May clinical neurophysiology help to predict the recovery of neurological early rehabilitation patients?

Background So far, the role of clinical neurophysiology in the prediction of outcome from neurological and neurosurgical early rehabilitation is unclear. Methods Clinical and neurophysiological data of a large sample of 803 early rehabilitation cases of the BDH-Clinic Hessisch Oldendorf in Northern Germany have been carefully reviewed. Most patients (43.5 %) were transferred to rehabilitation after stroke, mean age was 66.6 (15.5) years. Median somatosensory (SEP), auditory (AEP) and visual evoked potentials (VEP) along with EEG recordings took place within the first two weeks after admission. Length of stay (LOS) in early rehabilitation was 38.3 (37.2) days. Results Absence of SEP on one or both sides was associated with poor outcome, χ2 = 12.98 (p = 0.005); only 12.5 % had a good outcome (defined as Barthel index, BI ≥50) when SEP were missing on both sides. In AEP, significantly longer bilateral latencies III were observed in the poor outcome group (p < 0.05). Flash VEP showed that patients in the poor outcome group had a significantly longer latency III on both sides (p < 0.05). The longer latency III, the smaller BI changes (BI discharge minus admission) were observed (latency III right r = −0.145, p < 0.01; left r = −0.206, p < 0.001). While about half of the patients with alpha EEG activity belonged to the good outcome group (80/159, 50.3 %), only 39/125 (31.2 %) with theta and 5/41 (12.2 %) with delta rhythm had a favourable outcome, χ2 = 24.2, p < 0.001. Conclusions Results from this study suggest that loss of median SEP, prolongation of wave III in AEP and flash-VEP as well as theta or delta rhythms in EEG are associated with poor outcome from neurological early rehabilitation. Further studies on this topic are strongly encouraged.


Background
Patients entering neurological and neurosurgical early rehabilitation are severely impaired. Morbidity is high [1] and they are suffering from disorders of consciousness [2,3]. Early rehabilitation patients are dependent on nursing and may be colonized with multi-drug resistant germs [4,5]. Frequently, their outcome is poor, but it is quite difficult to predict outcome accurately [3].
The role of clinical neurophysiology, in particular electroencephalography (EEG) and evoked potentials (EP) in predicting outcome of these patients is still unclear.
Only a few studies are available on long-term rehabilitation results and clinical neurophysiology. The question is whether clinical neurophysiological techniques may help to distinguish between patients who benefit from neurological early rehabilitation and such who don't. In contrast to imaging techniques, neurophysiological measurements are easy to perform, cheap, safe and available in most rehabilitation facilities.
Some studies on long-term outcome of traumatic brain injury (TBI) patients are available (Table 2) [17][18][19][20][21][22][23]. While AEP may be of some prognostic value in this group of patients [17,18], most studies focus on SEP. Like with stroke, TBI patients with a loss of cortical SEP-responses seem to have a poor outcome [22].
There are also studies on long-term outcome of disorders of consciousness (DOC) patients, in particular after hypoxic brain damage (Table 3) [3,[24][25][26][27][28]. Results on the question whether SEP or AEP may be used to predict recovery are controversial [24][25][26][27]. EEG could be of some value [28]. Latest results from our group suggest that prolongation of wave III in flash VEP, theta or delta rhythm in EEG, and N20/P25 amplitude reduction in median SEP may be associated with poor outcome of hypoxic brain damage patients undergoing neurological early rehabilitation [3].
The present study tried to figure out whether clinical neurophysiology may help to improve outcome prediction of a large cohort of neurological and neurosurgical early rehabilitation patients.   Table 4. The main diagnosis was defined according to the G-DRG-(German Diagnosis Related Groups) system as the disease justifying referral for early rehabilitation. When stroke patients also suffered from a peripheral disorder like diabetic polyneuropathy, for instance, stroke was defined as main diagnosis because it necessitated rehabilitation treatment.
One patient was discharged against medical advice and another one was transferred to a hospice.

