Rapid Responses to:
|
|
Rapid Responses: Rapid Responses published:
-
![[Read Rapid Response]](/icons/misc/sectionDOWN.gif)
Mortality odds ratio for all data should be 0.94, not 1.13
- Larry S. Hobbs (15 August 2004)
-
Motor fluctuations, Selegiline and neuro-protection in Parkinson's disease
- Milind S Deogaonkar (11 September 2004)
-
Mortality odds ratio is 1.13, and not 0.94
- Natalie J Ives, Rebecca L. Stowe, Keith Wheatley, Carl E. Clarke, Richard Gray (11 October 2004)
-
Re: Mortality odds ratio is 1.13, and not 0.94
- Adam Jacobs (11 October 2004)
-
Total line on figure 1 is correct
- Natalie J Ives, Rebecca L. Stowe, Keith Wheatley, Carl E. Clarke, Richard Gray (19 November 2004)
|
|
|||
|
Larry S. Hobbs, researcher Pragmatic Research, Irvine, CA 92612 USA
Send response to journal:
|
Odds ratio for
deaths in trials other
than the UK-PDRG trial should be 0.85 rather than 1.02
It appears the odds ratio for the risk of deaths in trials other than the Parkinson's Disease Research Group of the United Kingdom (UK-PDRG) should be 0.85 rather than 1.02. I believe the correct calculation is 15.5% divided by 18.2% which equals an odds ratio of 0.85. This means there were 15 percent fewer deaths (1.00 - 0.85 = 15 percent) in people given MAO B inhibitors than those who were not. The odds ratio can also be calculated from the Subtotals shown Figure 1 at the bottom of page 4, which would be as follows: Deaths the MAOBI patients (211/1360) divided by Deaths in Placebo patients (213/1171) which equals 0.85. This odds ratio appears on page 2 of 7 in column 2 under the heading of "Mortality". Odds ratio for deaths using all data should be 0.94 rather than 1.13It appears the odds ratio for the risk of deaths using all data should be 0.94 rather than 1.13. I believe the correct calculation is 19.99% divided by 21.15% which equals an odds ratio of 0.94. This means there were 6 percent fewer deaths (1.00 - 0.94 = 6 percent) in people given MAO B inhibitors than those who were not. The odds ratio can also be calculated from the Totals shown Figure 1 at the bottom of page 4, which would be as follows: Deaths the MAOBI patients (287/1436) divided by Deaths in Placebo patients (257/1215) which equals 0.94. This odds ratio appears on page 2 of 7 in column 2 under the heading of "Mortality", and also appears in the abstract. Competing interests: None declared |
|||
|
|
|||
|
Milind S Deogaonkar, Fellow, Department of Neurosciences Cleveland Clinic Foundation, Cleveland, OH 44195, USA
Send response to journal:
|
In a progressive disease like Parkinson’s disease (PD), the aim of therapy is to slow down the progression and improve the quality of life. One important factor that governs the quality of life is motor fluctuations. The occurrence of these motor fluctuations is dictated by the extent of striato-nigral degeneration. In the later stages of disease due to extensive nigral neuronal depletion there are fewer striatal dopaminergic terminals. This leads to insufficient storage of dopamine to buffer fluctuations in plasma levodopa and to prevent striatal dopamine receptors being exposed to fluctuations in plasma levodopa concentration. Plasma levodopa concentration thus, can influence the striatal dopamine concentration and bring about the pulsatile stimulation of dopamine receptors [1]. Thus, protection or augmentation of striatal dopamine terminals can be a useful approach to prevent the dyskinesias. . ‘Neuro-protection’ to slowing down disease progression of PD is an important strategy to prevent levadopa induced dyskinesia [2] and thereby improves the quality of life in PD patients. One such drug that is supposed to have neuro-protective effect is Selegiline. Selegiline is shown to have neuro-protective effect in variety of laboratory animals [3]. Studies have shown that it delays the need of levodopa, allows sparing of levodopa dose and reduces levodopa induced motor fluctuations. The neuro-protective effect of selegiline is mediated through the anti-apoptotic mechanisms. In spite of this, long-term use of selegiline is not very common. The meta-analysis presented in this article definitely shows that selegiline reduces disability, the need for levodopa, and the incidence of motor fluctuations, without substantial side effects or increased mortality [4]. 1. J.A. Obeso, C.W. Olanow, J.G. Nutt, Levodopa motor complications in Parkinson's disease, Trends Neurosci 23 (2000) S2-7. 2. O. Rascol, Medical treatment of levodopa-induced dyskinesias, Ann Neurol 47 (2000) S179-188. 3. C.W. Olanow, C. Mytilineou, W. Tatton, Current status of selegiline as a neuroprotective agent in Parkinson's disease, Mov Disord 13 Suppl 1 (1998) 55-58. 4. Natalie J Ives, Rebecca L Stowe, Joanna Marro, Carl Counsell, Angus Macleod, Carl E Clarke, Richard Gray, and Keith Wheatley Monoamine oxidase type B inhibitors in early Parkinson's disease: meta-analysis of 17 randomised trials involving 3525 patients BMJ 2004; 329: 593-0 Competing interests: None declared |
|||
|
|
|||
|
