History

Electricity has been applied to human disease for centuries. As early as 400 BC, torpedo fish possessing paired organs capable of generating up to 220 V of electricity, and amber, a fossilized resin that produces electrical current when rubbed, were used to treat a wide variety of conditions ranging from headaches to hemorrhoids.^15,^23,⁸⁶ “Electricity” was derived from the Latin word for amber, “electrica.” In 1744 Kratzenstein, a German physician, reported the first use of static electricity to treat paralysis.⁶⁵ Beginning in 1745 with the invention of the Leyden jar, a precursor to the capacitor, technological advances provided a growing capacity to harness, store and modulate electricity. In 1831 Faraday constructed an electromagnetic machine, a forerunner of the electric generator and many electrotherapeutic devices that are currently in use.⁶⁵ Because of ongoing technological development and the invention of devices such as the battery, the latter half of the nineteenth century became an era where electrotherapeutic treatments expanded in a largely unregulated fashion, possibly contributing to a backlash of skepticism occurring later. Around 1840 electrical stimulation of muscle began its development as a diagnostic tool. Although electrical changes corresponding to the human heartbeat had been demonstrated decades earlier, Einthoven (inventor of the string galvanometer) in 1903 began electrical recordings of the human heart and coined the term “electrocardiogram.” Zoll demonstrated that an artificial pacemaker could sustain a human heartbeat in 1952, and in 1958 the first human heart was paced for 96 days. Liberson et al.⁵⁸ reported the first practical application of electrical stimulation in rehabilitation in 1961. He described the application of external current to the peroneal nerve to correct foot drop in a hemiparetic patient. Today there are three electrical stimulation systems to correct foot drop that are commercially available in the United States. A wide variety of electrical stimulation applications have been clinically implemented to address many rehabilitation issues, and many more are currently in development.

Physiology

A comprehensive discussion of the neurophysiologic basis of electrical stimulation in medicine is beyond the scope of this chapter. Several key issues warrant discussion, however, in order to provide a background for understanding the mechanisms of action and limitations of electrical stimulation interventions. Electrical stimulation in physiatric practice can be placed into four general categories whereby external current is applied to (1) the peripheral nervous system to induce muscle activation; (2) the central nervous system to induce muscle activation; (3) the central or peripheral nervous system to modulate sensory input, primarily to reduce pain; and (4) the skin to induce wound healing. Peripheral nerve stimulation to activate muscle, also known as neuromuscular electrical stimulation (NMES), is by far the most common.

With NMES the point of stimulation can occur anywhere along the axon of the lower motor neuron including the motor points or the terminal branches of the nerve. Stimulation-induced muscle activation requires an intact motor unit—that is, the lower motor neuron (α-motor neuron), neuromuscular junction and corresponding muscle fibers. The threshold for muscle fiber stimulation is 100 to 1000 times higher than that for nerve fiber stimulation.⁶⁹ For clinical purposes, direct stimulation of muscle fibers to induce contraction cannot be safely performed with available technology. Consequently NMES cannot induce muscle contraction in denervated muscle in clinical applications, and is generally best suited for upper motor neuron diseases to induce muscle activation in paralyzed muscle where the motor unit remains intact.

The generation of an action potential in the α-motor neuron follows the “all-or-none” principle for NMES, just as it does for normal physiologic excitation. The order of recruitment of motor units with NMES is reversed, however, in comparison with normal physiologic excitation. This is because the threshold for excitation with an externally applied current is inversely proportional to the nerve fiber diameter.⁹² During normal volitional muscle contraction, motor units with low conduction velocity, slow twitch force, long contraction time, low force generation, and greater resistance to fatigue are recruited first. With increasing volitional force generation, motor units with higher conduction velocity, greater force generation, and less resistance to fatigue are recruited.^7,^41,⁴² The orderly recruitment of motor units during normal volitional muscle contraction facilitates gradations in muscle force generation, smooth movement, and resistance to fatigue. With NMES the motor units containing larger nerve fibers are recruited first, followed by motor units containing smaller nerve fibers. As a result, NMES first recruits motor units with higher conduction velocities, greater force generation, and lower resistance to fatigue, followed by motor units with slower conduction velocities, lower force generation, and greater resistance to fatigue.⁹² NMES-induced recruitment, the reverse of normal recruitment, can pose challenges in clinical applications such as abrupt force generation and rapid onset of fatigue that can be only partially compensated for by adjusting stimulation parameters to modulate the delivery of the external current. For example, a longer stimulus ramp-on time can induce more gradual force generation, and lower stimulus frequencies can minimize muscle fatigue. The modulation of stimulation parameters relevant to the clinical use of NMES is discussed in subsequent sections.

