Section Editor
®
WriteClick Editor’s Choice
Robert C. Griggs, MD
Editors’ Note: In WriteClick this week, Dr. Kawada questions whether there were a sufficient number of events in the article “Predictors of dementia in Parkinson disease: A prospective cohort study” to ensure stable statistical estimates. Authors Anang and Postuma detail their statistical methodology. In reference to “Mirror therapy in unilateral neglect after stroke (MUST trial): A randomized controlled trial,” student doctors Ng and Singh and authors Pandian et al. discuss the cortical mechanisms underlying the efficacy of mirror therapy in unilateral neglect, phantom limb pain, and complex regional pain syndrome. —Megan Alcauskas, MD, and Robert C. Griggs, MD
PREDICTORS OF DEMENTIA IN PARKINSON DISEASE: A PROSPECTIVE COHORT STUDY
Tomoyuki Kawada, Tokyo, Japan: I read with interest the follow-up study by Anang et al.1 conducted to determine predictors of dementia in Parkinson disease. The authors used logistic regression analysis adjusting for disease duration, follow-up duration, age, and sex. I have some concerns. Among the 80 participating patients, 27 developed dementia. Odds ratios (ORs) (95% confidence intervals [CI]) of mild cognitive impairment and REM sleep behavior disorder were 18.13 (3.96– 83.05) and 49.7 (7.4–333.5), respectively. In general, the number of events per independent variable for logistic regression analysis should be kept greater than 10 for stable estimates.2,3 As the number of independent variables was 5 in their study, 50 events are needed. Although there have been opposing opinions,4 wide ranges of 95% CI of OR reflect insufficient number of events. Many significant predictors in tables 1–3 should also be used for the adjustment in regression model. Furthermore, the difference in Unified Parkinson’s Disease Rating Scale (UPDRS) 1.1 (cognition)5 between the 2 groups illustrates that at a baseline study free of dementia should be handled with caution. The authors should also consider the Cox regression analysis with follow-up duration as time functions. In any case, more events are needed for stable statistical estimates.
Author Response: Julius B.M. Anang, Ronald Postuma, Montreal, Canada: The authors thank Dr. Kawada for his interest in our study.1 The point regarding number of variables is valid, although not as straightforward as a simple 10:1 rule. There is no overall agreement on mandatory number of events, and other sources recommend 5–10 cases per variable.2,6 CI becomes narrower as sample size grows, but even at the lower limits of the 95% CI, there were still 4- to 7-fold increases in dementia risk for REM sleep behavior disorder and mild cognitive impairment, so our findings remain robust. With regards to covariates, many variables cosegregate and are highly correlated. We already adjusted for 4 variables, and it is statistically inappropriate to adjust for all predictors indiscriminately. This would only result in a loss of power and increase colinearity. The decision to perform multiple adjustments should be based on clinical reasoning or inference. We were not trying to understand a causal relationship (which requires strict adjustment for covariates); rather, we were interested in understanding all factors that are associated with risk of dementia. UPDRS 1.1 assesses patient recognition of intellectual impairment, including mild cognitive symptoms. We extensively assessed the presence of cognitive impairment on detailed neuropsychological examination, ruling out dementia, and providing a much more accurate potential covariate. Adjusting for baseline cognition is probably inappropriate to our research question. With regards to Cox regression, logistic regression and Cox hazards models are both used in such analyses. However, Cox proportional hazards depend on a reliable estimation of a time-dependent covariate, specifically an interval between baseline and dementia onset. We were seeing patients at 2 time points, often several years apart. We cannot know precisely when dementia started (and patients/caregivers could not provide information 100% reliably), potentially biasing results. Finally, as mentioned in our article, the size of the cohort and duration of follow-up are limitations, but this was the first prospective study of this nature looking extensively at both motor and nonmotor variables as predictors (and a 4.2-year follow-up of 80 patients is reasonably powered). Power issues mainly affect interpretation of negative results, and did not directly impact Neurology 84
March 24, 2015
1285
ª 2015 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.
the validity of our positive findings. We look forward to further studies to confirm our findings, and to discover new predictors of dementia. © 2015 American Academy of Neurology 1.
