RA Trials Have Significant Shortcomings in Reporting of Missing Data

The “amount, reporting, and handling” of missing data from rheumatoid arthritis (RA) drug therapy randomized clinical trials is analyzed.

About one-third of rheumatoid arthritis (RA) drug therapy randomized clinical trials (RCTs) have more than 20% missing data, according to study findings published in RMD Open.

Investigators searched the MEDLINE and Cochrane Central Register of Controlled Trials databases for the terms “rheumatoid arthritis” and “arthritis, rheumatoid” with limits of “Clinical Trials” and the years “2002–3”, “2006–7”, “2010–11”, and “2016.” The original report of a single RCT in participants with RA, parallel design, drug therapy intervention, and a clinical primary endpoint were the inclusion criteria. The study authors aimed to assess the trend in reporting in 2006 and 2016.

A total of 176 eligible RCTs with a median of 160 (interquartile range [IQR] 62-339) patients were included. Regarding actual analysis, 81 (46%) RCTs performed intention-to-treat (ITT) analysis, 23.9% conducted modified ITT analysis, and 30.1% performed non-ITT analysis. The researchers found that 59.8% RCTs that reported an ITT analysis actually performed it.

Among the trials, a median of 86% patients (IQR 74%-91%) were enrolled until trial completion, and 14 trial reports did not specify the number of patients who remained until completion. A total of 32.7% and 40.1% of RCTs had greater than 20% and 10% to 20% missing primary outcome data, respectively. In addition, a median of 100% (IQR 97.1%-100%) of patients were included in the efficacy analysis of primary trial outcome, and 13 RCTs did not specify the number of patients who were assessed regarding the primary outcome.

A total of 33.9% of RCTs did not report missing data handling, and 5 of 171 RCTs had no missing data. Also, 12 RCTs that performed ITT analysis did not specify the missing data handling method.

From 2006 to 2016, sensitivity analysis use improved from 14.7% to 21.4%. No significant improvement was observed in the parameters of missing data reporting, handling, and sensitivity analysis in the 2006 and 2016 comparisons.

Among several study limitations, the included RCTs were time and language restricted, and some of the deficiencies may result from poor reporting and not necessarily poor execution.

“Increased awareness and requirement of using the preferred imputation methods is needed,” stated the investigators. “Journals should require reporting of these methods as well as sensitivity analysis.”


Khan NA, Torralba KD, Aslam F. Missing data in randomised controlled trials of rheumatoid arthritis drug therapy are substantial and handled inappropriately. RMD Open. 2021;7(2):e001708. doi:10.1136/rmdopen-2021-001708