Clinical scales and assessments
On admission and at discharge, Barthel index (BI) [29] and Early Rehabilitation Index (ERI) [30] have been obtained. In line with previous studies, poor outcome was defined as a BI <50 points [3]. In addition, CRS [31], GCS [32] and Early Functional Abilities scale [33] have been analyzed.

Clinical neurophysiology
Electroencephalography (EEG), auditory evoked potentials (AEP), visual evoked potentials (VEP), somatosensory evoked potentials (SEP) of the median nerve were recorded usually within the first two weeks after admission. EEG was done using the international 10/20 system (Neurofax EEG 9000, Nihon Kohden Europe, Rosbach, Germany). Surface electrodes were used for evoked potentials (Nicolet Viking Select, Natus Medical, Middleton, WI, USA). VEPs were recorded with flashing light-emitting diodes (flash VEP, stimulation frequency 1.3Hz). Latencies and amplitudes of wave I-III were examined according to the guidelines of the American Clinical Neurophysiology Society [34]. Further, AEP latencies I-V and N20/P25 latencies and amplitudes of median nerve SEPs were analyzed. Neurophysiological examinations were performed by an experienced team of only four paramedics working in this field for many years.

Ethics
This is a retrospective data analysis, only (no intervention). Local ethics committee of the BDH-Clinic Hessisch Oldendorf gave approval to use facility's data. Patient records/information were anonymized and de-identified prior to analysis. No written informed consent for participation was obtained (retrospective data analysis, no intervention).

Results
Poor outcome patients were significantly older, had more co-diagnoses, a longer LOS, lower BI, ERI, GCS and CRS on admission (Table 5).

Imaging data
Computed tomography (CT) and/or magnetic resonance imaging (MRI) of the skull was available in about 2/3 of cases (Table 6). Most frequent lesion sites were temporal, parietal and frontal lobes.

Median SEP
Median SEP data was available in 449 cases (55.9 %). Loss of cortical SEP on one or both sides was associated with poor outcome ( AEP AEP data was available in 448 cases (55.8 %). Absence of AEP on one side was observed in two cases, bilateral loss of AEP responses in one case, only. All three cases belonged to the poor outcome group, but due to small sample size, χ 2 -test did not reveal significant differences. When comparing poor and good outcome groups, it turned out that a significantly longer latency III was observed on both sides in the poor outcome group (p < 0.05), Table 8. Latency IV also showed significant differences between the groups, but on the left side, only (p < 0.05). Since AEP are closely connected to brain stem function, a sub-analysis focusing on brain stem lesions was done: 60 patients with brainstem lesions (7.5 %) were identified. With intact brain stem, BI at discharge was significantly higher than in patients with unior bilateral lesion (F = 3.931, p = 0.009). Further, there was a small but significant correlation between age and AEP latency III (left: r = 0.209, p < 0.001; right: r = 0.132, p < 0.01).

VEP
Flash VEP data was available in 391 cases (48.7 %). Loss of cortical VEP was detected in six cases on one and in four cases on both sides, only. All bilateral loss cases belonged to the poor outcome group. With unilateral absence, 4/6 (66.6 %) had a poor outcome. Like with AEP, these differences did not reach a level of significance due to small group size. Patients belonging to the poor outcome group had a significantly longer flash VEP latency III on both   sides (p < 0.05), Table 8. The longer latency III, the smaller BI changes (BI discharge minus admission) could be observed (latency III right r = −0.145, p < 0.01; left r = −0.206, p < 0.001). As with AEP, there was a small but significant correlation between age and VEP latency III (left: r = 0.166, p < 0.01; right: r = 0.136, p < 0.01).

EEG
EEG recordings were available in 360 cases (44.8 %). EEG with alpha, theta or delta frequency was included in the analysis, Table 9. While about half of the patients with alpha activity belonged to the good outcome group (80/159, 50.3 %), only 39/125 (31.2 %) with theta activity and 5/41 (12.2 %) with delta rhythm had a favorable outcome, χ 2 = 24.2, p < 0.001. In ANOVAs, BI was significantly lower on admission and discharge when patients had theta or delta rhythms compared to alpha activity (p < 0.001). BI changes (BI discharge minus admission) were also smaller when patients had delta or theta activity, Fig. 1. The age of patients with alpha, theta or delta activity did not differ significantly (F = 1.274, p > 0.05).