Natalie J Ives, Senior Statistician University of Birmingham, B15 2RR, UK, Rebecca L. Stowe, Keith Wheatley, Carl E. Clarke, Richard Gray
Send response to journal:
|
In reply to the rapid response regarding the odds ratio for mortality in our meta-analysis, the odds ratio for all data is 1.13 (as reported in the paper), and not 0.94 as suggested by Hobbs. The method used by Hobbs to calculate the odds ratio of 0.94 (and 0.85 for deaths in trials other then the UK-PDRG) is incorrect, as it was based on the formula for calculating the odds ratio of an individual trial. For example, if one trial reported 287 deaths in 1436 patients in arm A and 257 deaths in 1215 patients in arm B, then indeed the odds ratio for that trial would equal 0.94 (287/1436 divided by 257/1215). However, the overall odds ratio in a meta-analysis is not calculated using this formula. The basic principle of meta-analysis is to make comparisons of treatment with control within a trial, and to avoid completely any direct comparisons of patients in one trial with patients in another. This is achieved by calculating for each trial, the standard quantity “observed minus expected” (O-E) (and its variance) for the number of deaths among treatment-allocated patients. The overall O-E (and variance) for the meta-analysis is then simply obtained by adding together the individual trial O-E’s (and similarly for the variance), which can then be translated into an odds ratio (e.g. ln (odds ratio) = O-E / Variance) [1,2]. Therefore, in Figure 1 of our paper, adding together the individual trial O-E and variances gives an overall O-E of 14.6 and variance of 123.5, which gives an overall odds ratio for mortality of 1.13 (= exponential (14.6 / 123.5)), and not 0.94. References 1. Early Breast Cancer Trialists’ Collaborative Group. Treatment of early breast cancer. Vol 1: Worldwide evidence 1985-1990. Oxford: Oxford University Press, 1990. 2. Sutton AJ, Abrams KR, Jones DR, Sheldon TA, Song F. Methods for meta-analysis in medical research. Chichester: Wiley, 2000. Competing interests: None declared |
|||
|
|
|||
|
Adam Jacobs, Director Dianthus Medical Limited, London SW19 3TZ
Send response to journal:
|
Just one brief observation that might help clear up the confusion about the odds ratios: the totals at the bottom of figure 1 are not actually the sum of the values given above, which I assume is an error. According to my calculations, they should be 287/1631 and 257/1420, rather than 287/1436 and 257/1215 as printed. Calculating the odds ratio from these figures in the manner suggested by Hobbs still doesn't give an odds ratio of 1.13, as calculated by the more appropriate method of Ives et al, but it does give a value (0.97) slightly closer to it. The difference between the two methods of calculating the odds ratio therefore takes on less importance. Competing interests: None declared |
|||
|
|
|||
|
Natalie J Ives, Senior Statistician University of Birmingham, B15 2RR, UK, Rebecca L. Stowe, Keith Wheatley, Carl E. Clarke, Richard Gray
Send response to journal:
|
In reply to the comment by Jacobs regarding our meta-analysis, the totals at the bottom of Figure 1 are correct (see footnote on figure). The reason that the totals do not equal the sum of the values is because the UK-PDRG study is included twice (as both hypothesis generating and subsequent data), but counted only once in the total denominator. The results from the UK-PDRG study were first published in 1995, when they reported 76 deaths among 271 selegiline treated patients compared to 44 deaths in 249 control patients (hazard ratio=1.57, 95% confidence interval=1.09 to 2.30; p=0.015) [1]. This report of excess mortality by the UK-PDRG was considered as hypothesis generating in our meta-analysis, and was compared to subsequent mortality data reported, which included data reported in 2001 after additional follow-up of the patients in the UK -PDRG study [2]. In the 2001 publication, they reported that 148 selegiline patients and 118 levodopa patients had died. Thus in the subsequent data analysis, the UK-PDRG study contributes 72/195 (148-76 / 271-76) vs. 74/205 (118-44 / 249-44) deaths to the analysis. So the UK- PDRG study is included in both the hypothesis generating and subsequent data analyses, but contributes only once to the overall total. Therefore, the total number of patients included in the overall analysis is calculated as 271+1360-195=1436 for the selegiline arm and 249+1171- 205=1215 for the control arm. And hence the total line is correct, and the overall odds ratio for mortality is 1.13. References 1. Lees AJ, on behalf of the Parkinson’s Disease Research Group of the United Kingdom. Comparison of therapeutic effects and mortality data of levodopa and levodopa combined with selegiline in patients with early, mild Parkinson's disease. BMJ 1995;311:1602-7. 2. Lees AJ, Katzenschlager R, Head J, Ben-Shlomo Y, on behalf of the Parkinson’s Disease Research Group of the United Kingdom. Ten-year follow- up of three different initial treatments in de-novo PD: a randomized trial. Neurology 2001;57:1687-94. Competing interests: None declared |
|||