Muscle fiber types can change in response to NMES. Although most data come from studies in mammalian animal models, the consistency of the results suggests that the data can be generalized to all mammals including humans. Skeletal muscle can be classified into types I, IIa, and IIb based on metabolic, histochemical, and functional characteristics (Table 21-1).⁶⁴ All muscle fibers are histologically identical within a motor unit, but muscle fiber types vary within a muscle and convey unique metabolic and contractile characteristics.¹⁴ Continuous (24 hr/day) low-frequency (10 Hz) stimulation results in the transformation of fast-twitch type IIb fibers into slow-twitch type I fibers. Histologic changes are seen within 2 to 4 days of stimulation onset, beginning in the sarcoplasmic reticulum.⁴⁰ Fiber type conversion is completed within 8 weeks.⁷⁷

Table 21-1 Muscle Fiber Types

The timing and sequence of histochemical and ultrastructural changes suggest that the regulation of fiber type transformation occurs at the level of gene transcription. The fiber type changes are accompanied by associated metabolic changes from the glycolytic pathway of type IIb fibers to the Krebs cycle pathway of type I fibers. Capillary density and oxygen utilization are increased.^13,⁴⁵ The fiber type transformation is accompanied by characteristic functional changes, including greater resistance to fatigue.⁴⁴ A reduction in daily stimulation time (noncontinuous, 8 hr/day) results in similar but delayed changes in muscle fiber type.⁷⁸ When the electrical stimulation is discontinued, muscle fibers return to their prestimulation type in a first-in, last-out sequence. Functional characteristics such as contractile properties return within 6 weeks, and metabolic changes are reversed within 12 weeks, but changes in capillary density may persist for several months.⁸⁷

Electrical stimulation delivered at higher frequencies (40 to 60 Hz) results in similar fiber type changes if the total number of stimuli remains constant.^46,⁹³ An increase in frequency (100 Hz) and number of stimuli has been shown to result in conversion of slow-twitch to fast-twitch fiber in denervated rodent muscle, but similar effects have not been demonstrated in innervated muscle.⁶⁰

Stimulation Variables and Parameters

Externally applied electrical current can be characterized by a number of parameters including waveform, amplitude, pulse width, frequency, and duty cycle. Electrical current can be direct (DC), alternating (AC), or pulsed. DC current involves a continuous, unidirectional flow of charge that is generally not applicable to NMES. However, stimulation of the brain cortex with DC current has been investigated for treatment of depression and various cognitive impairments.^9,^53,⁶¹ AC current involves a continuous, bidirectional flow of charge. Pulsed current is most commonly used in NMES and involves a pulsed, interrupted flow of charge that can take on a variety of waveforms. Pulsed waveforms can be monophasic or biphasic. Biphasic waveforms can be symmetric or asymmetric with respect to baseline. Waveform asymmetry can be related to any aspect of the waveform shape including current amplitude, duration, rise, and decay. Electrical current delivered via implanted electrodes requires special waveform considerations to avoid charge accumulation that can result in electrode corrosion, tissue injury, or both.⁶⁹ Consequently devices designed to deliver transcutaneous stimulation might not be safely used with percutaneous or other implanted electrodes. Pulsed or AC current can be delivered in trains or bursts, where the frequency of bursts is known as the carrier frequency. Waveforms have been studied extensively with respect to stimulation-induced pain and tolerance because transcutaneous stimulation, despite preferential recruitment of larger nerve fibers, tends to stimulate smaller cutaneous pain fibers because of their proximity. Results have varied, however, and no consensus has been reached regarding the most tolerable waveform. Electrical stimulation delivered via implanted electrodes can bypass cutaneous nociceptive nerve fibers and results in significantly less discomfort.¹⁰⁵ Waveform examples are shown in Figure 21-1.

