The neurobehavioural rating scale-revised: sensitivity and validity in closed head injury assessment.

Abstract

Objectives--To investigate the factor structure and psychometric properties of the neurobehavioural rating scale-revised (NRS-R) and to determine its usefulness in clinical trials.

Methods--A consecutive series of patients sustaining severe closed head injury were evacuated to one of 11 large regional North American trauma centres and entered into a randomised, phase III, multicentre clinical trial investigating the therapeutic use of moderate hypothermia. Acute care personnel were blinded to outcome and outcome personnel were blinded to treatment condition. The Glasgow outcome scale (GOS) was the primary outcome measure. Secondary outcome measures included the disability rating scale (DRS) and the NRS-R.

Results--Exploratory factor analysis of NRS-R data collected at 6 months after injury (n=210) resulted in a five factor model including: (1) executive/cognition, (2) positive symptoms, (3) negative symptoms, (4) mood/affect, and (5) oral/motor. These factors showed acceptable internal consistency (0.62 to 0.88), low to moderate interfactor correlations (0.19 to 0.61), and discriminated well between GOS defined groups. Factor validity was demonstrated by significant correlations with specific neuropsychological domains. Significant change was measured from 3 to 6 months after injury for the total score (sum of all 29 item ratings) and all factor scores except mood/affect and positive symptoms. The total score and all factor scores correlated significantly with concurrent GOS and DRS scores.

Conclusions--The NRS-R is well suited as a secondary outcome measure for clinical trials as its completion rate exceeds that of neuropsychological assessment and it provides important neurobehavioural information complementary to that provided by global outcome and neuropsychological measures.

Keywords: outcome measures; head injury; randomised clinical trials; neurobehavioural sequelae

The neurobehavioural rating scale (NRS) of Levin et al [1] is a 27 item, multidimensional clinician based assessment instrument designed to measure neurobehavioural disturbances. Based on the brief psychiatric rating scale, [2] the NRS included items which would be more specific to patients with neuropsychiatric symptomatology resulting from closed head injury. Levin et al [1] reported a four factor solution to the NRS using principal components analysis in a sample of 101 patients with closed head injury (covering a wide range of injury severity), including: cognition/energy, metacognition, somatic/anxiety, and a language factor. Significant changes in factor scores were reported from the initial assessment at 3 months after injury to follow up testing at 6 months after closed head injury for cognition! energy, metacognition, and language. The NRS has been shown to be a reliable and valid [1 3-5] instrument for quantifying behavioural disturbances and gross cognitive impairments in those with neuropsychiatri c syndromes, including head injury, [3 6-10] dementia, [4 11-16] and HIV disease. [5]

To enhance the reliability and content validity of the NRS, Levin et al [17] developed the NRS-Revised (NRS-R). Changes to the NRS-R include the addition of two items ("difficulties in mental flexibility" and "irritability"), consolidation of two previous items into a single item ("tension" and "anxiety" merged together as "anxiety"), and separation of other items ("inattention" divided into "reduced alertness" and "attention"). To improve scoring reliability, the anchor points were more clearly defined and the Likert scale was condensed from a 7 point to a 4 point scale (absent, mild, moderate, and severe). The ratings for each item are defined on the basis of the potential impact of the behaviour on social and occupational independence. The NRS-R is administered through a brief structured interview (typically requiring 15-20 minutes to complete) which includes a test of orientation and memory for recent events, questions regarding emotional state, postconcussional symptoms, focused attention, and concentra tion (performing serial sevens), explanation of proverbs, tasks of planning and mental flexibility, and delayed recall of three objects presented at the beginning of a session. Observations are also made regarding the patient's fatigability, visible signs of anxiety, disinhibition, agitation, hostility, difficulties in expressive and receptive communication, and disturbance of mood. About one third of the item ratings are based solely on examiner observation which are graded according to a behaviourally anchored four point rating scale. The balance of the items are rated according to the patient's performance on brief tasks and quality of answers to interview questions.

In a French multicentre study in which data were collected at rehabilitation hospitals on 286 patients with closed head injury, Vanier et al [18] assessed the interrater reliability of the NRS-R through videotaped interviews in a randomly selected subgroup of 70 patients who had sustained closed head injury ranging from mild to severe according to the Glasgow coma scale (GCS) score of Teasdale and Jennett. [19] These videotapes of the NRS-R structured interview and mental status examination were rated based on a single viewing by two independent examiners at other centres involved in the project. Vanier et al reported that no item had a substantially low percentage of interexaminer agreement or al value. The mean percentages of agreement among all 29 items was 73.6%. The mean k statistic for three independent observers was acceptable at 0.43 with a range in value across the NRS-R items from 0.22 ("difficulty in planning") to 0.77 ("memory difficulties").

Factor analysis of the NRS-R data for the total sample of 286 patients with closed head injury in the French multicentre study [18] disclosed five correlated factors (using oblique rotation) which explained 42.2% of the total variance. Factor I (intentional behaviour) consisted of items pertaining to intentional or goal oriented behaviour (for example, difficulties in self appraisal, initiative, and motivation, mental flexibility, and planning, blunted affect, conceptual disorganisation, disorientation, and memory problems). Factor II (lowered emotional state) included depressive mood, anxiety, and emotional withdrawal, factor III (survival oriented behaviour/heightened emotional state) consisted of items such as irritability, disinhibition, hostility, and hyperactivity-agitation, factor IV (arousal state) had items such as reduced alertness, inattention, and mental fatiguability and factor V (language) consisted mainly of expressive and comprehension language skills. Although the study by Vanier et al provi ded empirical support for the interrater reliability and validity of the NRS-R while identifying five factors, the cross sectional design precluded analysis of the sensitivity of the NRS-R to changes over time. In the course of a multicentre clinical trial which utilised the NRS-R, we investigated the factor structure and scale properties of this instrument in a sample of patients with severe head injury who were studied longitudinally.

Methods

PATIENTS

From October 1994 to the end of November 1998, 392 patients with severe closed head injury were enrolled in and eligible for an evaluation 6 months after injury in the national acute brain injury study: hypothermia (NABIS:H), a multicentre, phase III clinical trial investigating the therapeutic use of moderate hypothermia to treat severe closed head injury. Inclusion criteria consisted of an abnormal CT obtained within 24 hours of injury, and a post-resuscitation GCS motor score of 1 to 5 (total GCS [less than or equal to] 8). Exclusion criteria included evidence of hypotension (systolic BP [less than] 90 mm Hg for 30 minutes after resuscitation), hypoxia (saturation[less than]94%) for 30 minutes after resuscitation, estimated abbreviated injury severity [20] score [greater than]4 for any organ system, GCS of 3 with unreactive pupils, or inability to randomise within 6 hours of injury. Consent to randomise was obtained from appropriate family members or relevant others. Outcome personnel were blinded to the p atient's treatment condition, and the acute care personnel were blinded to the patient's outcome at the assessment 3 months after closed head injury and the 6 month evaluation which was the primary end point.

COMPLETION RATE

Of the 392 patients enrolled, 105 (26.8%) died before the 6 month end point. A total of 77 …