Abstract
Keywords
Severe traumatic brain injury (TBI) is characterized by prolonged posttraumatic amnesia (PTA) and scores <8 on the Glasgow Coma Scale (GCS; Teasdale & Jennett, 1974). Behavioral assessments following severe TBI aim to elicit, observe, and record behaviors that indirectly indicate the presence of neural networks and pathways presumed to be prerequisite for consciousness (Dolce & Lucca, 2010). These behaviors are not, however, an on–off phenomenon but part of a continuum that is captured by these assessments. Assessments are administered often and usually over a long period of time and are therefore thought to be sensitive to small and subtle changes in patient behaviors. The ability to understand observations made during assessment and the fit of these observations into a sequence of recovery may help clinicians more accurately diagnose the type and level of disorders of consciousness (DOC) and make decisions regarding management and intervention.
Despite significant developments in the past 10 yr in understanding DOC, bedside assessment of neurobehavioral function remains problematic (Bruno, Vanhaudenhuyse, Thibaut, Moonen, & Laureys, 2011; Dolce & Lucca, 2010; Gosseries et al., 2011; Greenwald & Rigg, 2009; Seel et al., 2010). Early and accurate assessment remains a practical challenge given the variable nature of the condition and the slow and incremental nature of change to be observed in patients with severe TBI (Seel et al., 2010). Although brain imaging has revealed new insights and questions regarding neuronal structure and function during recovery (Schiff, 2010) and is recommended to assist in the clinical diagnosis of DOC (Coleman et al., 2009), bedside behavioral assessment scales are more commonly used and bear the “burden of proof for establishing conscious awareness in individuals who sustain severe brain injury” (Giacino et al., 2009, p. 33).
Six bedside assessments have been recommended for clinical practice to assess brain injury patients who have DOC (Seel et al., 2010). Seel et al. (2010) identified and appraised relevant assessment tools on the basis of their reliability and validity and made recommendations on their use in clinical practice. These recommendations were made using a modified version of the American Academy of Neurology classification system in which assessment tools are recommended with minor, moderate, or major reservations or are not recommended for use (Seel et al., 2010). The leading assessment, with only minor reservations, was the Coma Recovery Scale—Revised (CRS–R; Giacino, Kalmar, & Whyte, 2004; Kalmar & Giacino, 2005). This assessment was developed for patients with a range of severity in brain injury. Four others were recommended with moderate reservations, and two of these assessments were specifically designed to assess patients following severe brain injury: the Wessex Head Injury Matrix (WHIM; Shiel et al., 2000) and the Western Neuro Sensory Stimulation Profile (WNSSP; Ansell & Keenan, 1989).
The CRS–R, WHIM, and WNSSP each present items with a hierarchy of weighted scores in which higher levels of function receive higher scores. Such hierarchies suggest a recovery sequence, and anecdotally therapists expect behaviors to emerge in sequences similar to those presented in the assessments. Sequence of recovery is critical to DOC prognostic accuracy because it is “combinations of signs and the timing of their observation [that] is more informative than any single sign” (Dolce & Lucca, 2010, p. 108). Sequence of recovery in patients with severe TBI was previously studied using the WNSSP (Ansell & Keenan, 1989; N = 9 patients) with a particular focus on the ability of the WNSSP to identify emergence from the DOC of a minimally conscious state (MCS). This group study found that functional communication and object manipulation were two key indicators of emergence from MCS; however, sequence of recovery across items was not the focus (Aspen Neurobehavioral Conference Work Group, 2002).
The WNSSP was specifically designed for slow-to-recover patients with severe head injury. The WNSSP has been identified as having acceptable standardized administration and scoring procedures and good content validity, with items in the test providing helpful information for differential diagnosis across DOCs using the Aspen Work Group criteria (Aspen Neurobehavioral Conference Work Group, 2002) from the lowest level of vegetative state to MCS or emergence from MCS (Seel et al., 2010). Accurate DOC diagnosis for this patient population is needed. Notwithstanding the earlier study in which the WNSSP was used to identify emergence from MCS, to date the actual sequence of neurobehavioral recovery in patients with severe TBI using the WNSSP has not been tested. The aim of this study was to understand the sequence of recovery in neurobehavioral function in patients with severe TBI and to determine whether this sequence is consistent with the sequence reflected in the WNSSP test item order.