FIGURE 21-1 Diagrammatic representations of waveforms and modulations.

Stimulation intensity is typically regulated via current amplitude, voltage, or pulse width. Increasing these parameters generates greater contractile force with associated stimulation-induced pain.^12,^43,⁵⁷ Stimulators are either current or voltage regulated. Current-regulated stimulation provides (adjustable) constant current despite changing resistance that can be because of reduced contact between electrode and skin interface or variable impedance resulting from tissue movement. Current-regulated stimulators have safety mechanisms that prevent the delivery of excessively high voltage when excessive impedance is encountered. Increasing stimulation frequency also results in greater contractile force but is associated with greater muscle fatigue and stimulation-induced pain.^11,⁴⁹ The lowest stimulation intensity and frequency that achieves the desired physiologic or functional effect is generally recommended. The lowest stimulation frequency that results in fused muscle contraction can be as low as 12 Hz. Frequencies of 12 to 16 Hz for upper limb applications and 18 to 25 Hz for lower limb applications have been recommended.⁸⁹ The duty cycle refers to “on” and “off” time during cyclic stimulation. The duty cycle can be described by on:off time (e.g., 10 seconds:10 seconds) or as a percentage of on time (e.g., 50%.) Stimulation is typically ramped on by linear increases in stimulus intensity, with effects on graduated force generation. Stimulation can also be ramped off.

System Components

NMES systems are basically composed of a stimulator, a power source, electrodes, and either wired or wireless connections between these components. Wireless connections commonly use a radiofrequency link. Electrodes are either external, placed on the skin surface (transcutaneous), or implanted, where the electrically active surface of the electrode is placed internally, near or on the target nerve structure. The latter are referred to as “implanted systems,” although not all the system components are necessarily implanted. Implanted systems have expanded rapidly because of inherent advantages that include less stimulation-induced pain, more reliable electrode placement, the capacity to selectively stimulate deeper structures, no need to don and doff implanted components, and greater practicality (particularly for interventions that require a larger number of stimulation channels). Many electrical stimulation applications, whether functional or therapeutic, require hours of daily use. In these cases the issues surrounding practical and reliable implementation at home and in the community are paramount.

Although implanted systems are expanding rapidly, transcutaneous NMES systems are more commonly used in clinical practice because of their availability, lower cost, safety, and noninvasive implementation. Optimal clinical use of transcutaneous NMES requires experience and knowledge beyond the scope of this chapter. Clinicians who use transcutaneous NMES in clinical practice are referred to a practical guide developed at Rancho Los Amigos National Rehabilitation Center.⁷ A number of factors determine whether current delivered through a transcutaneous electrode will be adequate to cause neural excitation and subsequent muscle activation. The neural excitation will occur if the accumulation of charge (charge density) at the target neural structure reaches a threshold for depolarization that initiates an action potential (stimulus threshold.) The sum of the resistive and capacitive elements (i.e., electrode or tissue impedance) that oppose the flow of current affects charge density at the target neural structure and can be influenced by electrode size, orientation, configuration, and location. Current flows in a path of least resistance (Kirchoff’s principle). Most of the applied current bypasses the target nerve and dissipates through interstitial fluids. Humans are composed of heterogenous tissues with variable electrical resistance and capacitance (i.e., impedance). Tissue impedance is inversely proportional to water content. The water content of muscle, fat, epidermis or bone are approximately 75%, 15%, and 5% to 16%, respectively.⁹¹ Even a given type of tissue such as muscle is nonhomogenous with respect to tissue impedance. Current conducts about four times better in parallel to muscle fiber orientation compared with transversely across fibers. The clinical implications of this for electrode orientation and configuration are most significant when stimulating larger muscles.^34,⁹¹

