Plethysmography: Safety, Effectiveness, and Clinical Utility in Diagnosing Vascular Disease

Oculoplethysmography

Atherothrombotic lesions of the cerebral circulation most frequently form where the common carotid artery bifurcates into the internal carotid artery (ICA) and external carotid artery (ECA). Strokes and transient ischemic attacks are caused primarily by emboli (dislodgement of cholesterol plaque or aggregated platelets)(22) that occlude blood flow distal to the circle of Willis or gradually stenose the artery at or around the common carotid bifurcation with extension into, or occlusion of, the ECA or the ICA.

Several imaging techniques exist to estimate the degree of stenosis or occlusion of extracranial vessels (Table 2). These include MRA, pulsed Dopper arteriography, real-time B-mode ultrasound, DU, color-coded echo flow arteriography, and CA. Both linear and area cross-sections of the carotid vessels can be calculated, and estimates of partial occlusions can be made.(23-25)

Nonimaging techniques (Table 3) may also be used to estimate changes in arterial blood volume or velocity distal to the suspected lesion by oculoplethysmography, oculopneumoplethysmography, periorbital Doppler examination, ophthalmodynomometry, SPPG, or, rarely, thermography. Although incapable of measuring flow directly, demonstrating anatomic information or quantifying percent of stenosis, OPG has been used as an initial semiquantitative evaluation of the patency of the common carotid artery, its bifurcation, and the internal and external carotid arteries and their tributaries (facial, ophthalmic) in symptomatic patients or in asymptomatic patients with bruits.

With OPG, it is possible to compare both the ocular arterial blood pressure (suction ophthalmodynamometry) and carotid blood flow (ocular waveform analysis) based on the following principles: 1) despite wide variation in human body size, there is little variance in the size of the eye; 2) external pressure applied to the eye increases the intraocular pressure; and 3) inward and outward distortion of the sclera elevates intraocular pressure. Also, a stenotic lesion poses a fixed resistance to flow, causes hemodynamic energy loss as blood flows through the stenotic area, slows arterial inflow (to the ophthalmic artery), and ultimately reduces perfusion to the eye.(26) A positive OPG test may include an estimate of ophthalmic pressure and/or volume within the globe. The reliability of OPG depends on its ability to detect the presence of stenosis or occlusion by recording a difference in flow (amplitude) between the left and right carotid circulation, which is expressed as a differential pulse arrival time, and a comparison of the ophthalmic and brachial systolic pressures, expressed as a ratio (SOB ratio).

Three methods of OPG have been introduced into clinical practice over the last 30 years by Kartchner (OPG-K), Gee (OPG-G), and Zira (OPG-Z).(24-58) The main differences among the three methods are the type of transducer (air or water) used with the eye cups and the reference pulse or pressure measurement used for comparison (brachial systolic pressure, ear pulse arrival time). For example, one method, OPG-G, compares the SOB ratio, whereas OPG-Z compares bilateral ocular filling times as well as bilateral eye-to-ear pulse waveforms. All OPG-G techniques are noninvasive, relatively simple to perform, require minimal technical expertise, and can be performed within a short period of time (less than 30 minutes).

Kartchner's OPG method (OPG-K) (Table 4) simultaneously records relative filling times of the distal carotid bed by ocular sensors attached to water-filled suction cups held in place on each anesthetized cornea and, concomitantly, by densitometers attached to the earlobes.(27) It compares the filling time of one ophthalmic artery with that of the contralateral artery.(28) Ideally, there should be no delay in either eye or ear pulse recordings. However, if an ocular delay is detected on one side, it reflects flow reduction in the ipsilateral ICA, whereas a delay in one of the ear tracings signifies ipsilateral ECA disease. This claim has been disputed.(29) The test is performed without carotid artery compression or induced ocular hypertension. In the studies published, patients with systemic hypertension, cardiac arrhythmias, glaucoma, or corneal opacification were not excluded from testing.

