The Moss Attention Rating Scale (MARS) was developed by John Whyte, MD, PhD, Tessa Hart, PhD, and colleagues at the Moss TBI Model System at MossRehab Hospital and Moss Rehabilitation Research Institute. Collaborators from other TBI Model System centers were also instrumental in testing the reliability of the MARS.
Attention deficits are nearly ubiquitous after traumatic brain injury, but can be hard to measure for several reasons.
First, “attention” is not a unitary construct but encompasses a number of cognitive processes from arousal and basic orienting to higher level processes overlapping with executive control—multitasking, error monitoring, etc.
Psychometric measures of some of these constructs are available, but an attention battery might not be feasible in the clinical setting, either because of time constraints or because patients are too impaired to undergo testing.
Traditional attention tests are typically highly structured, time-limited, and administered in an environment that minimizes distractions. Thus, they might not adequately measure clinically important dimensions of attention—such as resistance to distraction, the ability to make decisions about how to deploy one’s attention, or consistency of attention over time. One may make informal or bedside assessments of a patient’s attention in the real-world environment, but these might not be reliable, and do not allow for quantitative assessment of changes over time, or in response to treatment.
The MARS was designed as an observational rating scale to provide a reliable, quantitative and ecologically valid measure of attention-related behavior after TBI. Items were developed and refined using literature review and expert consultation, focus groups of expert TBI clinicians, and the results of two pilot studies (see Whyte et al., 2003).
A 45-item research version of the MARS was subjected to item response analysis (Whyte et al., 2003) and both exploratory and confirmatory factor analysis (Hart et al., 2006) to produce the current, 22-item MARS. Each item is a behavioral descriptor rated on a five-point Likert-type scale according to how well that behavior describes the patient, ranging from “definitely true” to “definitely false.” Items are phrased so that the rater considers behaviors indicative of both good and impaired attention.
Half the items relate to the three correlated factors found for the MARS: Restlessness/ Distractibility, Initiation, and Sustained/ Consistent Attention (Hart et al., 2006) and may be used to compute factor scores. The total raw MARS score may be converted to an interval metric from 0-100 (see Scoring section for how to compute factor and logit scores).
Since the MARS is based on observable behavior it is not appropriate for use with patients in the minimally conscious or vegetative states; validation studies thus far have been restricted to patients at Rancho level IV or higher. Item response analysis suggested that the “difficulty level” of the MARS is most appropriate for persons with moderate to severe attention deficits (Whyte et al., 2003). It has been tested thus far in inpatient TBI rehabilitation settings.
The MARS is designed to be completed by clinicians on the basis of their everyday, routine interaction with the patient; no special tests or questions are administered. In studies thus far, clinicians have been asked to think of the 2 or 3 previous days in which they have treated/ interacted with the patient to form their ratings. It is not known whether ratings based on a shorter period of time would show equivalent reliability.
There is no manual for administering the MARS because it was developed to be usable by people from a variety of disciplines and backgrounds and to provide a simple, rapid assessment of attention behavior. The only special instruction to raters is not to skip any items, because doing so prevents calculating a score. If an item mentions a behavior that the rater has not observed directly during the rating period, s/he should extrapolate from other observations of the patient to make a best guess about how that item should be rated. For example, patients who are entirely non-verbal must still be rated on the item “Tends to speak less than he/ she is capable of.” The scale has shown adequate reliability with a variety of patients under these instruction conditions.
A focus group of clinicians from a variety of rehabilitation disciplines who were experienced MARS users revealed that nurses given the MARS found it easier to complete their ratings when they “scheduled” a specific time for themselves to spend with (or observing) the patient, e.g., a half hour over lunch, time spent passing meds, etc., to make their ratings more comparable to those of team members who had specific times scheduled with the patient. Team members agreed that it was most useful to balance their ratings by considering not the best nor the worst attention observed over the rating interval, rather trying to capture the average level of attention over the proscribed time period.
Each item is rated on a 5-point Likert type scale indicating the degree to which the item describes the patient’s behavior. Since items include both positive and negative behaviors, some items must be “flipped” in direction before any scores are summed. The total, raw MARS score is the sum of the 22 items (ranging from 22-110), with higher scores indicating better attention.
Half of the MARS items may also be used to compute 3 factor scores: Restlessness/ Distractibillity (items 1, 10, 12, 17, 22), Initiation (items 7, 13, 19), and Sustained/ Consistent Attention (items 6, 14, 15). Mean item ratings must be used to compare factor scores due to the unequal number of items that compose the factors.
The attached worksheets (MARS Scoring Worksheet.xls) allow for semi-automatic MARS scoring. The first tab aligns reverse-scored and direct-scored items into the same scoring direction and calculates the total raw MARS score and mean item Factor scores. The second tab provides a Table for looking up the logit scores that correspond to raw total MARS scores. These logit scores convert the MARS to an interval scale for parametric manipulations.