Discussion
So far, reliable data on the usefulness of neurophysiological measurements in predicting the outcome from neurological and neurosurgical early rehabilitation is lacking. The present study analysed data of a large sample 803 patients. One would opt for clinical neurophysiology as a predictor because it is cheap, safe (no radiation), easy to perform and available in most rehabilitation facilities (in contrast to imaging like CT or MRI). As with previous studies, present results suggest that outcome of neurological early rehabilitation patients mainly depends on age, morbidity and functional status on admission [3]. However, neurophysiological data may be of some predictive value, in particular median SEP, AEP, flash VEP and EEG.
It turned out that patients with worse outcome had longer AEP III latencies. AEP wave III represents the cochlear nucleus which is located in the pontomedullary junction of the dorsolateral brainstem [35]. It is well known that brainstem lesions are associated with poor neurological outcome and fatality [36]. This finding could be reproduced in this study: Subjects with brain stem lesions on one or both sides showed a worse functional outcome from neurological early rehabilitation. As yet, a prolongation of AEP wave III latency has not been identified as a predictor of poor outcome and is a novel finding. Age, however, correlated significantly with wave III latency. Since age is a well-known predictor of poor outcome in neurological rehabilitation [3], it may partially explain this finding. In addition, it has to be Significant differences between subjects with good and poor outcome are indicated as follows: *p < 0.05, **p < 0.01 (t-tests for independent samples) pointed out that even in the normal ageing brain, a delay of evoked potentials, in particular VEP and AEP, may be observed [37,38]. Another finding of this study was a prolongation of wave III in flash VEP in the poor outcome group. This finding is in line with a previous study from our group which focused on hypoxic brain damage patients [3]. VEP wave III abnormalities might be a neurophysiological correlate of cortical dysfunction [3]. As with AEP, VEP wave III latency also correlated with age. Thus, age might influence AEP wave III, too.
Another finding was that loss of cortical median SEP responses on one or both sides was associated with poor outcome. We know from literature that long-term outcome of stroke patients is also worse with absent SEP [6,10]. Thus, it may be hypothesized that absence of SEP indicates poor outcome in early rehabilitation patients.
There are a couple of studies focusing on EEG and outcome prediction. As with previous studies [3,8], theta and delta activity was associated with poor outcome.
There are some limitations to this study. First of all, this was a retrospective data analysis, only. This explains why only a proportion of the sample has been studied with all four neurophysiological examinations (EEG, SEP, AEP, VEP). Secondly, the patients showed a wide heterogeneity. This, however, is a common finding when examining neurological early rehabilitation patients [1]. These patients suffer from a broad specter of neurological and neurosurgical disorders, central as well as peripheral disturbances. In addition, the study employed no control group and confounding medication (e.g. neuroleptics, benzodiazepines) has not been included in the analysis. Sedatives, however, are rarely used in our rehabilitation facility.
Results from this study defy ready summary, but EEG, median SEP, AEP and flash VEP may be of some predictive value in early rehabilitation patients. Further studies are strongly encouraged.

Conclusion
Results from this study suggest that loss of median nerve SEP, prolongation of wave III in early AEP and flash-VEP, as well as theta or delta rhythms in EEG are associated with poor outcome from neurological early rehabilitation. Clinical neurophysiology may help to improve outcome prediction of neurological and neurosurgical rehabilitation patients. EEG and evoked potentials are widely-used, cheap, easy to perform and non-invasive.

Competing interests
The author declares that he has no competing interests. Fig. 1 When patients had alpha EEG-activity, BI on admission, at discharge and changes of BI (discharge minus admission) were significantly higher than patients with theta or delta activity (ANOVAs with LSD-tests, p < 0.001)