Carotid Phonoangiography

Proponents of OPG-K introduced the sequential use of carotid phonoangiography (CPA) for the diagnosis of CAD. It entails the study of sound waves and flow turbulence originating from bruits along the carotid arteries. Also developed by Kartchner and McRae in the early 1970s,(27) it consists of electronic stethoscopic ausculation with a high-fidelity microphone and loudspeaker and direct visualization by oscilloscope, camera, and possibly a recorder.(37) The microphone captures a wide band of audiofrequencies that are not discernible to the human ear, and the oscilloscope and camera provide a permanent record. Readings are taken at three levels (low, mid, and high) along the carotid architecture, and the user can discern the location of the bruit by the increasingly turbulent sounds. The introduction of sequential CPA was believed to be necessary to increase the sensitivity of OPG-K. Table 5 lists the results of clinical studies that compared the accuracy of CPA alone with carotid arteriogram. Sensitivities ranged from 40 to 88, specificities 51 to 96, positive predictive values (PPVs) 38 to 88, and negative predictive values (NPVs) 67 to 93. A correspondingly high range of false-negative and false-positive rates (12 to 60 and 4 to 49, respectively) were reported. Carotid phonoangiography was not useful in patients with carotid stenosis without bruits or in patients with total occlusions and could not differentiate between ICA and ECA bruits when present.

It has been further demonstrated that the presence or absence of a bruit does not correlate with the degree of CAD.(39) Ginsberg et al(40) used CPA with OPG-K to assist in ruling out bilateral carotid disease when no delay between the left and right ocular waveforms is present and to assist in the detection of stenotic lesions that do not greatly reduce carotid artery flow. Three anatomic sites were sampled: immediately below the angle of the mandible, several centimeters below the mandible at the carotid bifurcation, and a supraclavicular carotid segment to sample the common carotid artery. The amplitude of the bruits were estimated semiquantitatively (small, moderate, or large), displayed on an oscilloscope, and permanently recorded on Polaroid film. They demonstrated that an ocular-to-ocular delay interval of at least 12 milliseconds (msec) was necessary to maximize the accuracy of the OPG-K system although sensitivity suffered (reduced from 91 to 60), whereas specificity rose from 45 to 90. When a 12-msec delay interval was used as the criterion for abnormality, rather than a 4-msec delay, false positives were reduced from 55 (17 patients) to 9 (3 patients). The accuracy of CPA in bilateral disease in this group of patients was 67. Table 4 summarizes the results of almost 2,700 vessels evaluated by OPG-K. Sensitivities ranged from 40 to 93, whereas specificities ranged from 46 to 96. The PPVs ranged from 38 to 94 and NPVs ranged from 67 to 97.

Oculopneumoplethysmography (Gee's OPG method) indirectly measures the ophthalmic artery systolic pressure in each eye by applying and calibrating air-filled suction cups on the anesthetized corneas to -300 mm Hg pressure (-500 mm Hg in hypertensive patients). Over a 30-second recording period, the negative pressure is reduced and the systolic endpoint is noted (the time at which globe pulsation reappears).(41) Correlation of the mean ophthalmic systolic pressures with each other and with the respective mean brachial systolic pressures (SOB index) is the principle mode of interpretation for the OPG-G. Secondarily, systolic ophthalmic arterial blood flow is then calculated from the intraocular pressure at which eye pulsations resume, along with two variables, amplitude of the ocular pulse tracing and the heart rate, and a constant which represents the volume of ocular tissue under the eye cup. (43)

Clinical studies that have examined the accuracy of OPG-G with carotid angiography (Table 6) reported sensitivities from 0 to 98 and specificities from 85 to 100. Posttest probabilites demonstrated that OPG-G was slightly better at predicting those with disease (PPVs from 85 to 100) than those without disease (NPVs from 58 to 99).

Zira's oculoplethysmography method (OPG-Z) is an automated, digital, air-filled system that measures ocular pulse-wave forms using ocular cups held against each cornea with slight negative pressure. Both ear pulse wave forms, used to assess the ECA, are measured simultaneously by applying PPG sensors on each ear. Elapsed time in milliseconds between the rising portion of the pulsatile wave form from each eye and ear is displayed digitally as ear-to-ear delay, eye-to-eye delay, and eye-to-ear or ear-to-eye delay.(39,49) The following criteria were used for an abnormal interpretation: 1) > 30 msec ear-to-ear pulse delay -- suggestive of unilateral ECA stenosis on the slower side; 2) >10 msec delay in the eye-to-eye tracing -- suggestive of severe ICA stenosis on the delayed side; and 3) >30 msec delay in the ear-to-eye or eye-to-ear tracing -- suggestive of bilateral ECA stenosis or bilateral ICA stenosis, respectively. Those studies that measured the accuracy of OPG-Z are cited in Table 7.