MARS Rating Form
The MARS rating form includes the instructions, date, rater, and items. It does not include information on which items are reverse scored and which items contribute to specific factors. The reason for this is that such information, if included on the rating sheet, might bias the rater.
You can download the MARS in Portable Document Format (PDF) and several other useful documents below.
MARS Frequently Asked Questions
1. What if a MARS item mentions a behavior I have never directly observed? For example, how do I score the items that have to do with verbal behavior if the patient is non-verbal or aphasic?
We have always encouraged clinicians to “take their best guess” under these conditions, rather than leave an item blank. The scale has been shown to be reliable across disciplines under this instructional set.
2. Can I use the MARS with an outpatient sample?
There is no reason not to use the MARS with outpatients although validation studies to date have been on rehabilitation inpatients. We would be very interested to know about outpatient data on the MARS.
3. Can I use the MARS with patients who did not have a TBI?
There is no reason why not, although again, validation studies have been done with samples having TBI only. Again, we would be very interested in MARS data collected on other populations.
MARS Training & Testing
There is no established procedure for training on the MARS and indeed, the scale was developed to require minimal training in its use and thus to be “user friendly” for people from a variety of disciplines.
The MARS has been subjected to two large sample studies of inter-rater reliability.
The first (Whyte, Hart, Bode & Malec, 2003) compared ratings of patients’ Occupational Therapists (OTs) and Physical Therapists (PTs) done independently over the same three-day window. Agreement was good (Pearson r = .64). When there were disagreements, OTs tended to rate attention as slightly worse than PTs.
The second study (Whyte, Hart, Ellis & Chervoneva, 2008) compared ratings across four disciplines: OT, PT, Speech Therapy (ST) and Nursing, using ratings conducted both early and late in the inpatient rehabilitation stay. Agreement across disciplines was fairly robust, with Nursing showing somewhat lower concordance with therapy disciplines. This was interpreted as possibly relating to the less structured and more variable observations available to nursing vs. therapy staff on an inpatient rehabilitation unit.
Hart, Whyte, Ellis, & Chervoneva (2009) showed that total scores on the MARS administered to rehabilitation inpatients in the subacute phase of TBI:
- were more strongly correlated to concurrent measures thought to assess attention (e.g., (Digit Span) than those thought to be less demanding of attention (e.g., grip strength)
- were more strongly correlated to Cognitive than Motor FIM scores;
- and predicted 1-year outcomes of TBI (Disability Rating Scale scores) better than a battery of psychometric measures of attention administered concurrently with the MARS.
These findings provide support for the MARS as a measure of cognitive function, specifically attention, which performs well as a predictor of outcome.
The MARS is highly sensitive to the effects of “natural recovery plus rehabilitation” in the subacute phase of TBI, i.e., MARS scores are significantly higher near rehabilitation discharge compared to rehabilitation admission several weeks earlier (Whyte et al., 2008). However, its sensitivity to treatments specifically targeting attention remains to be determined.
The MARS did not discriminate groups of inpatients with TBI with attention deficits who had received a three-day course of methylphenidate or placebo (Hart et al., 2009). However, it was unclear whether the MARS was insensitive or the treatment was not sufficiently powerful—e.g., too short, not strong enough to add-on to natural recovery-to show an effect.
Whyte J, Hart T, Ellis CA, Chervoneva. (2008). The Moss Attention Rating Scale for traumatic brain injury: further explorations of reliability and sensitivity to change. Arch Phys Med Rehabil, 89, 966-73. More information is available from PubMed at this link, PMID 18452747
Hart T, Whyte J, Millis S, Bode R, Malec J, Richardson RN, Hammond F. (2006). Dimensions of disordered attention in traumatic brain injury: further validations of the Moss Attention Rating Scale. Arch Phys Med Rehabil, 87, 647-55. More information is available from PubMed at this link, PMID 16635627
Whyte J, Hart T, Bode RK, Malec JF. (2003). The Moss Attention Rating Scale for traumatic brain injury: initial psychometric assessment. Arch Phys Med Rehabil, 84, 268-76. More information is available from PubMed at this link, PMID 12601660
Hart T, Whyte J, Ellis C, Chervoneva I. (2009). Construct validity of an attention rating scale for traumatic brain injury. Neuropsychology, 23, 729-35. More information is available from PubMed at this link, PMID 19899831
The MARS download consists of the following 3 files:
- MARS Introduction.doc
- MARS Test Form.pdf
- MARS Scoring Worksheets.xls
- Download MARS Japanese translation (PDF: 90kb / 2 pages)
- Download MARS Italian translation (PDF: 84kb / 2 pages)