In response to The effect of study design and analysis methods on recovery rates in Bell's palsy

✍ Scribed by Thomas Berg; Lars Jonsson

• Publisher: John Wiley and Sons
• Year: 2010
• Tongue: English
• Weight: 38 KB
• Volume: 120
• Category: Article
• ISSN: 0023-852X
• DOI: 10.1002/lary.20792

✦ Synopsis

We appreciate the comments of Dr. Lee regarding our article titled ''The Effect of Study Design and Analysis Methods on Recovery Rates in Bell's Palsy.'' 1 We agree with Dr. Lee that it is very important to describe exactly the method of analyzing facial function, the follow-up time, and the time of evaluation of facial function in a Bell's palsy study. As we show in our article, these methodological factors substantially affect recovery rates and should therefore not be overlooked.

When analyzing a randomized clinical trial, analyzing all randomized patients, including dropouts, according to the intention-to-treat principle, has been strongly recommended. 2 This is the method we used in the Scandinavian Bell's palsy study. 3 Many previous reports have excluded dropouts in their final analysis according to the complete-case analysis. Our aim was therefore to compare these two different approaches to missing data. We also evaluated the impact of two different definitions of facial recovery. This strategy resulted in four different analysis methods being applied to our data. All four methods gave significant differences in recovery rates between the prednisolone and no-prednisolone groups.

Dr. Lee stated that the results of these four methods seemed to correlate well. However, in both completecase analysis groups, some of the patients included in the last-observation-carried-forward analysis are excluded from the final analysis. We therefore believe that it may be hazardous to estimate the correlation between the four analysis methods because they include different numbers of patients.