Rivastigmine for Dementia Associated with Parkinson's Disease
http://www.100md.com
《新英格兰医药杂志》
To the Editor: Emre et al. (Dec. 9 issue)1 report "moderate improvements" with rivastigmine in patients with dementia associated with Parkinson's disease. We believe that the small size of the effect, significant side effects, and overlooked biases in the study do not support the use of rivastigmine for dementia associated with Parkinson's disease.
The absolute difference in the rate of "clinically meaningful improvement" on the Alzheimer's Disease Cooperative Study–Clinician's Global Impression of Change (ADCS–CGIC) was only 5.3 percentage points between rivastigmine and placebo. Although the number needed to treat to achieve this benefit is 19, the "number needed to harm" to cause nausea is 6; to cause vomiting, 7; and to cause tremor, 16. For every 19 patients treated with rivastigmine, only 1 would improve, but 4 could potentially be harmed.
Several potential biases introduced by the higher rate of adverse effects with treatment may exaggerate the treatment benefits: discounting symptoms of Parkinson's disease when rating impressions of change (symptoms of tremor were ignored at evaluation); using no change as the imputed value in a last-observation-carried-forward analysis for a condition in which deterioration is expected; and "unblinding" due to high rates of nausea and vomiting among patients taking rivastigmine.2
Caroline N. Harada, M.D.
Joseph W. Shega, M.D.
Greg A. Sachs, M.D.
University of Chicago Hospitals
Chicago, IL 60637
caroline.harada@uchospitals.edu
References
Emre M, Aarsland D, Albanese A, et al. Rivastigmine for dementia associated with Parkinson's disease. N Engl J Med 2004;351:2509-2518.
DuBeau CE, Khullar V, Versi E. "Unblinding" in randomized controlled drug trials for urinary incontinence: implications for assessing outcomes when adverse effects are evident. Neurourol Urodyn 2005;24:13-20.
The authors reply: We think that the analysis provided by Dr. Harada et al. is incomplete. Their analysis of the number needed to treat to achieve a benefit disregards the difference in the proportion of patients who had clinically meaningful worsening, which was 10.1 percentage points. Dementia associated with Parkinson's disease is a progressive disorder and slowing of worsening is also a favorable outcome. The analysis of the number needed to harm is based on the frequency of adverse events, disregarding their consequences. It should be based instead on the difference in the proportion of patients who discontinued treatment because of adverse events, which was 3.0 percentage points for nausea, 1.3 percentage points for vomiting, and 1.7 percentage points for tremor. The ADCS–CGIC was meant to measure only changes in symptoms of dementia, not in motor function, which was separately assessed. In an attempt to minimize bias due to imputed values in the last-observation-carried-forward analysis, the intention-to-treat-retrieved-dropout population was chosen for the primary efficacy analysis. Finally, in order to minimize potential unblinding as a result of adverse events, ratings on the ADCS–CGIC and neuropsychological assessments were performed by independent assessors, who were otherwise not involved in the clinical evaluation of patients.
Murat Emre, M.D.
Istanbul Faculty of Medicine
34390 Istanbul, Turkey
muratemre@superonline.com
Dag Aarsland, M.D., Ph.D.
University of Bergen
4095 Stavanger, Norway
Werner Poewe, M.D.
Medical University of Innsbruck
A-6020 Innsbruck, Austria
The absolute difference in the rate of "clinically meaningful improvement" on the Alzheimer's Disease Cooperative Study–Clinician's Global Impression of Change (ADCS–CGIC) was only 5.3 percentage points between rivastigmine and placebo. Although the number needed to treat to achieve this benefit is 19, the "number needed to harm" to cause nausea is 6; to cause vomiting, 7; and to cause tremor, 16. For every 19 patients treated with rivastigmine, only 1 would improve, but 4 could potentially be harmed.
Several potential biases introduced by the higher rate of adverse effects with treatment may exaggerate the treatment benefits: discounting symptoms of Parkinson's disease when rating impressions of change (symptoms of tremor were ignored at evaluation); using no change as the imputed value in a last-observation-carried-forward analysis for a condition in which deterioration is expected; and "unblinding" due to high rates of nausea and vomiting among patients taking rivastigmine.2
Caroline N. Harada, M.D.
Joseph W. Shega, M.D.
Greg A. Sachs, M.D.
University of Chicago Hospitals
Chicago, IL 60637
caroline.harada@uchospitals.edu
References
Emre M, Aarsland D, Albanese A, et al. Rivastigmine for dementia associated with Parkinson's disease. N Engl J Med 2004;351:2509-2518.
DuBeau CE, Khullar V, Versi E. "Unblinding" in randomized controlled drug trials for urinary incontinence: implications for assessing outcomes when adverse effects are evident. Neurourol Urodyn 2005;24:13-20.
The authors reply: We think that the analysis provided by Dr. Harada et al. is incomplete. Their analysis of the number needed to treat to achieve a benefit disregards the difference in the proportion of patients who had clinically meaningful worsening, which was 10.1 percentage points. Dementia associated with Parkinson's disease is a progressive disorder and slowing of worsening is also a favorable outcome. The analysis of the number needed to harm is based on the frequency of adverse events, disregarding their consequences. It should be based instead on the difference in the proportion of patients who discontinued treatment because of adverse events, which was 3.0 percentage points for nausea, 1.3 percentage points for vomiting, and 1.7 percentage points for tremor. The ADCS–CGIC was meant to measure only changes in symptoms of dementia, not in motor function, which was separately assessed. In an attempt to minimize bias due to imputed values in the last-observation-carried-forward analysis, the intention-to-treat-retrieved-dropout population was chosen for the primary efficacy analysis. Finally, in order to minimize potential unblinding as a result of adverse events, ratings on the ADCS–CGIC and neuropsychological assessments were performed by independent assessors, who were otherwise not involved in the clinical evaluation of patients.
Murat Emre, M.D.
Istanbul Faculty of Medicine
34390 Istanbul, Turkey
muratemre@superonline.com
Dag Aarsland, M.D., Ph.D.
University of Bergen
4095 Stavanger, Norway
Werner Poewe, M.D.
Medical University of Innsbruck
A-6020 Innsbruck, Austria