Not quite a breath of fresh air

Clinical versus statistical significance

Readers must interpret annualized rates of exacerbations and subsequent reductions with caution.³ In the TORCH trial, the annual rate of exacerbation at baseline was approximately 1 per year.² After treatment, LABA-alone patients had a rate of 0.97 per year and LABA-ICS patients had a rate of 0.85 per year.² The clinical significance of a statistical reduction of 0.12 exacerbations per year is unclear. Does this mean you would need to treat a patient for 8 years with LABA-ICS combination therapy to prevent one additional exacerbation versus LABA alone? This result is reported as a 12% relative risk reduction in a recent Cochrane review, which is misleading.^2,3

We are also concerned about the emphasis in the guideline that there is a clinical benefit with LABA-ICS combination therapy versus LABA alone with respect to health status. Nannini et al report a St George’s Respiratory Questionnaire (SGRQ) score improvement of −1.64 points (95% CI −2.28 to −1) in 4 studies (N = 4700).^3–5 However, the minimum clinically important difference in SGRQ is thought to be a change of at least 4 points.⁵ In our opinion, clinicians should be alerted to the fact that trials might have shown a statistical improvement in SGRQ scores with LABA-ICS combination therapy, but this difference might not be clinically perceptible. Again, owing to differential rates of withdrawal, statistical differences in SGRQ scores (health status) could be the result of confounding factors and might be unrelated to the intervention. Furthermore, even if there is a statistical difference in SGRQ scores between patients receiving LABA-ICS combination therapy versus those receiving LABA alone, there is insufficient evidence to support a clinically important change in health status.³

A Cochrane systematic review³ highlights the contribution of the TORCH trial and an earlier trial, TRISTAN (TRial of Inhaled STeroids ANd long-acting β₂ agonists),⁴ which had similar results to TORCH, to the overall body of evidence for the use of LABA-ICS combination inhalers in COPD patients. These trials accounted for 70% of the overall weighting of the 5 included trials for the outcomes of exacerbation frequency and health status.^2–4 Therefore, we should be examining the outcomes and the authors’ analysis of the outcomes used in these particular trials with careful scrutiny.

The guidelines¹ state the following regarding the use of LABA-ICS therapy in combination with tiotropium⁶: For patients with moderate to severe COPD with persistent symptoms and a history of exacerbations … a combination of tiotropium plus a LABA and ICS therapy product … is recommended to improve bronchodilation and lung deflation, to reduce the frequency and severity of exacerbations and to improve health status.¹

We assume that the Optimal trial⁶ data provide the basis for these guideline recommendations. The Optimal trial studied the addition of LABA-ICS combination therapy, LABA, or placebo to patients who were receiving tiotropium.⁶ The authors concluded that “the addition of fluticasone-salmeterol to tiotropium therapy did not statistically influence rates of COPD exacerbation but did improve lung function, quality of life, and hospitalization rates in patients with moderate to severe COPD.”⁶

The aforementioned issues regarding clinically versus statistically significant differences, as well as the lack of accountability of outcomes for patients who withdrew, also apply to interpretation of the Optimal trial. The recommendation from the guideline above is incorrect, as the Optimal trial did not show a significant difference in the frequency and severity of exacerbations among any of the treatment arms.⁶ Also, it appears LABA-ICS combination therapy added to tiotropium improves health status more than placebo plus tiotropium (−4.1 points on the SGRQ).⁶ However, this analysis is difficult to interpret given the number of patients who were not evaluated for this outcome at week 52, the lack of confidence intervals around the change in SGRQ scores for each treatment arm, and the fact that there did not appear to be a clinically important difference when comparing all other treatment arms.⁴

Bottom line

Clinical practice guidelines are an essential resource for front-line clinicians. However, in order for “bottom line” guideline recommendations to have a positive effect on patient care, we must ensure that these recommendations are based on clinically, rather than statistically, meaningful differences in outcomes.