Table 2

Summary of analyses.*

Meta-Analyses

Sensitivity Analyses

Overall Median Contrasts

Best-case and worst-case scenarios

4 possible combinations of contrasts including one control treatment per trial

Cleary [47] trial excluded

All 20 contrasts

Stratified Median Contrasts

A. Control Treatment

1. Active treatment or placebo control

Best-case and worst-case scenarios

2 possible combinations of contrasts including one control treatment per trial

Cleary [47] trial excluded

All 16 contrasts

2. No treatment

B. Country Where Trial was Performed

3. United Kingdom

Best-case and worst-case scenarios

2 possible combinations of contrasts including one control treatment per trial

Cleary [47] trial excluded

All 10 contrasts

4. United States

Best-case and worst-case scenarios

2 possible combinations of contrasts including one control treatment per trial

All 10 contrasts

C. Duration of Follow-Up

5. Short-term

Best-case and worst-case scenarios

All 9 contrasts

6. Intermediate-term

4 possible combinations of contrasts including one control treatment per trial

7. Long-term

2 possible combinations of contrasts including one control treatment per trial

Cleary [47] trial excluded

*There were insufficient contrasts to perform sensitivity analyses for the no treatment stratified analysis. For the short-term stratified analysis, the median contrast was defined to be that corresponding to the eighth combination when effect sizes for the 16 possible contrast combinations were rank ordered from least to greatest. For the intermediate-and long-term stratified analyses, the median contrasts defaulted to the all-contrasts analyses because there were no repeated measures within these time intervals in any trial. All possible contrast combinations were included in the sensitivity analyses for intermediate-and long-term follow-up because of the limited numbers of combinations for these analyses.

Licciardone et al. BMC Musculoskeletal Disorders 2005 6:43   doi:10.1186/1471-2474-6-43