Method
A retrospective clinical cohort design was used. The University of Western Sydney Human Research Ethics Committee and Royal Rehabilitation Centre Sydney Human Research Ethics Committee granted institutional ethics approvals before the study commenced.
Participants
We reviewed all archived records of admission to a specialist brain injury rehabilitation service at the Royal Rehabilitation Centre Sydney in New South Wales, Australia, from January 2001 to December 2006 and included all patients diagnosed by the treating physician as having a severe TBI, determined by GCS score <8 at admission and duration of posttraumatic amnesia >1 wk using the Westmead PTA Scale (based on the archived data, which measured PTA prospectively; Marosszeky, Ryan, Shores, Batchelor, & Marosszeky, 1997). Medical records were deidentified, and all records with GCS scores plus admission and discharge clinical assessments using the WNSSP (Ansell & Keenan, 1989) were included in the analysis. The WNSSP and FIM™ (Uniform Data Set for Medical Rehabilitation, 1997) were routinely administered during center admission even if coma was evident. Coma state was defined clinically as the clinical state in which a patient is nonrousable and does not respond to stimuli (Rosenberg, 2009).
Instruments
The WNSSP is an observational test of 33 items administered bedside and then scored on type of stimulation (general or specific), latency of reaction, and need for cueing. The total score is the sum of performance across all items (range = 0–113, with low scores indicating poorer function; Ansell & Keenan, 1989). The authors of the WNSSP developed the sequence of recovery of behaviors from clinical observation and reading of the literature. Nine areas of performance are measured in the following order: arousal and attention (4 items), auditory response (2 items), auditory comprehension (6 items), expressive communication (3 items), visual tracking (7 items), visual comprehension (5 items), tactile response (2 items), object manipulation (3 items), and olfactory responses (1 item). The WNSSP has excellent internal consistency (Cronbach’s α = .95) and good content validity (Seel et al., 2010); further research is required to describe additional psychometric properties.
The FIM was used to provide information about the level of disability of the participant group. The FIM describes level of disability in a range of domains including personal care, sphincter control, mobility and locomotion, communication, and social cognition. Each item is given a score ranging from 1 = total assistance to 7 = complete independence (Keith, Granger, Hamilton, & Sherwin, 1987). The FIM has good construct and concurrent validity, high internal consistency (Cronbach’s α = .93–.95), and high interrater and test–retest reliability (Mackintosh, 2009).
Data Collection
The GCS was administered by treating doctors. The WNSSP and FIM were administered as part of routine clinical data collection by experienced speech pathologists and physical and occupational therapists who were specialists in neurological rehabilitation. All therapists were routinely trained in WNSSP administration through a familiarization session facilitated by a senior therapist, observation of test administration by the experienced therapist, and practice administration of the test with feedback on scoring. Items were administered according to manual instructions at bedside; length of time of assessment was not held constant across patients in this retrospective, pragmatic study. We used a data extraction form to record data from the deidentified medical file. Data extracted included GCS scores, length of coma, length of PTA, FIM score, Rancho Los Amigos Scale (Hagen, Malkmus, & Durham, 1972) results, and WNSSP scores.
Analysis
Demographic data were analyzed using descriptive statistics. The recovery sequence was investigated using the paired preference technique (PPT), which previously has been used to examine the sequence of recovery following severe head injury (Watson & Horn, 1992; Watson, Horn, Wilson, Shiel, & McLellan, 1997). This technique involved comparing the date of achievement of each item on the WNSSP with the date of achievement of every other item on the WNSSP in a series of pairwise comparisons. For example, Item 1 is compared with Items 2, 3, 4, 5, and so on. Following this, Item 2 is compared with items 1, 3, 4, 5, and so on until comparisons are made with all other items. Each pairwise comparison is made to label the item that each patient recovered first. If the first item in the pair was recovered before the second item, pairwise comparison was coded as 1. If the first item did not recover before the second item, the pair was coded 0. Pairwise comparisons were recorded in a matrix and adjusted to provide a best-fit sequence of test-item recovery across all participant pairs to identify an observed order-of-recovery trend on WNSSP items. The PPT was used to analyze WNSSP data from admission and discharge data and all other progress points. The technique also provided information regarding item redundancy (ties) and scale components needing refinement.
