Author's Reply: Absence of evidence is still not evidence of absence
Dear Sir
Thank you for taking the time to read our article and submit your
comments. Here are our responses to your questions regarding the
interpretation of data.
1.In all our mixed model analyses we adjusted for the baseline
differences between the groups. We absolutely agree that this should be
done irrespective of whether or not the differences at baseline are
statistically significant... in fact, that is the reason why we adjusted
for these differences in our study.
2. The minimal important significant difference (MID)is estimated at
0.081 (SD 0.319) for the EQ-5D.
Szpalski, M. et al. (Eds) Surgery for Low Back Pain. Chap. 6.1.
Springer. 2010.
Walters, S.J. and Brazier, J.E. Comparison of the minimally important
difference for two health state utility measures: EQ-5D and SF-6D. Qual
Life Res. 2005 Aug; 14(6) 1523-32.
3. We agree that the power calculations show that we have too low
power to give a valid estimate of the between group differences concerning
sick leave. The P value given in our paper is therefore not so
informative. A more valuable presentation of the data could have been to
give the number of patients on sick leave in the three intervention
groups.
4. At the 52 week follow up 1 patient in the caudal epidural steroid
group, 6 patients in the caudal epidural saline group, and 8 patients in
the sham group had undergone back surgery during the previous year. We did
a within group comparison and used the number of patients in each group at
inclusion to give the within group differences with a P value of 0.07. We
agree that the presentation in the paper could have been misleading
because we have given the percentage in each group at the 52 week follow
up with a P value as reported above.
Rapid Response:
Author's Reply: Absence of evidence is still not evidence of absence
Dear Sir
Thank you for taking the time to read our article and submit your
comments. Here are our responses to your questions regarding the
interpretation of data.
1.In all our mixed model analyses we adjusted for the baseline
differences between the groups. We absolutely agree that this should be
done irrespective of whether or not the differences at baseline are
statistically significant... in fact, that is the reason why we adjusted
for these differences in our study.
2. The minimal important significant difference (MID)is estimated at
0.081 (SD 0.319) for the EQ-5D.
Szpalski, M. et al. (Eds) Surgery for Low Back Pain. Chap. 6.1.
Springer. 2010.
Walters, S.J. and Brazier, J.E. Comparison of the minimally important
difference for two health state utility measures: EQ-5D and SF-6D. Qual
Life Res. 2005 Aug; 14(6) 1523-32.
3. We agree that the power calculations show that we have too low
power to give a valid estimate of the between group differences concerning
sick leave. The P value given in our paper is therefore not so
informative. A more valuable presentation of the data could have been to
give the number of patients on sick leave in the three intervention
groups.
4. At the 52 week follow up 1 patient in the caudal epidural steroid
group, 6 patients in the caudal epidural saline group, and 8 patients in
the sham group had undergone back surgery during the previous year. We did
a within group comparison and used the number of patients in each group at
inclusion to give the within group differences with a P value of 0.07. We
agree that the presentation in the paper could have been misleading
because we have given the percentage in each group at the 52 week follow
up with a P value as reported above.
Regards
Trond Iversen
Competing interests: No competing interests