Another variety of OPG (OPG-LS manufactured by Life Sciences, Inc.) is a pulse-volume recorder with an oculoplethysmographic attachment. Based on the same operating principles as OPG-G but measuring one eye at a time, OPG-LS was studied in 219 carotid arteries to determine its accuracy when compared with carotid angiograms.(50)

Lateralizing criteria for a positive OPG-LS were: systolic ocular pressure (SOP) difference >= 5 mm Hg between eyes; pulse-volume amplitude difference >= 3 mm at 45 mm Hg SOP; pulse-volume difference >= 1 mm Hg when SOP was >110 mm Hg. The lateralizing criteria alone correctly identified 9 out of 130 (7) less severely stenosed vessels (<= 60), and 51 out of 89 (57) severely stenosed arteries (>= 60). Forty-three percent of the vessels with severe stenosis had accompanying bilateral disease (38 out of 89), and the lateralizing criteria correctly identified 13 of those vessels (37). When the SOB index was added to the lateralizing criteria to signify a positive test, the yield in identifying less severe carotid stenosis increased from 9 to 22 out of 130 vessels (17) and from 51 to 54 out of 89 vessels (60) with severe stenosis. No mention was made of how many positive vessels were identified by SOB alone. Oculoplethysmography-LS is an insensitive diagnostic test with an overall sensitivity of 57, specificity of 7, and PPVs and NPVs of 30 and 19, respectively. It is particularly insensitive to bilateral carotid disease (37).

Spontaneous scleral ecchymosis does occur in about 3 of patients(43) tested with OPG and resolves spontaneously over several days without treatment. Transient loss of vision may occur in some patients as a result of the negative pressure generated during application of the ocular cups. This resolves spontaneously within several seconds. All of the OPG techniques require that patients be able to stay relatively quiet and motionless for several minutes. Calibration of the transducers and patient position are critical to correct interpretation of the results.

Improvements in both duplex scanning (direct real-time B-mode ultrasound with Doppler spectral analysis) and Doppler ultrasonic color-flow mapping have rendered plethysmography and other nonimaging techniques obsolete in those cases in which information on patency, anatomic location, and flow is essential.(52,53)

Supraorbital Photoplethysmography

Supraorbital photoplethysmography is not strictly a plethysmographic technique.(15) Rather, its operation is based on the principles of photodensitometry, using a near-infrared light-emitting diode (LED) light source and a phototransistor with a linear response over a wide range of intensities. The absorption of light by hemoglobin reduces the amount of backscatter to the receiver and is dependent on the volume of blood in the superficial dermal venous plexus.(54,55) A flat probe that contains both an infrared light source and phototransistor receiver is attached to the patient's forehead. Qualitative information about the patency of the ICA can be elicited by various compression maneuvers of the ECA. Supraorbital photoplethysmography has been used in conjunction or in sequence with other noninvasive diagnostic methods such as OPG, CPA, and Doppler ultrasound, but is not used as a stand alone diagnostic test.

Barnes et al(56) studied 156 ICAs in 78 consecutive patients "evaluated for cerebrovascular disease." They chose a very liberal definition of stenosis (>= 50) which accounted for the high sensitivity (100) and poor PPV (67). The authors claimed that in 10 of these 18 vessels, angiography underestimated the stenosis by 33 to 49, and, therefore, the SPPG results should not be considered false positives. Barnes later demonstrated that the predictive value of SPPG, validated by surgical specimens of carotid artery stenosis, was about 40.(56) Duke et al(57) studied 272 vessels with SPPG and carotid angiogram and found a similarly high false-positive rate (15).

Magnetic Resonance Angiography

Investigators are presently attempting to establish MRA as a reference method for the study of extracranial CAD (Table 8).

In an independent technology assessment on the clinical utility of MRA for the determination of blood flow and vessel morphology, Handelsman et al(75) concluded that MRA has demonstrated accuracies comparable to CA, but CA remains the definitive preoperative study.