Results
Thirty-seven eligible participants (84% men) were admitted between 2001 and 2006 (Table 1). Two patients had been excluded because of incomplete medical records. Included participants had a mean age of 29 yr, 4 mo (range = 15 yr, 11 mo–66 yr, 1 mo) and a mean of 77.6 days postinjury on admission (range = 15–243 days, standard deviation [SD] = 51.8); 81% had been transferred from another hospital. Participants all had severe TBI with a mean duration PTA of 110 days and were clinically diagnosed as being in MCS at the time of the initial WNSSP assessment. Seventy percent of study participants had not emerged from PTA by discharge.
Participant Characteristics (N = 37)
Note. Percentages may not total 100 because of rounding. M = mean; MCS = minimally conscious state; PTA = posttraumatic amnesia; SD = standard deviation.
Patients who did not emerge from coma (n = 6) were not included in the calculation of length of PTA.
GCS scores for 35 (95%) participants were ascertained at the accident scene (mean GCS score = 4.2, SD = 2.2) and for 22 (59%) at admission (mean GCS score = 3.9, SD = 1.1). Only 3 had discharge GCS scores reported (of 6, 8, and 8).
WNSSP admission scores were available for 33 participants (89%) at admission and 32 (86%) at discharge (Table 2). Mean WNSSP total score was 38.3 at admission (SD = 29.6) and 70.2 at discharge (SD = 31.6), reflecting a mean gain of 31.9 points. The distribution of scores was large (e.g., discharge WNSSP total scores ranged from 2 to 112). On admission, the majority of participants scored below the score range considered indicative of rehabilitation-ready status (i.e., 65–75). More than half exceeded this range by discharge. Mean and median admission scores were reasonably high on the Arousal/Attention subscale, but significant impairment was found on five other subscales, particularly Visual Response, Expressive Communication, and Olfactory Response (Table 2). The distribution of WNSSP discharge subscale scores was large with the exception of Arousal/Attention. Mean discharge subscale scores indicated improvement on all subscales, particularly Visual Response and Auditory Response. Despite improvements, impairment was still apparent at discharge with the exception of Arousal/Attention.
WNSSP, GCS, FIM, and RLAS Descriptive Statistics
Note. GCS = Glasgow Coma Scale; M = mean; max = maximum; NA = not applicable; RLAS = Rancho Los Amigos Scale; SD = standard deviation; WNSSP = Western Neuro Sensory Stimulation Profile.
We explored the sequence of recovery of cognitive–sensory function with WNSSP data sets that had admission, discharge, and progress points for each of 31 participants; the 33 items were compared, resulting in 528 pair comparisons. An index was computed for each pair; this index represented the number of cases in which the first item of a pair recovered before the second (i.e., scored 0 or 1). For each of the 33 items, indexes of all pairs were summed. The summed index was used to order an item by comparing it with the summed index of other items. This ordering then represented the observed sequence of recovery in our sample. Table 3 presents the observed sequence of recovery of items and the order in which the test presented the item (WNSSP test order). Observed WNSSP recovery of function in our sample was not the same as the order of recovery presented in the WNSSP instrument.
Hierarchy of Behaviors Assessed Using the WNSSP Reflecting a Statistically Derived Order of Recovery From Coma
Note. WNSSP = Western Neuro Sensory Stimulation Profile.
With the exception of Arousal/Attention, observed subscale item order also was not consistent with the instrument item order (Table 4). Arousal/Attention observed recovery did follow the test item sequence proposed (i.e., Item 1 recovered first, followed by Items 2, 3, and then 4).
Observed Order of Recovery on the WNSSP Subscales
Note. WNSSP = Western Neuro Sensory Stimulation Profile.
