Sensitivity analysis by worst and best case assessment: is it really sensitive?
The question of how to perform an analysis according to intention-to-treat in the presence of missing information is of major importance. Several procedures have been suggested, but there is no generally accepted rule. Sensitivity analysis has been recommended as one solution. Extreme case analysis,...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article (Journal) |
| Language: | English |
| Published: |
1999
|
| In: |
Drug information journal
Year: 1999, Volume: 33, Issue: 3, Pages: 835-839 |
| ISSN: | 2164-9200 |
| DOI: | 10.1177/009286159903300324 |
| Online Access: | Verlag, Volltext: http://dx.doi.org/10.1177/009286159903300324
|
| Author Notes: | Kristina Unnebrink, and Jürgen Windeler |
| Summary: | The question of how to perform an analysis according to intention-to-treat in the presence of missing information is of major importance. Several procedures have been suggested, but there is no generally accepted rule. Sensitivity analysis has been recommended as one solution. Extreme case analysis, consisting of worst case and best case analysis, has been widely requested to assess the sensitivity of results to the method of handling missing values., In a simulation study the consequences of worst case and best case analysis on type I error and power of clinical trials with continuous endpoints are evaluated. The strategies of worst case and best case analysis are too extreme for a meaningful sensitivity analysis. Only for very small dropout rates are the two strategies able to discriminate at all between different situations. The demand for sensitivity analysis is no satisfactory solution to the intention-to-treat problem unless its methods and application are evaluated carefully, it is specified in advance, and its interpretation is clear-cut. |
|---|---|
| Item Description: | Gesehen am 19.04.2021 |
| Physical Description: | Online Resource |
| ISSN: | 2164-9200 |
| DOI: | 10.1177/009286159903300324 |