Transcutaneous electrodes come in a variety of sizes and shapes. The impedance within the water-based gel on the conducting surface of the electrode is much less than that of epidermis and subcutaneous fat. Electrode size, shape, and orientation influence the electrical field as it enters the body. The cathode is the negative electrode. Positively charged ions are drawn to the cathode, creating a negatively charged environment external to the neuron that results in a reduction in the transmembrane potential. If the stimulus threshold is reached, explosive opening of voltage-dependent sodium ion (Na⁺) channels results in the generation of an action potential. The relatively greater neuromuscular activation at the cathode compared with the anode should be considered in clinical applications. For example, during transcutaneous NMES to reduce shoulder subluxation in hemiplegia, the cathode is applied to the posterior deltoid where the greatest muscle activation is desired. The anode is applied over the supraspinatus fossa to minimize activation of the upper trapezius muscle, which lies superficial to the supraspinatus muscle, thereby minimizing uncomfortable shoulder shrugging. Electrode configuration is often referred to as monopolar or bipolar. In a monopolar configuration, the cathode (“active”) is placed over the target stimulation site and the anode (“inactive”) is placed remotely, usually on the muscle tendon. Current density is high near the cathode, and volumetric dispersion is large, with only a fraction of the current returning through the anode. In a bipolar configuration, where the anode and cathode are placed in close proximity (typically a few centimeters), the current density between electrodes is maximized.³⁹ Maximizing current density to the target neural structure is desirable to minimize the overall current delivered, because muscle activation will occur with less pain. When the target neural structure is deep, current dispersion over a larger volume might be needed, requiring a greater distance between electrodes.

Implanted NMES systems take on many different forms. The number and variety of implanted systems have expanded rapidly. A standardized nomenclature has yet to be developed. Conventional terms tend to be descriptive but can lead to confusion. For example, percutaneous can refer to the method of implantation or an electrode with percutaneous leads. All implanted systems use electrodes where the stimulating surface is placed internally, but other components might be internal or external. Implanted systems range from those that use electrodes with percutaneous leads in conjunction with other components that are not implanted, to systems that are fully implanted within the body. Implantation of some devices requires surgery with general anesthesia, whereas other devices can be implanted using a percutaneous approach with local anesthesia during a brief outpatient procedure. From a clinical practice standpoint, a detailed discussion for implanted systems is not as relevant as for transcutaneous systems, because implanted systems are implemented either under research protocols or with special training and technical support. Various implanted systems are described in subsequent sections.

Clinical Applications for the Upper Limb

Clinical NMES applications have been historically categorized as therapeutic or functional. Although the goal of most, if not all, rehabilitation interventions is to enhance function, therapeutic electrical stimulation (TES) applications are designed to directly provide therapeutic benefits such as promoting neurologic recovery or reducing pain. Therapeutic benefits are intended to persist after TES is discontinued. Improvement in function occurs indirectly as the result of the therapeutic effects. Functional electrical stimulation (FES) provides functional benefits directly while the electrical stimulation is being delivered, but the benefits do not persist when stimulation is discontinued. The concepts of TES and FES will likely merge in the near future, particularly for stroke motor recovery, because adaptive neural plasticity is optimized when movement occurs within a functional context. Typical FES applications are based on stimulation of various muscles in a coordinated sequence to perform functional tasks under the user’s control. The classic example of FES is the upper limb neuroprosthesis, which has been generally applied to persons with cervical spine injury and more recently to survivors of hemiparetic stroke.

Neuroprostheses

An advanced, surgically implanted neuroprosthesis has been developed for spinal cord–injured individuals with C5–C6 motor levels, or more precisely, groups 0, 1, and 2 of the International Classification for Surgery of the Hand in Tetraplegia.⁵⁹ These are individuals who retain sufficient volitional movement of the shoulder and elbow (often with posterior deltoid to triceps tendon transfer to improve elbow extension) but who lack volitional control of muscles below the elbow to benefit from tendon transfer surgery to improve prehensile function. The system, formerly known as the Freehand System (NeuroControl Corp, Valley View, OH), was commercially available in the United States and Europe for several years (Figure 21-2). The Freehand system included eight channels of stimulation to provide grasp and release in palmar prehension for larger objects, such as a drinking glass, and lateral pinch for smaller objects such as an eating utensil. The electrodes, receiver-stimulator, and leads were surgically implanted. A radiofrequency inductive link to the receiver-stimulator provided power and real-time proportional control of prehensile function. The system was controlled by the user through a device worn on the contralateral shoulder that could detect various shoulder movements. A multicenter clinical trial of 50 subjects implanted with the neuroprosthesis demonstrated significant improvement in activities, regular use of the device in the community, and greater than 90% satisfaction among users.⁷⁴ Satisfaction and patterns of use persisted beyond 4 years for most users.¹⁰⁴ There were no reported cases of neuroprosthesis failure. Lead failure and infection rates were less than 1% and 2%, respectively.⁸⁹ The Freehand was commercially available for several years, but eventually was removed from the market by the manufacturer for many reasons that need to be understood by those who seek clinical implementation of such devices. Suffice it to say that developing an effective electrical stimulation intervention and conducting sound science to demonstrate efficacy and safety are not enough to ensure societal benefit. Many factors beyond science and engineering affect medical utilization and commercial viability.