2.
3.
4.
5.
6.
Anang JB, Gagnon JF, Bertrand JA, et al. Predictors of dementia in Parkinson disease: a prospective cohort study. Neurology 2014;83:1253–1260. Peduzzi P, Concato J, Kemper E, et al. A simulation study of the number of events per variable in logistic regression analysis. J Clin Epidemiol 1996;49:1373–1379. Bagley SC, White H, Golomb BA. Logistic regression in the medical literature: standards for use and reporting, with particular attention to one medical domain. J Clin Epidemiol 2001;54:979–985. Vittinghoff E, McCulloch CE. Relaxing the rule of ten events per variable in logistic and Cox regression. Am J Epidemiol 2007;165:710–718. Scanlon BK, Katzen HL, Levin BE, et al. A formula for the conversion of UPDRS-III scores to Hoehn and Yahr stage. Parkinsonism Relat Disord 2008;14:379–380. Courvoisier DS, Combescure C, Agoritsas T, GayetAgeron A, Perneger TV. Performance of logistic regression modeling: beyond the number of events per variable, the role of data structure. J Clin Epidemiol 2011;64:993–1000.
MIRROR THERAPY IN UNILATERAL NEGLECT AFTER STROKE (MUST TRIAL): A RANDOMIZED CONTROLLED TRIAL
Melissa J. Ng, Prateush Singh, Cambridge, UK: Pandian et al.1 reported that mirror therapy improves unilateral neglect in stroke patients. Mirror therapy has also been shown effective in the treatment of complex regional pain syndrome and phantom limb pain.2 The mechanisms driving these positive effects are unclear. Chan et al.3 proposed that the mechanisms are similar to those suggested by Pandian et al., namely the activation of mirror neurons in the contralateral hemisphere or modification of cortical activity via visual feedback of the imagined action. Other theories of why mirror therapy is effective in conditions with no apparent peripheral cause for pain (such as complex regional pain syndrome and phantom limb pain) suggest that visual illusion of normal, pain-free movement results in cortical reorganization of the homunculus, which reduces pain, or that it corrects the mismatch among motor control, proprioception, and vision.2
It would be interesting to reconcile the cortical mechanisms driving effectiveness of mirror therapy in pain syndromes and neglect in stroke. If this is achieved, mirror therapy could be more widely applied to other cortical conditions aiming to decrease pain or to improve motor function. Author Response: Jeyaraj D. Pandian, Rajni Arora, Paramdeep Kaur, Ludhiana, India: The authors read with interest the comments by Ng and Singh about the effectiveness of mirror therapy in treating unilateral neglect and also about the underlying mechanisms in regards to our article.1 Mirror therapy was first used in the treatment of phantom limb pain.4 The principle of mirror therapy is based on the mirror neuron system. In animal experiments, the mirror neuron system discharges when an animal performs a goal-directed hand action and also when it observes someone else performing a similar action.5 The activation of mirror neuron system leads to recruitment of functionally interconnected cortical structures coupling action execution and observation.6 In the present report, patients were asked to look into a mirror and practice nonparetic side wrist and finger flexion and extension movements, watching the image of their unaffected hand. When the nonparetic limb is engaged during motor training, crossed facilitatory drive from the intact hemisphere will give rise to increased excitability in the homologous motor pathways of the paretic limb, facilitating recovery of function. Similar mechanisms have been postulated in the alleviation of complex regional pain syndrome. © 2015 American Academy of Neurology 1.
2.
3. 4. 5. 6.
