(Stroke. 1999;30:1731-1733.)
© 1999 American Heart Association, Inc.
Letters to the Editor |
Three-Dimensional Vector Component Analysis of Neurological Stroke Scales
Non-invasive Neurocybernetic Flow Laboratory, International Institutes of Advanced Research and Training, Chidicon Medical Center, Owerri, Nigeria
 |
Introduction |
|---|
 Top Introduction ReferencesIntroduction References |
|---|
To the Editor:
Neurological Stroke Scales (NSS) are impairment measurement levels of different domains essentially including consciousness/orientation, motor power, and speech/verbal communication. These domains reflect global effects of stroke and changes in specific vascular territories. Orgogozo et al1 have advocated a Unified Form for Neurological Stroke Scales (UNSS), including the Scandinavian Neurological Stroke Scale (SNSS) and the Middle Cerebral Artery Neurological Scale (MCANS). However, monitoring patients' progress using the UNSS involves calculating cumulative scores from each domain. Comparison of patients using scores from the UNSS is rather difficult, because similar cumulative scores may represent widely different abnormalities in different domains. Changes in a patient's condition marked by progress in one domain but regression in another may not be reflected in the cumulative score. For example, consider 2 items on the UNSS, one measuring language, the other measuring motor power. Adding the 2 together is obviously like adding "apples and oranges." Doing this implies that a patient who is aphasic but has minor motor deficits is comparable overall with a person who is hemiplegic but not aphasic, given similar cumulative scores. Again, the problem with simply correlating 2 sums (for example, before treatment and after treatment) is that one might lose important information in the process, and in the worst case actually "lose" the treatment effects, thereby reducing the sensitivity of the UNSS (that is, its ability "to translate meaningful clinical changes into numerical differences that are statistically significant"1 ). The above examples show as well how cumulative scores also reduce the validity of the UNSS, that is, its ability to measure the dimension of interest. What is then required is a method to represent the magnitude and direction of change in the various domains of the UNSS.
We present here our findings in 16 patients, 10 men and 6 women of mean age 67±3.98 years, who suffered a stroke, as defined by WHO,2 of the middle cerebral artery and underwent a standardized neurological assessment with use of the UNSS1 on admission and at discharge, after a mean duration of hospitalization of 39.31±10.66 days. The scores for items on SNSS and MCANS reflecting global severity (consciousness/orientation), motor power, and speech/verbal communication were averaged between both scales and totaled for each domain and for each patient. The mean values for consciousness/orientation on admission and at discharge were 14.25±0.82 and 16.25±0.17, respectively (P=0.02 by t test). The motor power scores on admission and at discharge were 31.28±4.41 and 55.96±4.18, respectively (P=0.0002). The speech/verbal communication domain values on admission and discharge were 5.84±0.83 and 8.34±0.78, respectively (P=0.008).
The individual values were plotted on a 3-D scatterplot with a
statistical software package (Statistica, StatSoft, Inc). Scores were
plotted for consciousness/orientation on the z axis, motor
power on the x axis, and speech/verbal communication on the
y axis. This 3-coordinate system provided a unique placement
of each case, irrespective of the cumulative score. However, 3-D
rotation was applied to choose the view that best displayed the
scatterplot, with minimal visual overlap of data points. The 3-D
scatterplot shown in the
Figure
revealed the relationship between 3 domains: each point in the plot
represents 1 case on admission (small open circles) and at
discharge (large closed circles). The Figure revealed that the strength
of the relationship between consciousness/orientation, motor power, and
speech/verbal communication increased at discharge compared with that
on admission. This complex (interactive) relationship between these
variables would be almost impossible to identify in a numerical
exploratory data analysis of the cumulative scores. The 3-D
scatterplot allowed a magnitude and directional assessment of
patients' progress, hence a vector component analysis. It
demonstrated the efficacy of treatment modalities in specific domains.
For example, the need for continued physiotherapy and/or speech therapy
in 4 patients with low scores at discharge (4 large closed circles
among predominantly small open circles) is readily evident.
|
Therefore, we suggest the use of this approach we refer to as "3-D vector component analysis of neurological stroke scales" in further studies. However, how this approach will improve sensitivity and validity of NSS remains to be demonstrated in future stroke trials.
|
References |
|---|
|
TopIntroductionReferencesIntroduction References |
|---|
1. Orgogozo JM, Asplund K, Boysen G. A unified form for neurological scoring of hemispheric stroke with motor impairment. Stroke.. 1992;23:1678–1679.
2. World Health Organization. Proposal for the Multinational Monitoring of Trends and Determinants in Cardiovascular Disease (MONICA Project). Geneva, Switzerland: World Health Organization; 1983. WHO document MNC/82.1, Rev 1.
