Stroke Rehabilitation (III): Impairments in Somatosensation

(1) Prevalence of somatosensory loss post-stroke

Evidence-based practice and research have established that impairments resulting from a stroke happen not only in motor domains (e.g. inability to perform reaching, loss of balance, and slurred speech) but also in somatosensory domains (loss of tactile sensation, limb position sense, and perceiving force). Carey LM (1995) stated that the loss of somatosensation occurs in about 60% of stroke survivors and has detrimental impacts on the quality of life, e.g. in spontaneous use of the hand and object manipulation. 

Intact somatosensation is essential for motor control since it is reliant on both intact feedforward and intact feedback from afferent inputs. It has been suggested that a learned non-use phenomenon with sensory loss leads to further deterioration of motor abilities. Despite this fact, the association between sensory impairment and outcomes following stroke has received limited focus in rehabilitation research. One reason is that most clinicians assume that spontaneous motor recovery occurs in the first 4 weeks post-stroke in the acute phase (Jia-Ching et al, 2005; Wing et al, 1990; Heller A et al, 1991; Lincoln NB et al, 1991), and that somatosensory recovery will arguably follow suit and therefore receives less attention. Another reason is the complexity in measuring the sensing ability of different modalities.

More recently, Connell et al (2008) conducted a newer prospective study with 70 patients with a first stroke assessed on admission day, and 2, 4, and 6 months post-stroke. Their findings did not contradict the earlier findings by Carey LM (1995). Of the sample collected, the authors found that 7–53% had impaired tactile sensations, 31–89% impaired stereognosis, and 34–64% impaired proprioception. Specifically, proprioception and stereognosis (the ability to perceive 3D shape and depth) were more frequently impaired than tactile sensations. This is in contrast to a study by Kim et al. [16] on acute stroke patients, who found that proprioception was less impaired when compared with localization and two-point discrimination regardless of the lesion.

Connell et al further said that the different somatosensory modalities showed only slight agreement between impairment within the same body areas, suggesting that the modalities are independent of each other. This suggests that it is necessary to include all somatosensory modalities while assessing one body part. In contrast, the high agreement between sensory modalities in adjacent body areas means that it is probably not necessary to assess all related body parts, e.g. there was redundancy between the wrist and hand, or between the ankle and foot.

(2) NSA and RASP scales

One challenge of sensory assessment post-stroke is the variety of sensing modalities of somatosensation, ranging from tactile or touch, pressure, position sense, movement direction, pain, to temperature. Another important barrier is the lack of standardization and low reliability of the clinical assessment scale (Winward, et al 1999). At the moment, there are three common clinical assessment scales for sensory impairment: the sensation parts of the Fugl-Meyer Assessment for UE/LE, Nottingham Sensory Assessment (NSA), and Rivermead Assessment of Somatosensory Performance (RASP). All sensory tests are conducted in the absence of vision.

NSA was developed as a standardized clinical sensory assessment, assessing both sides of the body and all areas. The original version uses a 5-scale rating system, has good intra-rater reliability (the same clinicians did multiple times), but poor inter-rater reliability (different clinicians did the same assessment) and was time-consuming (Lincoln NB, et al, 1991). The NSA measures tactile sensations (light touch, temperature, pinprick, pressure, tactile localization, and bilateral simultaneous touch), on the face, trunk, shoulder, elbow, wrist, hand, hip, knee, ankle, and foot, on both the paretic and normal side. The poor reliability has prompted a revision of NSA according to the Erasmus MC version (Em-NSA) (Stolk-Hornsveld, 2006) with lesser items to test but uses 3-scale rating system. Although NSA was shown to have concurrent validity with the more established and gold standard Fugl-Meyer Assessment for sensorimotor impairments (Scalha et al, 2011), this scale is still less attractive, with some clinicians view this to be a mere screening tool at best.

The RASP is a multi-modal sensory tool that tests six sensations (sharp/dull discrimination, surface pressure, tactile localization, temperature discrimination, joint movement, and joint movement direction discrimination), and two secondary sensations (sensory extinction and two-point discrimination) (Winward et al, 2002). The scale for proprioception was shown to have excellent test-retest reliability among sub-acute survivors and good concurrent validity with Motricity Index and Barthel Index, an ordinal scale used to measure performance in activities of daily living (ADL).