Pandian JD, Arora R, Kaur P, et al. Mirror therapy in unilateral neglect after stroke (MUST trial): a randomized controlled trial. Neurology 2014;83:1012–1017. Al Sayegh S, Filen T, Johansson M, et al. Mirror therapy for complex regional pain syndrome (CRPS): a literature review and an illustrative case report. Scand J Pain 2013;4:200–207. Chan BL, Witt R, Charrow AP, et al. Mirror therapy for phantom limb pain. N Engl J Med 2007;357:2206–2207. Ramachandran VS, Rogers-Ramachandran D, Cobb S. Touching the phantom limb. Nature 1995;377:489–490. Buccino G, Binkofski F, Riggio L. The mirror neuron system and action recognition. Brain Lang 2004;89:370–376. Carson RG. Neural pathways mediating bilateral interactions between the upper limbs. Brain Res Brain Res Rev 2005;49:641–662.
Author disclosures are available upon request ([email protected]). 1286
Neurology 84
March 24, 2015
ª 2015 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.
Predictors of dementia in Parkinson disease: A prospective cohort study Tomoyuki Kawada, Julius B.M. Anang and Ronald Postuma Neurology 2015;84;1285-1286 DOI 10.1212/WNL.0000000000001408 This information is current as of March 23, 2015 Updated Information & Services
including high resolution figures, can be found at: http://www.neurology.org/content/84/12/1285.full.html
References
This article cites 6 articles, 2 of which you can access for free at: http://www.neurology.org/content/84/12/1285.full.html##ref-list-1
Permissions & Licensing
Information about reproducing this article in parts (figures,tables) or in its entirety can be found online at: http://www.neurology.org/misc/about.xhtml#permissions
Reprints
Information about ordering reprints can be found online: http://www.neurology.org/misc/addir.xhtml#reprintsus
Neurology ® is the official journal of the American Academy of Neurology. Published continuously since 1951, it is now a weekly with 48 issues per year. Copyright © 2015 American Academy of Neurology. All rights reserved. Print ISSN: 0028-3878. Online ISSN: 1526-632X.
®
WriteClick Editor’s Choice
Robert C. Griggs, MD
Editors’ Note: In WriteClick this week, Dr. Kawada questions whether there were a sufficient number of events in the article “Predictors of dementia in Parkinson disease: A prospective cohort study” to ensure stable statistical estimates. Authors Anang and Postuma detail their statistical methodology. In reference to “Mirror therapy in unilateral neglect after stroke (MUST trial): A randomized controlled trial,” student doctors Ng and Singh and authors Pandian et al. discuss the cortical mechanisms underlying the efficacy of mirror therapy in unilateral neglect, phantom limb pain, and complex regional pain syndrome. —Megan Alcauskas, MD, and Robert C. Griggs, MD
PREDICTORS OF DEMENTIA IN PARKINSON DISEASE: A PROSPECTIVE COHORT STUDY
Tomoyuki Kawada, Tokyo, Japan: I read with interest the follow-up study by Anang et al.1 conducted to determine predictors of dementia in Parkinson disease. The authors used logistic regression analysis adjusting for disease duration, follow-up duration, age, and sex. I have some concerns. Among the 80 participating patients, 27 developed dementia. Odds ratios (ORs) (95% confidence intervals [CI]) of mild cognitive impairment and REM sleep behavior disorder were 18.13 (3.96– 83.05) and 49.7 (7.4–333.5), respectively. In general, the number of events per independent variable for logistic regression analysis should be kept greater than 10 for stable estimates.2,3 As the number of independent variables was 5 in their study, 50 events are needed. Although there have been opposing opinions,4 wide ranges of 95% CI of OR reflect insufficient number of events. Many significant predictors in tables 1–3 should also be used for the adjustment in regression model. Furthermore, the difference in Unified Parkinson’s Disease Rating Scale (UPDRS) 1.1 (cognition)5 between the 2 groups illustrates that at a baseline study free of dementia should be handled with caution. The authors should also consider the Cox regression analysis with follow-up duration as time functions. In any case, more events are needed for stable statistical estimates.
