Artifact in Autonomic Neurophysiology Studies
William P. Cheshire Jr.
The past few decades have witnessed major advances in increasingly sophisticated, noninvasive neurophysiological tests to probe one of the most developmentally ancient of human body systems. The autonomic nervous system is a vast network of neurons that coordinate visceral activity throughout the body to ensure internal homeostasis and respond to external stress. Most autonomic responses are involuntary and occur below the level of conscious awareness. As autonomic dysfunction can be a hallmark or epiphenomenon of many diseases, as well as an important cause of distress and debility for patients, diagnostic testing of autonomic responses is a valuable aspect of clinical neurophysiology.
THE AUTONOMIC NERVOUS SYSTEM
Structurally the autonomic nervous system comprises central and peripheral interconnected components. The central autonomic network includes neurons located in the hypothalamus; insular, anterior cingulate, and ventromedial prefrontal cortices; stria terminalis, amygdala; periaqueductal gray matter; parabrachial nucleus, nucleus of the solitary tract; ventrolateral medulla; medullary lateral tegmental field; and intermediolateral cell column of the spinal cord.
The peripheral autonomic nervous system is organized into sympathetic, parasympathetic, and enteric divisions. The sympathetic nervous system, which is often characterized as coordinating “fight-or-flight” responses, flows from thoracolumbar segments of the spinal cord and contrasts with the parasympathetic nervous system, which has more of a role in “rest-and-digest” functions and flows from cranial nerves III, VII, IX, and X, and sacral spinal segments. Although sympathetic and parasympathetic behaviors are often antagonistic, some are functionally distinct and unopposed, whereas others are interdependent. Additionally, the enteric nervous system consists of ganglionic plexuses intrinsic to the wall of the gut.
Autonomic responses entail a sequential, two-neuron efferent pathway in which a preganglionic neuron synapses onto a postganglionic neuron that innervates the target organ. Sympathetic synapses are found in the paravertebral sympathetic chain ganglia, whereas parasympathetic synapses are found in ganglia near their target organs.
AUTONOMIC REFLEX TESTING
This chapter focuses on the components of autonomic testing of greatest importance to neurologic practice (1). In clinical neurophysiology laboratories, three categories of autonomic testing are well established. Sudomotor testing (CPT code 95923) evaluates sympathetic cholinergic function in the secretion of sweat from eccrine glands. Cardiovagal testing (CPT code 95921) evaluates parasympathetic function in the control of heart rate. Vasomotor adrenergic testing (CPT code 95922) evaluates sympathetic noradrenergic function in the regulation of blood pressure. The methodology and proper performance of these tests have been reviewed elsewhere (2).
238Many other tests of autonomic function are available in other specialties. These include tests of circulating catecholamine levels and renin for the endocrinologist, pharmacologic tests of pupillary dilation and constriction for the ophthalmologist, electrocardiography or electrophysiologic testing of the heart for the cardiologist, urodynamic testing for the urologist, and gastric and bowel motility testing for the gastroenterologist. As these are not typically part of the neurologist’s repertoire of clinical tests, they are not addressed here.
In routine clinical practice, autonomic neural activity cannot be measured directly. Rather, the autonomic end-organ response is measured. Clinically meaningful information may be present at rest or, more typically, is obtained by measuring the evoked response to a standardized stimulus under controlled conditions.
SOURCES OF ARTIFACTS
Six categories of potential artifacts may occur during autonomic testing (Table 10.1). For tests that require the patient’s participation, inadequate cooperation can decrease autonomic responses. Also, the patient may have concurrent medical conditions that influence autonomic responses or impair his or her ability to participate in the test. Importantly, a wide variety of medications can potentially increase or decrease autonomic responses. Additionally, ambient conditions in the testing room may influence the patient’s autonomic responses. Further, testing equipment malfunctions can occur. Finally, the physician who lacks sufficient training or experience in autonomic disorders may misinterpret the test results. Examples of these will be pointed out under each type of testing.
ARTIFACTS AT THE BEDSIDE
Although the focus of a textbook on artifacts in clinical neurophysiologic testing is necessarily on the ways in which technology is used and its data are interpreted in the course of patient care, the experienced physician is mindful that the history and physical examination are typically the most important elements of the diagnostic process. In regard to autonomic disorders, a careful history is the most important diagnostic test (3). Here too artifact can intervene.