Discussion
The WNSSP identifies a hierarchy of items indicating a proposed sequence of recovery (Ansell & Keenan, 1989). We found that the observed sequence of recovery matched test item order for the Arousal/Attention subscale but not for the other subscales. A revised item order may be required to reflect the observed order presented in Table 4. Visual Response, for example, with the exception of Item 18, “horizontal track picture,” suggested that participants could track stimuli in the upper and lower visual fields (vertically) before they could track items in the left and right visual fields (horizontally). It was also apparent that patients demonstrated an ordered preference for certain stimuli. In both the vertical and horizontal tracking items, participants were first able to track a picture, then a mirror, then an object. Moreover, observed order showed that patients were able to track stimuli before they could respond to visual commands. This finding complements and places in context the recommendation (Vanhaudenhuyse, Schnakers, Bredart, & Laureys, 2008) that a mirror should be used in assessment of postcomatose states.
Almost one-third of total-scale WNSSP items relate to visual performance with items requiring visual tracking or visual recognition. These items can be problematic if the patient has a visual impairment, something that is practically difficult to detect when he or she also has a disorder of consciousness. Notwithstanding this inherent limitation, study findings contribute to the growing body of evidence relating to the sequence of recovery of visual fixation, tracking, and pursuit following severe brain damage. Visual tracking is presumed to indicate the return of functional networks in the brain required for consciousness; specifically it “reflect[s] (partial) recuperation of the brainstem–cortical interaction and overall brain functional organisation that are thought to be necessary to sustain consciousness” (Dolce et al., 2011, p. 1152). For decades, observation of visual performance has been a critical clinical marker regarding DOC informing diagnostic, prognostic, and classification decisions (Ansell, 1993, 1995; Bruno et al., 2010; Dolce et al., 2011, p. 1152). In particular, visual performance is now identified as critical in differentiating between patients in a vegetative state or an unresponsive wakefulness syndrome, in which visual tracking is present in only 20% of patients (Gosseries et al., 2011), and those with MCS (Dolce, Quintieri, Serra, Lagani, & Pignolo, 2008) or low-level MCS (Giacino & Kalmar, 1997), in which visual tracking is present in 80% of patients.
Although the quest for better diagnosis and classification of disorders using technological approaches continues (Coleman et al., 2009), particularly for patients in whom no behavioral signs of consciousness are present (Bruno et al., 2011), in reality most patients will continue to be assessed using behavioral assessments. At the heart of behavioral assessments, visual performance and the order of recovery of neurobehavioral function are critical clinical signs of reorganizing neuronal networks. To help determine the likely behaviors that will next recover, researchers should pay careful attention to the sequence of recovery of neurobehavioral function. The slow, incremental sequence of behavioral change in response to standardized performance items reveals the neurocortical change within. As Dolce and Lucca (2010) suggested, “Boundaries between the vegetative and minimally conscious states are somehow blurred and experience suggests sequential phases in a recovery process, rather than independent conditions” (p. 107). The findings from our study suggest that WNSSP subscales, rather than total scores, provide a level of information that may be more helpful for practitioners faced with the challenge of differentiating types of DOC.
Limitations and Future Research
The paired comparison analysis method has limitations. Participants do not necessarily demonstrate recovery of all test items, nor do they necessarily recover items in exactly the sequence found in the results of the current study. The finding, for example, that patients recovered the simple motor command “move your arm” before “open your mouth” is not meant to indicate that one task is more difficult to achieve than the other. Rather, the method of analysis is used to suggest a trend for order of recovery rather than a fixed sequence. The strength of the paired preferences technique lies in its ability to suggest an order of recovery in certain domains that can then be used clinically to suggest test items the patient might reasonably be expected to recover next.
Our study is the first, to our knowledge, to use observational data to suggest trends in WNSSP order of recovery. Further research, preferably prospective in design and involving a larger number of participants, is required to verify these findings. Furthermore, the use of Rasch analysis to verify these results would provide additional useful information.
Implications for Occupational Therapy Practice
The results of this study have the following implications for occupational therapy practice:
The sequence of recovery from coma among study participants was not consistent with the sequence proposed in the WNSSP.
Occupational therapists should be aware of the inconsistency between what they may observe in clinical practice and the sequence of recovery reflected in the WNSSP.
Furthermore, occupational therapists should carefully monitor changes in the patient’s neurobehavioral function rather than relying on assessment scores alone.