Author Response: Julius B.M. Anang, Ronald Postuma, Montreal, Canada: The authors thank Dr. Kawada for his interest in our study.1 The point regarding number of variables is valid, although not as straightforward as a simple 10:1 rule. There is no overall agreement on mandatory number of events, and other sources recommend 5–10 cases per variable.2,6 CI becomes narrower as sample size grows, but even at the lower limits of the 95% CI, there were still 4- to 7-fold increases in dementia risk for REM sleep behavior disorder and mild cognitive impairment, so our findings remain robust. With regards to covariates, many variables cosegregate and are highly correlated. We already adjusted for 4 variables, and it is statistically inappropriate to adjust for all predictors indiscriminately. This would only result in a loss of power and increase colinearity. The decision to perform multiple adjustments should be based on clinical reasoning or inference. We were not trying to understand a causal relationship (which requires strict adjustment for covariates); rather, we were interested in understanding all factors that are associated with risk of dementia. UPDRS 1.1 assesses patient recognition of intellectual impairment, including mild cognitive symptoms. We extensively assessed the presence of cognitive impairment on detailed neuropsychological examination, ruling out dementia, and providing a much more accurate potential covariate. Adjusting for baseline cognition is probably inappropriate to our research question. With regards to Cox regression, logistic regression and Cox hazards models are both used in such analyses. However, Cox proportional hazards depend on a reliable estimation of a time-dependent covariate, specifically an interval between baseline and dementia onset. We were seeing patients at 2 time points, often several years apart. We cannot know precisely when dementia started (and patients/caregivers could not provide information 100% reliably), potentially biasing results. Finally, as mentioned in our article, the size of the cohort and duration of follow-up are limitations, but this was the first prospective study of this nature looking extensively at both motor and nonmotor variables as predictors (and a 4.2-year follow-up of 80 patients is reasonably powered). Power issues mainly affect interpretation of negative results, and did not directly impact Neurology 84
March 24, 2015
1285
ª 2015 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.
the validity of our positive findings. We look forward to further studies to confirm our findings, and to discover new predictors of dementia. © 2015 American Academy of Neurology 1.
2.
3.
4.
5.
6.
Anang JB, Gagnon JF, Bertrand JA, et al. Predictors of dementia in Parkinson disease: a prospective cohort study. Neurology 2014;83:1253–1260. Peduzzi P, Concato J, Kemper E, et al. A simulation study of the number of events per variable in logistic regression analysis. J Clin Epidemiol 1996;49:1373–1379. Bagley SC, White H, Golomb BA. Logistic regression in the medical literature: standards for use and reporting, with particular attention to one medical domain. J Clin Epidemiol 2001;54:979–985. Vittinghoff E, McCulloch CE. Relaxing the rule of ten events per variable in logistic and Cox regression. Am J Epidemiol 2007;165:710–718. Scanlon BK, Katzen HL, Levin BE, et al. A formula for the conversion of UPDRS-III scores to Hoehn and Yahr stage. Parkinsonism Relat Disord 2008;14:379–380. Courvoisier DS, Combescure C, Agoritsas T, GayetAgeron A, Perneger TV. Performance of logistic regression modeling: beyond the number of events per variable, the role of data structure. J Clin Epidemiol 2011;64:993–1000.
