Neurology -- Correspondence for Miyasaki et al., 66 (7) 996-1002

Correspondence published:

Practice Parameter: Evaluation and treatment of depression, psychosis, and dementia in PD: Dag Aarsland, Murat Emre, Andrew Lees, Werner Poewe, Clive Ballard (17 August 2006)

Practice Parameter: Evaluation and treatment of depression, psychosis, and dementia in PD: Erwin B. Montgomery (17 August 2006)

Reply from the Authors (to Aarsland et al and Montgomery): Janis M Miyasaki, and Gary Gronseth on behalf of the author panel (17 August 2006)

Practice Parameter: Evaluation and treatment of depression, psychosis, and dementia in PD 17 August 2006

Dag Aarsland,
Stavanger University Hospital
Arm Hansen v 20, 4005 Stavanger, Norway,
Murat Emre, Andrew Lees, Werner Poewe, Clive Ballard
Send Correspondence to journal:
Re: Practice Parameter: Evaluation and treatment of depression, psychosis, and dementia in PD

daa{at}sir.no Dag Aarsland, et al.

We were surprised to see the the dementia section of the AAN Practice Parameter. [1] A large (n=541) randomized, controlled parallel-group trial with rivastigmine (“EXPRESS”) [2] is categorized as Class II evidence whereas a small (n=22) cross-over study with donepezil [3] is classified as Class I evidence.
The EXPRESS study fulfills all the criteria as defined for Class I studies by the Quality Standards Subcommittee of the AAN. [1] The cohort is representative, the randomization concealed, and primary endpoints defined a priori. More than 73% on rivastigmine completed the study and all drop-outs were accounted for. In the primary analysis, missing data were imputed using “last observation carried forward”, and observed cases and sensitivity analysis revealed concordant results. The parallel-group design used in the EXPRESS trial is more suitable for a chronic progressive condition than the cross-over design used in the donepezil study. We wonder why the EXPRESS study was classified as Class II evidence.
This also conflicts with the conclusions in a recent Cochrane meta-analysis. [4] The EXPRESS study also constituted the basis of regulatory approval by the European Medicines Agency and a recommendation for approval by a recent FDA Advisory Committee.
We would also like to highlight incomplete and misrepresentation of the rivastigmine data, with regard to the calculation of NNT and NNH. The number needed to treat and obtain clinically meaningful outcome was solely based on marked-moderate improvement and ignored the difference in the number of patients with marked-moderate worsening between placebo and rivastigmine (10% in favor of rivastigmine), which is relevant in a chronic progressive disorder. The NNT for a clinically meaningful outcome is 6 rather than 19.
With regard to NNH which was based on discontinuations the authors state that ".. eight patients must experience worsening of parkinsonism as assessed by the UPDRS for each patient experiencing clinically meaningful improvement, as measured by the ADC-CGIC." This is not correct, as there was no difference in the mean UPDRS scores between the two groups [2] and the majority of discontinuations were not due to worsesning of PD symptoms. The number of patients who discontinued because of worsening tremor was 1.7%, resulting in an NNH of 66.
Finally, we do not understand how a large RCT with favorable outcome in both primary and all secondary efficacy measures, and thorough assessment of safety parameters is given less weight than a small, cross-over study where the primary outcome measure was negative. We feel that correction of these errors would be important as AAN Practice Parameters are valued and relied upon worldwide.
References
1. Miyasaki JM, Shannon K, Voon V, et al. Practice Parameter: evaluation and treatment of depression, psychosis, and dementia in Parkinson disease (an evidence-based review): report of the Quality Standards Subcommittee of the American Academy of Neurology. Neurology 2006;66:996-1002.
2. Emre M, Aarsland D, Albanese A, et al. Rivastigmine for dementia associated with Parkinson's disease. N Engl J Med 2004;351:2509-2518.
3. Ravina B, Putt M, Siderowf A, et al. Donepezil for dementia in Parkinson's disease: a randomised, double blind, placebo controlled, crossover study. J Neurol Neurosurg Psychiatry 2005;76:934-939.
4. Maidment I, Fox C, Boustani M. Cholinesterase inhibitors for Parkinson's disease dementia. Cochrane Database Syst Rev 2006:CD004747.
Disclosure: The EXPRESS study mentioned in this Correspondence was sponsored by Novartis. Dr. Emre reports having served as a paid consultant for, having received grant support from, and having received lecture fees from Novartis, Pfizer, Eisai, and Lundbeck. Dr. Aarsland reports having served as a paid consultant and speaker for Pfizer, AstraZeneca, Eli Lilly, Eisai, Lundbeck, and Janssen and having received grant support from Novartis, Pfizer, and Janssen. Dr. Lees reports having served as a paid consultant and speaker for Novartis. Dr. Poewe reports having acted as a consultant for and received lecture fees from Novartis, Pfizer, and Lundbeck. Dr. Ballard has not been involved in any of the two studies discussed. He has received honoraria and research support from Novartis, AstraZeneca, Janssen-Cilag, Lundbeck and Pfizer.

