Just a moment, the page is loading...

Assessment of bias from treatment discontinuation on mortality and exacerbations in the Towards a Revolution in COPD Health (TORCH) randomized trial

Assessment of bias from treatment discontinuation on mortality and exacerbations in the Towards a Revolution in COPD Health (TORCH) randomized trial

Samy Suissa

McGill University


S Suissa has received research grants from Boehringer Ingelheim and Novartis, and has participated in advisory board meetings or as speaker for AstraZeneca, Boehringer-Ingelheim, Novartis, and Pfizer. Dr. Ernst has not conflicts of interest to disclose.

14 September 2016

The randomized controlled trial (RCT) is the preferred method to study a drug’s effectiveness and have it approved for use. In an ideal situation, all patients in a trial take their assigned drug and continue to follow the trial until the end. However, in real life, compliance to the assigned treatment in a RCT is often poor and some studies stop to follow patients when treatment is discontinued, possibly leading to biased results.

The Towards a Revolution in COPD Health (TORCH) is a large RCT where patients with COPD were randomized to receive salmeterol and fluticasone propionate, either alone or in combination, or to receive placebo, and followed for 3 years. Data were collected on the outcome of COPD exacerbations until treatment discontinuation. Since 44% of patients in the placebo group discontinued their assigned treatment, compared with 34% in the salmeterol-fluticasone propionate combination group, the comparison of these treatments on the outcome of exacerbation is likely biased.

In this request, we propose to use new techniques of data analysis to evaluate the extent of this bias by comparing the standard analyses with adherence-adjusted analyses for the outcomes of exacerbation and mortality in the TORCH trial. We will thus first replicate the analysis conducted in the publication of the TORCH trial in order to verify if it leads to the same results and confirm that the data are the same. Second, we will perform an adherence-adjusted analysis in which we will use inverse probability weighting to produce results adjusted for non-adherence, for both outcomes of mortality and exacerbations. We selected the TORCH trial for this analysis particularly because it is a trial where treatment adherence is a concern, so that we can expect differences in the results from the two methods.

This study will benefit upcoming and future trials, providing methods that account for treatment non-adherence, thus avoiding biased findings in evaluating the effectiveness of a drug. The findings will be communicated in peer-reviewed journal articles.

[{ "PostingID": 416, "Title": "GSK-SCO30003", "Description": "A multicentre, randomised, double-blind, parallel group, placebo-controlled study to investigate the long-term effects of salmeterol/fluticasone propionate (SERETIDE® inhaler) 50/500mcg BD, salmeterol 50mcg BD and fluticasone propionate 500mcg BD, all delivered via the DISKUS®/ACCUHALER® inhaler, on mortality and morbidity of subjects with chronic obstructive pulmonary disease (COPD) over 3 years of treatment

Medicine: fluticasone propionate/salmeterol, Condition: Pulmonary Disease, Chronic Obstructive," }]

Statistical Analysis Plan

The publication citation will be added after the research is published.