An artifact can be defined as something crafted by art or as information generated by an artificial device. Insofar as taking a good history is an 239essential aspect of the art of medicine, artifact can occur if communication between the patient and the physician is incomplete, if pertinent questions are not asked, or if questioning follows too rigidly an algorithm or predetermined care process instead of being open-ended and inquisitive about the patient’s perspective. Artifacts in the medical history take the form of gaps in information or misinterpretations based on flawed assumptions. Technology also can introduce artifacts into the medical history. Computer software that constructs a medical document by transcribing, without understanding, text from the digital patterns of spoken audio input frequently introduces errors of language that can alter the intended meaning of a clinical note, sometimes in surprising ways. The autonomic term lacrimation, for example, might be typed “like cremation.” Treatment with a course of steroids might be mutated into “coarse old stories” (4).
Source of Artifact
Concurrent medical condition
Respiratory muscle weakness
Dehydration with intravascular hypovolemia
Cold limb causing vasoconstriction
Vestibular dysfunction activated by tilt table motion
Carbonic anhydrase inhibitors
Uncomfortable heat or chill
Bright overhead light
Loud or sudden noise
Intravenous catheter causing vasovagal response
Electrical power failure
Loose electrical connections
Improperly sized blood pressure cuff
Pressure or fluid leakage from cables
Misinterpretation of data
Being unfamiliar with the range of normal
Noticing peaks or troughs rather than means
When evaluating the patient who presents with episodic collapse, an invaluable element of the art of the physical examination is the ability to measure very quickly the pulse and blood pressure at the moment that symptoms occur. Some healthy patients will experience a transient fall in blood pressure upon standing that recovers to normal values within 20 seconds. The patient whose systolic blood pressure falls below about 70 mmHg or in whom recovery is prolonged may, after 7 seconds of cerebral hypoperfusion, lose consciousness. One of the diagnostic challenges in evaluating such spells is distinguishing orthostatic syncope, which is caused by hypotension, from psychogenic pseudosyncope, in which blood pressure remains normal during apparent loss of consciousness (5). If one knows the blood pressure at baseline, with practice one can develop the skill of ascertaining a significant change in blood pressure very rapidly using a sphygmomanometer and stethoscope. Anticipating that a spell may occur, the neurologist has the blood pressure cuff already in place when asking the patient to stand up. At the moment the patient starts to collapse, and while preventing the patient from falling, the neurologist inflates the cuff to just above the baseline value and, while auscultating the brachial artery, deflates the cuff rapidly. In this way, accurate blood pressure measurements can usually be obtained within a few seconds. In many clinical settings, however, physicians and nurses increasingly are relying on automated blood pressure devices that take about 35 seconds to obtain a blood pressure reading. That time may be too long to correlate with symptoms if the blood pressure has already recovered by the time the machine registers a value. If the blood pressure is below the range of pressures the machine is designed to measure, it may return an error message (Figure 10.1), an artifact that is of little value compared to personally obtained blood pressure measurements.
ARTIFACTS DURING SUDOMOTOR TESTING
Tests of sudomotor function are commonly utilized in assessing small fiber neuropathy and disorders of thermoregulation. A number of methodologies are in common use (6). The thermoregulatory sweat test subjects the patient to a controlled, gradual heat stimulus to produce a generalized sweating response, the anatomical distribution of which is visualized by the aid of a colorimetric indicator powder, recorded, and quantified. The quantitative sudomotor axon reflex test utilizes acetylcholine, which is applied at four standardized limb sites by transcutaneous iontophoresis to reach subepidermal peripheral postganglionic sympathetic cholinergic sudomotor neurons, which then activate innervated eccrine glands. The latency and volume over time of the resulting secretion of sweat outflow is recorded. Another method is the silastic imprint technique, which evaluates the number and size of evoked sweat droplets per measured area of skin.