MIRROR THERAPY IN UNILATERAL NEGLECT AFTER STROKE (MUST TRIAL): A RANDOMIZED CONTROLLED TRIAL
Melissa J. Ng, Prateush Singh, Cambridge, UK: Pandian et al.1 reported that mirror therapy improves unilateral neglect in stroke patients. Mirror therapy has also been shown effective in the treatment of complex regional pain syndrome and phantom limb pain.2 The mechanisms driving these positive effects are unclear. Chan et al.3 proposed that the mechanisms are similar to those suggested by Pandian et al., namely the activation of mirror neurons in the contralateral hemisphere or modification of cortical activity via visual feedback of the imagined action. Other theories of why mirror therapy is effective in conditions with no apparent peripheral cause for pain (such as complex regional pain syndrome and phantom limb pain) suggest that visual illusion of normal, pain-free movement results in cortical reorganization of the homunculus, which reduces pain, or that it corrects the mismatch among motor control, proprioception, and vision.2
It would be interesting to reconcile the cortical mechanisms driving effectiveness of mirror therapy in pain syndromes and neglect in stroke. If this is achieved, mirror therapy could be more widely applied to other cortical conditions aiming to decrease pain or to improve motor function. Author Response: Jeyaraj D. Pandian, Rajni Arora, Paramdeep Kaur, Ludhiana, India: The authors read with interest the comments by Ng and Singh about the effectiveness of mirror therapy in treating unilateral neglect and also about the underlying mechanisms in regards to our article.1 Mirror therapy was first used in the treatment of phantom limb pain.4 The principle of mirror therapy is based on the mirror neuron system. In animal experiments, the mirror neuron system discharges when an animal performs a goal-directed hand action and also when it observes someone else performing a similar action.5 The activation of mirror neuron system leads to recruitment of functionally interconnected cortical structures coupling action execution and observation.6 In the present report, patients were asked to look into a mirror and practice nonparetic side wrist and finger flexion and extension movements, watching the image of their unaffected hand. When the nonparetic limb is engaged during motor training, crossed facilitatory drive from the intact hemisphere will give rise to increased excitability in the homologous motor pathways of the paretic limb, facilitating recovery of function. Similar mechanisms have been postulated in the alleviation of complex regional pain syndrome. © 2015 American Academy of Neurology 1.
2.
3. 4. 5. 6.
Pandian JD, Arora R, Kaur P, et al. Mirror therapy in unilateral neglect after stroke (MUST trial): a randomized controlled trial. Neurology 2014;83:1012–1017. Al Sayegh S, Filen T, Johansson M, et al. Mirror therapy for complex regional pain syndrome (CRPS): a literature review and an illustrative case report. Scand J Pain 2013;4:200–207. Chan BL, Witt R, Charrow AP, et al. Mirror therapy for phantom limb pain. N Engl J Med 2007;357:2206–2207. Ramachandran VS, Rogers-Ramachandran D, Cobb S. Touching the phantom limb. Nature 1995;377:489–490. Buccino G, Binkofski F, Riggio L. The mirror neuron system and action recognition. Brain Lang 2004;89:370–376. Carson RG. Neural pathways mediating bilateral interactions between the upper limbs. Brain Res Brain Res Rev 2005;49:641–662.
Author disclosures are available upon request ([email protected]). 1286
Neurology 84
March 24, 2015
ª 2015 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.
Predictors of dementia in Parkinson disease: A prospective cohort study Tomoyuki Kawada, Julius B.M. Anang and Ronald Postuma Neurology 2015;84;1285-1286 DOI 10.1212/WNL.0000000000001408 This information is current as of March 23, 2015 Updated Information & Services
including high resolution figures, can be found at: http://www.neurology.org/content/84/12/1285.full.html
References
This article cites 6 articles, 2 of which you can access for free at: http://www.neurology.org/content/84/12/1285.full.html##ref-list-1
Permissions & Licensing
Information about reproducing this article in parts (figures,tables) or in its entirety can be found online at: http://www.neurology.org/misc/about.xhtml#permissions
Reprints
Information about ordering reprints can be found online: http://www.neurology.org/misc/addir.xhtml#reprintsus
Neurology ® is the official journal of the American Academy of Neurology. Published continuously since 1951, it is now a weekly with 48 issues per year. Copyright © 2015 American Academy of Neurology. All rights reserved. Print ISSN: 0028-3878. Online ISSN: 1526-632X.