(3) Sensory impairments over time

One focus area in stroke rehab is the ability of a clinical assessment to predict recovery. The power of predictability helps clinicians to assess the stroke severity and to provide the most accurate intervention given a particular condition. Connell et al (2008) found the initial somatosensory impairment was significantly related to sensing ability at 6 months, accounting for 46–71% of the variance. The authors argued that the remaining factors were attributed e.g. to more cognitive factors, perceptual ability, and motivation. The spontaneous recovery over time was more obvious in the upper limb compared to the lower limb. 

In a study by Meyer and colleagues in Belgium (Meyer S. et al, 2016), the authors recruited acute stroke patients (< 1 week) and conducted sensory tests. Confirming earlier studies, they found that 41–63% of stroke survivors in the acute phase had a sensory loss in one of the modalities within the first week, but the deficits improved to 3–50% when assessed 6-month post-stroke. Proprioception score of Em-NSA moderately predicts motor ability at 6-month post-stroke as measured by the Fugl-Meyer UE and Action Research Arm Test. As a comparison, stereognosis moderately predicted motor ability at six months post-stroke as measured by Fugl-Meyer UE, the Motricity Index, and Action Research Arm Test. 

Using RASP, the somatosensory subtest of proprioception demonstrated the greatest level of recovery, but no patient achieved full recovery on all somatosensory subtests (Winward et al, 2007). In another study using the original NSA, Connell et al (2008) reported that most recovery of the upper limb tactile sensations and stereognosis occurred in the first 4 months, whereas recovery in proprioception continued over 6 months. Although the motor and functional recovery demonstrated continual improvement over time, somatosensory recovery showed marked variation in subtests both within and between patients. Both Connell's and Winward's studies reported that individual somatosensory modality recovery may be independent of other somatosensory modalities, despite the existence of parallel processing in the central nervous system.

Functional MRI has been beneficial to elucidate plastic changes in the brain following a stroke. Recently Carey et al. (2002) demonstrated in a single case study of severe sensory loss that there was little evidence of neural plastic changes in the early stages after stroke (2 weeks). However, a return of activation in ipsilesional primary and bilateral secondary somatosensory cortices was observed at 3 months and this was maintained at 6 months.

(4) Somatosensory recovery

Kessner et al (2016): Most cortical reorganization within the motor system occurs within 2–3 months post-stroke and stabilizes after 6 months. However, much less is known about the time course of recovery from somatosensory deficits, and about the mutual interaction of somatosensory and motor recovery. The deficits recovered at least partially, mostly within 3 months. Interestingly, some modalities (graphesthesia impairment and movement detection) even appeared to deteriorate during the time course after initial recovery (Julkunen, et al 2005). After about 6 months of recovery, the prevalence of somatosensory deficits seems to be lower. The initial somatosensory deficit was the strongest predictor for long-term somatosensory ability. Taken together, most patients recover at least partially from their somatosensory deficits, mainly during the first three to six months after stroke (Fig. 1). However, not all modalities necessarily recover with positive results. 

The probability and duration to reach a good rehabilitative plateau phase (Barthel index > 60, ambulation > 150 ft) are significantly worse, if stroke patients had combined motor and somatosensory deficits compared to motor deficits only. The authors summarized that somatosensory deficits after stroke have an important negative effect on motor and functional performance, especially proprioceptive impairments!

(5) Somatosensory-specific interventions

Retraining focusing on somatosensory-specific interventions is lacking. Stroke survivors with sensory impairments share that they often feel this is often a neglected aspect in their rehab program (Doyle, Bennet, Fasoli, & McKenna, 2010). 

A randomized controlled trial of sensory retraining has shown improvement in stroke patients’ sensory discrimination functions following a series of tactile and proprioceptive training (Carey et al., 2011). Such improvement was maintained and slightly increased at 6 weeks and 6 months follow-up. Evidently, the impact of somatosensory interventions on the recovery of sensation post-stroke is thought to be positive and significant. The authors have also argued that such a form of intervention is clinically beneficial, where reduction in deficit was targeted to improve lost abilities. 

On the other hand, Gopaul et al. (2018) proposed that the combination of somatosensory training with motor training may result in greater improvements in both motor and somatosensory functions, as compared to interventions that focus on individual function recovery. Most studies of such combined interventions have comprised active training components as it has the potential to drive neural plasticity (produce greater cortical activation extending to multiple areas), particularly when delivered insufficient dose. For example, Byl and colleagues (2008) demonstrated that patients who received higher-dose (72 hours) of integrated active-somatosensory and passive-motor training performed sensory discrimination tasks more accurately as compared to those receiving lower-dose (12-13.3 hours) of training. Despite the improvements observed, motor improvements were considered smaller than somatosensory improvements due to the reinforcement of passive movement training (mental practice and mirror therapy) in their study.