

2 Calverley PM, Anderson JA, Celli B, et al. Salmeterol and fluticasone propionate and survival in chronic obstructive pulmonary disease. N Engl J Med 2007;356:775–89.

3 Gøtzsche PC. Questionable research and marketing of a combination drug for smoker’s lungs. J R Soc Med 2014;107:256-7.

4 La Vecchia C and Fabbri LM. Prevention of death in COPD. N Engl J Med 2007;356:2211–2.

5 Suissa S, Ernst P, Vandemheen KL, et al. Methodological issues in therapeutic trials of COPD. Eur Respir J 2008;31:927–33.