Response
Department of Neurology, University Hospital Pellegrin, INSERM Epidemiology and Statistics Research Unit U-330, University of Bordeaux II, Bordeaux, France
|
Introduction |
|---|
|
TopIntroductionReferencesIntroduction References |
|---|
The rationale to quantify neurological impairments for stroke outcome research, particularly in stroke trials, is that the neurological signs and symptoms (impairments) are the most direct basis to measure the consequences of stroke for prognostic and follow-up purposes. They correlate closely with the damage to cerebral structures (face validity) on the one hand, and they are the main determinant of the functional disabilities (concurrent validity) on the other. These neurological features can be assessed in a standard manner by use of neurological scales, which assign numerical (ordinal) values to the elementary deficits and add them as a cumulative score. Several stroke scales have been produced, most of which are based on normative scaling, ie, on an empirical (based on experience) choice and weighting of items.
The United Form for Neurological Scoring of Hemispheric Stroke With Motor Impairment (UNSS)1 was made as a collective effort to help comparisons between stroke trials using either the Scandinavian Stroke Scale (SSS) or the Middle Cerebral Artery Stroke Scale (MCAS), and also to promote harmonization in the assessment of neurological impairments in stroke trials. Despite the fact that use of the UNSS as a new scale was not suggested, later independent publications showed that its items have a high interrater reliability2 and that their combination is a valid measure of clinical consequences of both ischemic and hemorrhagic strokes.3
The interesting data and comments from Njemanze and Chidi-Ebere are a helpful contribution to the use and interpretation of impairment scales in stroke. Beyond the criticism that "similar cumulative scores may represent widely different abnormalities in different domains"—a view shared by the authors of the paper under discussion1 —the question of a better way to assess the usefulness of impairment scales is rarely addressed.
The responsiveness of another neurological scale (the NIHSS) was shown to be equivalent to that of disability and global scales in the only fully positive stroke trial,4 and the MCAS was found to be more responsive than the Barthel Index (disability) in the positive subgroup of another trial.5 Therefore, stroke scales seem to be valid and sensitive instruments by which to assess treatment effect in relatively large trials.
What the data of Njemanze and Chidi-Ebere show, in a limited series of 16 stroke patients, is that the UNSS is in fact a construct containing several dimensions. This was also shown in a larger series by Edwards et al,3 and also that the SSS and MCAS have slightly different structures, so that their combination in the UNSS results in a scale with different properties. The relevance of these findings depends on the purpose of using a stroke scale. In principal component analyses, when the first principal component (in most stroke scales, the motor component) accounts for the majority of the total variance, the scale can be considered unidimensional,6 and the weighting of items is then irrelevant. This allows its use for simple statistical comparisons between groups.
But if there is a rationale for comparing different treatment options in more homogenous groups of patients—such as selective physiotherapy in the example given—then a factorial analysis, such as the 3-D vector component method described, could be a better option than comparisons of cumulative scores. As the authors themselves state, this "remains to be demonstrated in future stroke trials," but this could also be assessed from the databases of already-completed stroke trials.
|
References |
|---|
|
TopIntroductionReferencesIntroduction References |
|---|
1. Orgogozo JM, Asplund K, Boysen G. A unified form for neurological scoring of hemispheric stroke with motor impairment. Stroke.. 1992;23:1678–1679.
2. Treves TA, Karepov VG, Aronovitch BD, Gorbulev AY, Bornstein NM, Korczyn AD. Interrater agreement in evaluation of stroke patients with the Unified Neurological Stroke Scale. Stroke.. 1994;25:1263–1264.[Abstract]
3.
Edwards DF, Chen Y-W, Diringer MN. Unified neurological
stroke scale is valid in ischemic and hemorrhagic stroke.
Stroke.. 1995;26:1852–1858.
4.
The NINDS rt-PA Stroke Study Group. Tissue
plasminogen activator for acute
ischemic stroke. N Engl J Med.. 1995;333:1581–1587.
5.
De Deyn PP, De Reuck J, Deberdt W, Vlietink R, Orgogozo
JM. Treatment of acute ischemic stroke with piracetam.
Stroke.. 1997;28:2347–2352.
6. Bebbington AC. Scaling indices of disablement. Br J Prev Soc Med.. 1977;31:122–126.[Medline] [Order article via Infotrieve]
This article has been cited by other articles:
 |
|
 |
|
  P. C. Njemanze, J. Anozie, and I. Okadike 3D Vector Component Analysis of the Modified National Institutes of Health Neurological Stroke Scale Stroke, December 1, 2001; 32(12): 2958 - 2960. [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||