Brief communicationA simple measure of neglect severity
Introduction
Since being introduced by Denny-Brown (1963) and popularized by Albert (1973), cancellation tasks have become one of the most widespread methods for diagnosing spatial neglect. Classically, the individual is shown a piece of paper with a cluttered array of items, and asked to mark all of the target items, while ignoring other distractors. The prevalence of these tests stems from many factors including the ease of describing the task to participants in a clinical situation and speed of administration (a couple of minutes). In addition, cancellation tests appear to be strong predictors of other clinical manifestations of neglect such as errors on copying or drawing tasks and biased spontaneous exploratory behavior (Ferber & Karnath, 2001).
Despite widespread usage, the interpretation of these tests has remained somewhat arbitrary. Often, these tests are used as a simple binary classifier to detect the presence or absence of neglect. However, behavioral performance of neglect patients as well as visual inspection of cancellation tests suggests that this disorder exhibits a continuous spectrum of severity desiring quantification.
One simple method that has been used to derive a continuous measure from the cancellation task is to count the number of target omissions (or hits) and use this value as a measure of neglect severity. An illustration of this approach is the Behavioral Inattention Test (BIT, Wilson, Cockburn, & Halligan, 1987). Unfortunately, the standardized scoring system of the BIT does not reflect lateralized bias. As a result, this analysis cannot distinguish between spatially biased performance versus inattentive performance. For example, some patients may miss items specifically on the contralesional side of the test sheet whereas others may miss the same number of targets but evenly distributed across the sheet. While the first observation is indicative of spatial neglect; the latter does not support this diagnosis because it does not exhibit the spatial bias that is so unique to spatial neglect. Therefore, the simple number of omissions (or hits) alone is not an unambiguous measure of the disorder. Differential diagnosis on this measure still requires an experienced observer evaluating the pattern of omissions produced in addition.
To deal with this problem, Halligan et al. (1991) proposed a continuous measure that should reflect the spatial bias observed in spatial neglect. They suggested using Friedman's (1992) lateralization index to measure spatial biases in cancellation tasks (see van Kessel, van Nes, Brouwer, Geurts, & Fasotti, 2010 for a similar measure). This ratio reflects the number of targets detected on the left half of the test divided by the total number of targets detected (so the score ranges from zero to one with values near 0.5 suggesting unbiased performance). However, this measure may not be a particularly reliable measure of severe neglect. For example, consider two patients, one who misses all items on the left half of the page, versus another more severe patient who misses items on the left three-quarters of the test. Both patients receive a score of 0 even though one appears to have a more severe deficit. Further, this index can even misclassify the relative severity of patients. For example, consider two patients, one who misses all but one item on the left half of the page, versus another more severe patient who misses all but one item on the left half of the page as well as most of the leftmost items on the right side of the page. In this case, the more severe patient will get a less severe laterality index. Moreover, it should be noted that Halligan et al. (1991) actually described the laterality index as a binary classifier for neglect, and a continuous classifier for milder biased inattention. According to their method, for a patient to be considered to have spatial neglect they must have a laterality index of zero, missing all of the items on the left side of the page. Therefore, the laterality index does not actually attempt to gauge the severity of neglect.
Chatterjee and colleagues have suggested using power functions (1992) and later logistic regression (1999) to provide parametric analyses of cancellation tasks. In this context, the logistic regression attempts to model the probability of detecting targets across a continuous variable (such as horizontal position of each target) using a sigmoid function. This sigmoid curve is described by three parameters, one that determines the steepness of slope (how quickly performance improves) a constant and finally a goodness-of-fit. The utility of this approach is that one can model multiple interacting factors, for example the influence of both near-far as well as left-right position of a target. However, the output of the logistic regression cannot be easily quantified as a single intuitive measure that can be used as an index of neglect severity and used in subsequent analyses such as lesion-behavior mapping (Rorden et al., 2007). In theory, for a single variable (horizontal position of targets), one can determine the 50% crossing point (e.g. the horizontal location where the participant detects half of all targets.) However, the range of this value can be unintuitive (this hypothetical location could be outside the bounds of the paper test) and this value will not be meaningful in individuals where the model is not accurately fitting the data (e.g. patients without neglect or with non-spatial attentional deficits leading to errors evenly distributed across space). Therefore, while this method is useful for sophisticated analyses, it is not suitable for most studies as well as daily clinical usage.
Mark and Monson (1997) suggested measuring the ‘neglect center’ – that is, the center of mass for the neglected items. Specifically, this allowed them to calculate both the angle and distance from the center of the display to the center of the neglected items. The intention of this measure was to examine changes in the direction of neglect, and the authors noted that this measure is not a direct measure of neglect severity. For example, consider a patient who misses three targets directly to the left of the display center at −6, −4, and −2 cm. This patient receives an identical neglect center score as a less severe patient or even a healthy subject who only misses a single target located at −4 cm. This feature limits the utility of this measure.
In a paper focused on the anatomy of neglect, Binder, Marshall, Lazar, Benjamin, & Mohr (1992) described a clever method for measuring the severity of neglect. Specifically, they measured the mean horizontal location of the cancelled items. At first glance, this ‘center of cancellation’ (CoC) appears to offer the same information as Mark and Monson's ‘center of neglect’ measure. However, the CoC provides a direct measure of neglect severity, and it is capable of distinguishing between the two patients described in the previous paragraph. In the case of the (healthy) individual who simply forgot to mark a single target, the ‘center of neglect’ will be pulled all the way to this location, which might be on the extreme edge of the test. In contrast, the CoC for this individual is heavily weighted by the (large) number of correct marks, and therefore will not be substantially different from the center of the display.