FIGURE 21-2 The Freehand system, a surgically implanted neuroprosthesis for spinal cord–injured individuals with C5–C6 motor levels.

Upper limb neuroprostheses for high tetraplegia (C1–C4 neurologic level) have been evaluated to a much lesser extent. One case report describes a C3 tetraplegic individual who was able to feed himself using a percutaneously implanted eight-channel neuroprosthesis, a mobile arm support, and a universal cuff after setup following an 8-week program of stimulated muscle conditioning.¹⁰⁸ Musculoskeletal modeling suggests that 24 channels might be needed, including four channels for electromyographic (EMG) control from face and neck muscles.⁵² The overall feasibility of an upper limb neuroprosthesis for high tetraplegia remains questionable for a number of reasons, including the need for stimulation of muscles to provide shoulder and elbow control, and the need for more stimulation channels and fewer options for user control. Further, denervation to critical myotomes is more prohibitive, compared with lower cervical neurologic levels. Traumatic spinal cord injury (SCI) often results in a zone of injury where injury to anterior horn cells results in lower motor neuron injury that can encompass three to four spinal segments adjacent to the neurologic level. Although this is a potential challenge for FES at any neurologic level, C1–C4 neurologic levels are often accompanied by an injury zone that results in denervation to critical myotomes responsible for control of the shoulder and elbow. An effective upper limb neuroprosthesis for individuals with high tetraplegia has not yet been developed.

Several upper limb neuroprostheses have been developed for survivors of hemiparetic stroke by using transcutaneous or percutaneously placed intramuscular electrodes, with suggested improvements in motor impairment and selected laboratory-based activities in small, nonblinded clinical trials.^3,^19,^67,⁸⁵ These studies provide proof-in-concept that a neuroprosthesis can be developed for stroke survivors with upper limb paresis. The barriers to developing an effective upper limb neuroprosthesis for persons with stroke need to be addressed, however, before significant improvements in community-based activities can be expected. First, synergistic patterns of muscle activation and spasticity inhibit volitional movement after stroke and tend to be exacerbated during the performance of functional tasks, particularly when greater amounts of effort are required. As a result, electrically stimulated movement tends to be more easily controlled when the limb is at rest, but abnormal muscle activation patterns can become prohibitive during neuroprosthesis-facilitated, goal-oriented movement. Traditional spasmolytic medications have not been effective adjuncts to improve targeted movement while using a neuroprosthesis.⁸⁹ Focal, dose-dependent chemodenervation with botulinum toxin and other neurolytic procedures warrant further evaluation as adjuncts to the use of upper limb neuroprostheses in hemiparetic persons. Second, control strategies should not limit functional use of the neurologically intact limb. Third, the functional demands of a neuroprosthesis in the setting of unilateral (hemiplegia) as compared with bilateral (tetraplegia) upper limb paresis might be higher. At the minimum a neuroprosthesis for stroke must facilitate simple bimanual tasks. The NESS H200 (Bioness, Inc, Valencia, CA) is an exoskeletal device marketed as an upper limb neuroprosthesis (Figure 21-3). The H200 is the only commercially available device that has been shown to improve selected laboratory-based activities in stroke and tetraplegia, and bears mentioning for this reason.^2,³ Transcutaneous electrodes are fastened into the exoskeleton, resulting in more reliable electrode placement and thus easier donning that facilitates repetitive stimulation to prehensile muscles in a coordinated sequence. However, the device remains subject to the clinical barriers described earlier. Available control strategies limit its utility for performing functional activities in the real world. An upper limb neuroprosthesis for stroke that improves community-based activities has yet to be developed.