240All sudomotor tests are subject to the potentially confounding effects of medication. If the patient is taking an anticholinergic agent or a drug that blocks carbonic anhydrase, it may not be possible to distinguish hypohidrosis or anhidrosis on testing from a small fiber peripheral neuropathy affecting sudomotor nerves. For this reason, it is preferred that patients hold all anticholinergic agents (including tricyclics, bladder and gut antispasmodics, and antihistamines) and carbonic anhydrase inhibitors (topiramate and acetazolamide) for five half-lives prior to testing. These medications do not always interfere with sudomotor responses, but in some patients they can exert a potent effect, and it is not possible to predict in advance which patients are susceptible. The many drugs that can affect sudomotor responses have been extensively reviewed elsewhere (6,7).
It is sometimes assumed that, whereas a distal small fiber neuropathy might impair the sudomotor response only at the foot, a drug effect would be seen uniformly across sites; this does not always hold true. Distally predominant drug effects sometimes occur (Figure 10.2).
Other medications, particularly opioids, selective serotonin and norepinephrine reuptake inhibitors, and stimulants can enhance sudomotor responses. Opioid-induced hyperhidrosis is not uncommon. As drug-induced stimulation of sweating is typically mediated centrally, assessment of peripheral sudomotor nerve function is generally not impaired. Pyridostigmine is another example of a pharmaceutical agent that can enhance the sweating response (Figure 10.3).
ARTIFACTS OF HEART RATE DURING CARDIOVAGAL TESTING
Assessment of parasympathetic cardiovagal function is accomplished by quantifying the variation in heart rate (or change in the RR interval). A number of clinical techniques exist. The most sensitive, and the least specific, measures the difference between the increase in heart rate during inspiration and the decrease in heart rate during expiration (respiratory sinus arrhythmia) under conditions that maximize this difference (Figure 10.4). The Valsalva ratio is an index of baroreflex–cardiovagal function and is obtained by having the patient exhale against resistance and calculating the ratio of the increase in heart rate to the decrease in heart rate that occurs when straining ceases. Another test is the 30:15 ratio, in which the longest RR interval at the 30th beat after standing is divided by the shortest RR interval at the 15th beat. Cardiovagal function can be assessed also by fast Fourier transformation of ECG spectra in the time or frequency domain.
When testing heart rate variability to deep breathing, if the patient does not take slow, complete, deep breaths in timing with the visual sinusoidal stimulus that, like the conductor of an orchestra, directs the patient’s breathing, the magnitude of the response will be decreased (Figure 10.5). Thus, it is advisable to monitor simultaneously chest wall movement, and for the Valsalva maneuver, exhaled respiratory pressure to ensure that they are adequate in volume and force to induce the desired autonomic response.
The timing of chest wall movement in relation to changes in heart rate should also be heeded in order not to mistake a cardiac dysrhythmia for a normal autonomic response. The patient with atrial bigeminy, for example, will exhibit a heart rate that oscillates rhythmically between a high value and a low value, but the rate of change varies with the heartbeat, not with the slower respiratory rate (Figure 10.6).
In other patients, atrial bigeminy may be continuous, in which case it may be unclear how the maximum and minimum heart rates should be defined for purposes of evaluating neurally mediated heart rate variability. In such cases, it may be reasonable, when calculating the Valsalva ratio, to draw measurements from every other beat (Figure 10.7).
Valid assessment of cardiovagal function requires accurate assessment of beat-to-beat heart rate, which is typically accomplished by measuring the RR interval on a single-lead ECG. It is sometimes necessary to reposition the precordial or limb leads or select among electrode pairs in order to obtain distinct R waves that are sufficiently high in amplitude and consistently higher than the T waves. Occasionally, the ECG waveform will change during testing, and if the R waves decrease in amplitude or the T waves increase in amplitude, the software may erroneously count some of the T waves as R waves and falsely report a doubling of heart rate (Figure 10.8).
Atrial fibrillation or flutter is recognized by the characteristically erratic pattern on the beat-to-beat heart rate tracing (Figure 10.9). Sometimes, but not always, the slower, regular oscillation of neurally mediated heart rate variability can be discerned in the background.