Curiously, Binder et al.’s attractive measure for cancellation tasks has not been widely adopted. We speculate that four factors may have contributed to this. First, the paper by Binder and colleagues was focused on the anatomical differences between individuals with biased performance on the line bisection task relative to cancellation tasks, with no discussion regarding the novel method for measuring cancellation. Second, computing the CoC score by hand as they did is time consuming and the authors provided no means for replicating their results. Third, they did not provide any normative data for interpreting this measure. Finally, even though the CoC is a continuous measure, Binder and colleagues used it as a binary classifier to merely identify the presence or absence of neglect.
Our aim is to further develop this measure and directly address these factors. We developed an easy to use piece of software, which allows the scientist or clinician to intuitively enter data from (paper-and-pencil based) cancellation tasks in less than one minute. This program calculates a calibrated center of cancellation statistic that is easy to interpret. With this measure, healthy individuals as well as individuals who omit items with an unbiased spatial distribution (identifying as many objects on both sides of the test) score near zero, while scores approaching positive or negative one indicate left- or right-sided neglect, respectively. This continuous score is sensitive to both the number of omissions as well as the location of these omissions. Therefore, this score will not generate some of the paradoxical results seen from other measures such as the laterality index. In our previous thought example, consider two hypothetical patients who each only find one item on the left half of the page: the less severe patient sees all items on the right half of the page and has a score near 0.5 (mean for targets on the right half), whereas the more severe patient who misses all other targets on the left 3/4 of the page will receive a score near 0.75 (mean for targets on the right 1/4). Fig. 1 shows performance for a patient who scores near 0.5 on this measure. The range of the CoC index is easy to interpret and is similar to popular measures such as correlation ratios.
While our software can be trained on any cancellation task, we present normative data on two popular cancellation tasks: the Bells Test (Gauthier, Dehaut, & Joanette, 1989) and the Letter Cancellation Task (Weintraub & Mesulam, 1985).
Section snippets
Methods
We re-evaluated data from 110 stroke patients with focal right-hemisphere brain lesions admitted to the Center of Neurology at Tübingen Univesity, Germany (Table 1). The lesions were demonstrated by magnetic resonance imaging (MRI) or by computed tomography (CT). Patients with diffuse or bilateral brain injury, patients with tumors, as well as patients in whom MRI or CT scans revealed no obvious lesion were excluded.
We examined the performance of patients on three traditional paper-and-pencil
Results
Our first analysis was designed to see if the CoC is a robust continuous measure of neglect severity. To accomplish this, we examined each individual's CoC scores on two clinical tests. Across both groups, an individual's CoC for the Letter Cancellation Task was significantly correlated with his/her CoC for the Bells Test, R = 0.9387, p < 0.0001 (t = 28.31, DF = 108), as shown in Fig. 2A. The slope of this correlation was 0.8661. While this result demonstrates that CoC performance on one cancellation
Discussion
We examined the CoC for both the Bells Test and the Letter Cancellation Task in a large group of 110 individuals who had acute right-hemisphere stroke. We used independent binary classifiers from two clinical tests to generate normative data for the CoC scores. These values proved accurate at detecting the neglect patients in our sample of acute stroke patients. Crucially, our data suggests that this measure is a robust continuous measure of neglect severity; we wish to emphasize the strong
Acknowledgements
This work was supported by the National Institutes of Health (R01 NS054266) and the Bundesministerium für Bildung und Forschung (BMBF-Verbund 01GW0641 “Räumliche Orientierung”) and the Deutsche Forschungsgemeinschaft (KA 1258/10-1). We are grateful to Johannes Rennig and Sabrina Oelschlaeger, both Center of Neurology Tuebingen for their help with Table 1 and with entering data by using the new CoC software.
References (26)
- et al.
A stimulus-response relationship in unilateral neglect: The power function
Neuropsychologia
(1992) - et al.
Quantitative analysis of cancellation tasks in neglect
Cortex
(1999) - et al.
Two-dimensional cancellation neglect – A review and suggested method of analysis
Cortex
(1997) - et al.
Is the posner reaction time test more accurate than clinical tests in detecting left neglect in acute and chronic stroke?
Archives of Physical Medicine and Rehabilitation
(2009) - et al.
Disturbed line bisection is associated with posterior brain lesions
Brain Research
(2006) - et al.
Task-related modulation of visual neglect in cancellation tasks
Neuropsychologia
(2009) - et al.
Visuospatial asymmetry and non-spatial attention in subacute stroke patients with and without neglect
Cortex
(2010) A simple test of visual neglect
Neurology
(1973)- et al.
Sensitivity of clinical and behavioural tests of spatial neglect after right hemisphere stroke
Journal of Neurology Neurosurgery and Psychiatry
(2002, August) - et al.
Distinct syndromes of hemineglect
Archives of Neurology
(1992)
The physiological basis of perception and speech
Assessment of spatial attention after brain damage with a dynamic reaction time test
Journal of the International Neuropsychological Society
How to assess spatial neglect-line bisection or cancellation tasks?
Journal of Clinical and Experimental Neuropsychology
Cited by (187)
Spatial Neglect Therapy With the Augmented Reality App “Negami” for Active Exploration Training: A Randomized Controlled Trial on 20 Stroke Patients With Spatial Neglect
2023, Archives of Physical Medicine and RehabilitationWord-centred neglect dyslexia as an inhibitional deficit: A single case study
2023, Neuropsychologia