Practice Parameter: Evaluation and treatment of depression, psychosis, and dementia in PD 17 August 2006

Erwin B. Montgomery,
University of Wisconsin - Madison
H6/538 CSC, 600 Higland Ave., Madison Wisconsin 53792
Send Correspondence to journal:
Re: Practice Parameter: Evaluation and treatment of depression, psychosis, and dementia in PD

montgomery{at}neurology.wisc.edu Erwin B. Montgomery

Would any physician treat an elderly depressed Parkinson’s disease patient with tricyclic antidepressants first rather than any of the numerous better tolerated alternatives? Would any physician commit a psychotic Parkinson’s disease patient to the logistical and economic costs and the risks of agranulocytosis associated with the use of clozapine without first trying quetiapine? Unlikely, yet these are the reasonable conclusions non-experts might draw from the practice parameters reported by Miyasaki et al. [1] The authors state, "Although the highest level of evidence is for amitriptyline, it is not necessarily the first choice...". Yet they do not offer any alternatives.
The authors could state that this practice parameter, and others recently published [5,6], only represent a review of the epistemic basis of certain practices based on a narrow, and perhaps arbitrary, interpretation of evidence based medicine and not actual recommendations for daily practice. The report does an excellent job of outlining the limitations of current evidence and that generalization to Parkinson’s disease patients from studies of non-Parkinson patients is problematic. However, to suggest the recommendations are not for actual practice would be disingenuous particularly in view of the promotion of these reviews as guidelines by the American Academy of Neurology.
The concern is that this application of evidence-based medicine could result in recommendations that are unreasonable. The particular application of evidenced-based medicine used a classification of evidence as a “filter” to disregard other forms of evidence. However, the original concept of evidenced-based medicine held that case reports and the consensus of experts also constitute evidence.
It is not clear a priori that randomized clinical trials necessarily trump other forms of evidence when there is disagreement. [7] Further discussion of the best use of evidence based medicine is needed rather than accepting current applications as a fait accompli. [7]
References

5. Suchowersky O, Reich S, Perlmutter J, Zesiewicz T, Gronseth G, Weiner WJ. Practice Parameter: Diagnosis and prognosis of new onset Parkinson disease (an evidence-based review): Report of the Quality Standards Subcommittee of the American Academy of Neurology. Neurology 2006;66:968-975.
6. Suchowersky O, Gronseth G, Perlmutter, J, Reich Zesiewicz T, Weiner WJ. Practice Parameter: Neuroprotective strategies and alternative therapies for Parkinson disease (an evidence-based review): Report of the Quality Standards Subcommittee of the American Academy of Neurology. Neurology 2006;66:976–982
7. Montgomery Jr. EB, Turkstra LS. Evidenced based medicine: let’s be reasonable. J Med Speech Lang Path 2003;11:ix-xii.
Disclosure: The author reports no conflicts of interest.

Reply from the Authors (to Aarsland et al and Montgomery) 17 August 2006

Janis M Miyasaki,
University of Toronto
Toronto Western Hospital, 399 Bathurst Street, 7 McL, Toronto Ontario M5T 2S8,
and Gary Gronseth on behalf of the author panel
Send Correspondence to journal:
Re: Reply from the Authors (to Aarsland et al and Montgomery)

miyasaki{at}uhnres.utoronto.ca Janis M Miyasaki, et al.

We thank Aarsland et al for their comments. The AAN classification of evidence scheme is used to measure a study's risk of bias (systematic error). Following pre-established, rules, we graded the rivastigmine study Class II since more than 20% of enrolled subjects dropped out. The smaller size of the donepezil study does not increase its risk of systamtic error. The smaller sample size does increase the risk of random error in the donepezil study reflected by the wider confidence intervals of the observed effect sizes in the donepezil study.
The rivastigmine study employed a last observation carried forward analysis (LOCF) in reporting efficacy. This technique uses the last available data point as the final data point for patients prematurely terminating from the study. Given that parkinson associated dementia is a progressive condition, the LOCF analysis will artifactually show less worsening in the treatment group with more drop outs (in this case rivastigmine). For this reason, rather than reporting the number needed to treat to prevent worsening, we chose to report the number needed to treat for clinically meaningful improvement.
The Number Needed to Harm was calculated using the published figures of "Parkinsonisn symptoms were reported as adverse events". Therefore the Number Needed to Harm (eight) is correct.
We also thank Dr. Montgomery for his comments. The practice parameters are evidence-based systematic reviews of the literature. Recommendations reflect the strength of evidence to support use, not a cookbook algorithm for care.
This parameter does not suggest that tricyclic antidepressants are the first-line choice for depression associated with Parkinson disease (PD) nor that clozapine is the first-line choice for psychosis associated with PD. However, clozapine is considered first line therapy for PD associated psychosis treatment by many knowledgeable experts outside of North America.
Further, randomized-controlled studies are available for many agents, but may not answer the relevant clinical questions. Therefore, such studies are not used to make recommendations. In these instances, the best evidence would be evidence potentially based on other study designs. Again, the parameter uses the highest quality, relevant information rather than searching only for randomized controlled studies.
We agree that the reviews are most helpful in filtering large volumes of evidence for the practicing physician. Guidelines do not replace individual physician judgement or the patient-physician collaboration in decision-making. Rather, AAN Practice Parameters are as the name implies: parameters within which neurologists and other physicians can make informed decisions. They are teaching tools and practice guides - not an autocratic rule book to restrain practice.
Each guideline has this acknowledged in the manuscript - "This statement is provided as an educational service of the American Academy of Neurology. It is based on an assessment of current scientific and clinical information. It is not intended to include all possible proper methods of care for a particular neurologic problem or all legitimate criteria for choosing to use a specific procedure. Neither is it intended to exclude any reasonable alternative methodologies. The AAN recognizes that specific patient care decisions are the prerogative of the patient and the physician caring for the patient, based on all of the circumstances involved."
Disclosure: The authors report no conflicts of interest.