Premature ventricular contractions are another source of artifact arising from the cardiac conduction system that may be mistaken for neurally mediated autonomic responses arising from baroreflex–cardiovagal pathways. An occasional premature ventricular contraction may appear on the beat-to-beat heart rate tracing as a sudden blip or series of blips of high amplitude in which an apparent single-beat increase in heart rate is immediately followed by an apparent single-beat decrease in heart rate to below baseline (Figure 10.10). When a series of premature ventricular contractions (PVCs) occurs during testing of heart rate variability, the peak and trough values associated with the PVCs should not be counted. Rather, the slower, smoother, sinusoidal profile of heart rate variability may be appreciated in the background (Figure 10.11).
241Muscle artifact is another potential source of electrical interference with the ECG signal. As autonomic studies do not typically include electromyography, this artifact may appear only on the ECG tracing, typically as a single or clustered, arrhythmic group of heart rate peaks that are much higher than the baseline recording (Figure 10.12).
The recent introduction of personal biosensor technologies that allow patients and healthy individuals to detect heart rate changes during daily life has created a new source of artifact. Whereas, until now, detailed heart rate information had been available only to the autonomic nervous system or to medical professionals, now these devices have made heart rate information accessible to patients’ conscious awareness and amateur scrutiny. It should be recognized that data acquired from personal health monitoring devices are subject to selection bias, as patients may be more likely to look at the device’s display when they are experiencing palpitations or other symptoms. Patients without medical or sports physiology background may not know the normal range of heart rate and may misinterpret a normal transient increase in heart rate as a dysautonomia (8).
ARTIFACTS OF BLOOD PRESSURE DURING THE VALSALVA MANEUVER
The assessment of both baroreflex–cardiovagal and baroreflex–sympathoneural function require the rapid assessment of blood pressure on a beat-to-beat basis. This is accomplished by technologies such as photoplethysmography, which quantifies blood pressure indirectly by measuring the absorption of infrared light transmitted through the finger as an indicator of changes in the volume of blood flowing through. This method is insensitive as a measure of absolute blood pressure, for which reason autonomic testing utilizes simultaneous sphygmomanometer recordings, but it is an exquisitely sensitive test of what is most important for autonomic testing: moment-to-moment changes in blood pressure.
Beat-to-beat devices will display not only information about blood pressure, but also artifact. A sudden increase in apparent blood pressure in response to the patient’s emotional state, for example, could reflect a withdrawal of parasympathetic influence on heart rate, a surge of circulating catecholamines causing vasoconstriction in the fingers, or a subtle limb movement causing the device to lose track of its signal (Figure 10.13).
Additionally, coughing, sneezing, or laughing will momentarily alter the beat-to-beat blood pressure tracing. Merely speaking can, in some patients, elevate the heart rate by 5 to 10 beats/min and the blood pressure by 5 to 10 mmHg. The hemodynamic response to coughing or sneezing parallels that of the Valsalva maneuver (Figure 10.14). Profiles seen during laughing are brief and more variable, as the increased respiratory force is typically of shorter duration (Figure 10.15).
When performing the Valsalva maneuver, if the patient does not maintain a sufficient expiratory force (usually 30–40 mmHg), the magnitude of the elicited reflexes will be decreased. In these cases the physician might mistakenly conclude that the patient who has respiratory muscle weakness or who is uncooperative has autonomic failure.
ARTIFACTS OF BLOOD PRESSURE DURING TILT TABLE TESTING
Orthostatic hypotension, as defined by an international consensus committee representing the American Autonomic Society and the American Academy of Neurology, is a sustained reduction of systolic blood pressure of at least 20 mmHg or diastolic blood pressure of 10 mmHg within 3 minutes of standing or heat-up tilt to at least 60° on a tilt table (9). A number of variations on this broad definition are encountered in clinical practice and can be discerned by autonomic testing (10). Among them is an artifactual pattern that only appears to be orthostatic hypotension because initial supine blood pressure values are not verified to be a true baseline. In order to establish a valid baseline, the patient should be recumbent at rest for at least 10 minutes, if not longer. A blood pressure that decreases from an elevated baseline, barely meets the threshold for orthostatic hypotension when the table is upright, and does not increase when the table is returned to the horizontal position usually does not indicate a neurogenic form of orthostatic hypotension (Figure 10.16).
Another important distinction is between artifactual drops in blood pressure and the transient fall that occurs during neurally mediated syncope. In neurally mediated syncope, a gradual increase in heart rate typically precedes a gradual, smoothly contoured fall in the blood pressure tracing that occurs over about 30 to 60 seconds (Figure 10.17). Artefactual drops, by contrast, exhibit a different temporal profile. Artefactual deviations in blood pressure may abruptly decrease or increase during slight changes in finger cuff position, they may exhibit a stuttering pattern of inconsistent pulse detection, or they may taper off very slowly as pulse detection diminishes in the patient with peripheral vasoconstriction (Figure 10.16). Monitoring of beat-to-beat blood pressure during tilt table testing represents a special challenge, as the movement of the table may cause the patient’s arm position to shift, and even a slight shift of position can interrupt the delicate positioning of the photoplethysmographic finger cuff in relation to the patient’s pulse. For valid determinations, the finger cuff should be maintained at the level of the heart, and the blood pressure should always be confirmed by sphygmomanometry in the opposite arm.
242Not only decreases, but also increases in blood pressure require interpretation of factors beyond the autonomic responses of primary interest. Among the situational factors that can cause an increase in blood pressure during autonomic testing are pressor drugs, anxiety, and pain (Figure 10.18). The differential diagnosis of paroxysmal hypertension also encompasses the common phenomenon of “white coat” hypertension and the very rare condition of pheochromocytoma.
Oscillations in blood pressure are frequently observed during tilt table testing and can be a sign of orthostatic intolerance in the patient whose autonomic nervous system is less capable of responding to the upright posture with the physiologic changes necessary to adapt to the force that gravity exerts on the cardiovascular system. As limb muscle tensing also can produce oscillations in blood pressure, it is important to observe the patient throughout testing and ensure that the arms and legs remain relaxed. Tremors in patients with movement disorders are much faster (3–5 Hz for Parkinsonism, 4–11 Hz for essential tremor) than sympathoneural blood pressure oscillations (Figure 10.19) and generally do not interfere with the assessment of beat-to-beat blood pressure unless they are of sufficiently high amplitude to interrupt detection of the finger pulse.
For blood pressure oscillations at frequencies of 5 to 20 Hz, correlation with respiratory rate may demonstrate a corresponding profile. In such cases the oscillations are a manifestation of respiratory sinus arrhythmia, which consists of time-locked oscillations in both beat-to-beat heart rate and blood pressure (Figure 10.20).
ARTIFACTS OF HEART RATE DURING TILT TABLE TESTING
Spontaneous variations in heart rate during tilt table testing, both supine and upright, that correspond in frequency to chest wall movement very often reflect respiratory sinus arrhythmia and are thus normal and physiologic (Figure 10.21). It is important to bear in mind this physiological pattern when assessing the overall change in heart rate accompanying the upright posture.
The postural tachycardia syndrome, which is a form of orthostatic intolerance, is defined as a sustained heart rate increment of more than 30 beats/min within 10 minutes of standing or head-up tilt in the absence of orthostatic hypotension (9). On occasion, this diagnosis is made incorrectly on the basis of a transient peak in heart rate exceeding 30 beats/min above baseline rather than a sustained increment. When analyzing tilt table hemodynamic profiles, one’s eye can be drawn to the peak heart rates, which stand out from the other values. The published consensus definition, however, was based on normative values of mean heart rates during tilt table testing, for which reason the mean values, and not just the peaks, must be considered when distinguishing abnormal from normal heart rates (Figure 10.22).
The potential influence of medications on heart rate must also be taken into account when deciding whether a heart rate change is normal or abnormal. Among the common medications that can cause tachycardia, and hence the appearance of postural tachycardia syndrome on autonomic testing, are caffeine, pseudoephedrine, amphetamine, methylphenidate, phentermine (Figure 10.23), norepinephrine reuptake inhibitors, and anticholinergic agents (11).
This chapter has described and explained some of the most common artifacts encountered when performing autonomic testing. Many other types of artifacts can also occur, so varied are patients’ disorders and the conditions of testing among neurophysiology laboratories. In all of these cases, there can be no substitute for knowledge about autonomic physiology and pharmacology, astute clinical observation, and boundless curiosity about the complexities of the human nervous system.