This collection of studies, most of which were performed retrospectively in study populations "at risk or symptomatic for carotid artery disease," reveals that the reliability of OPG-K, OPG-G, and OPG-Z in broadly defined populations is extremely variable. The effectiveness of OPG in the detection of bilateral disease is dismal (Table 9).

Even with the sequential use of CPA, as is the case with OPG-K, sensitivity does not significantly increase. It is also interesting to note that OPG attains its highest sensitivity when both the differential time delays of the eyes (or eye and ear) and the SOB index are used to define a positive test. Some earlier studies used only the pulse differential criteria and missed bilateral disease in particular. Is it possible to differentiate the clinical utility of one OPG device over another, even in well-described subgroups, i.e., asymptomatic patients with or without carotid bruits? Based on the published information to date, this question cannot be answered. Each OPG method is significantly less accurate than the "gold standard" of carotid angiogram, although OPG-G has been demonstrated to attain 100 specificity in small study populations (66-72 patients) in which the prevalence of disease is high (55 to 88).

AbuRahma and Diethrich,(38) in a comparison of the three OPG methods, concluded that all OPG methods are "valuable," and the OPG-Gee was particularly useful in bilateral CAD. They reported overall accuracies of 88 to 95, and accuracy in bilateral disease ranged from 66 (OPG-Z) to 90 (OPG-G). They neglected to report the sensitivity and specificity of OPG in patients with bilateral disease and the number of those patients in each group. They failed to report any patient selection criteria, the comparability of patients among the three testing groups, and whether the investigators were blinded. The comparability of the OPG devices would have been better demonstrated had the test subjects been randomly selected patients who had undergone testing with all three OPG devices as well as with the same reference method.

The prevalence of carotid disease in the sample populations was very high (range, 65 to 70), and the false-negative rates ranged from 9 to 23. When prevalence is very high in a study population, it is possible for a comparatively poor diagnostic test to have a very high PPV. For example, at a disease prevalence of 70, even a test with 60 sensitivity will have a PPV of 93. This study and many others that were published supporting the clinical utility of OPG demonstrate this mathematical relationship quite well (Tables 4, 5, and 7).

The clinical utility of OPG has been particularly difficult to address for two reasons. First, recent innovations in imaging systems, such as DU, have not only replaced the reference method of carotid arteriography in many cases but have also become the initial diagnostic screening test for those patients believed to be at risk of stroke. Second, very few studies reported posttest outcomes in symptomatic patients. Even though the correlation of imaging and nonimaging systems may seem fraught with complexities, some authors did report comparisons of presurgical OPG and an imaging test (duplex, arteriography) to the ultimate "gold standard," the surgical specimen.(25) The accuracy of OPG was approximately 60 (severe stenosis on arteriogram = significant delay in eye-to-eye pulse wave forms and SOB pressure indices below .68 mm Hg).

If stroke or neurologic impairment of vascular origin are the critical clinical endpoints to be avoided in symptomatic patients, a highly sensitive noninvasive test that indicates reduced flow or mechanical obstruction in carotid vessels is of great value. As previously noted from Table 8, sensitivity and yield (the number of true positives found in the sample population) for OPG-K, OPG-G, and OPG-Z were 72 and 31, 81 and 45, and 74 and 35, respectively.

In reviewing additional clinical evidence to evaluate the utility of diagnostic tests in CAD, two controversies were identified with respect to imaging tests: the use of duplex imaging as a reference method and the most accurate method of estimating intraarterial stenosis.(77) Alexandrov et al,(25) in reaction to the belief that ultrasound methods have been inaccurate (when compared with cerebral angiography) in screening for carotid stenosis, prospectively analyzed 45 patients undergoing carotid endarterectomy from June 1992 to March 1993, using the North American Symptomatic Carotid Endarterectomy Trial (NASCET) and the European Carotid Surgery Trial (ECST) methods of measuring carotid stenosis and compared those results with direct visualization of the arteriogram ("eyeballing") and color duplex ultrasound. The reference method was the carotid artery plaque that was surgically removed during endarterectomy from 15 of the 45 patients. All patients had severe carotid stenosis and were symptomatic, having suffered either transient ischemic attacks or minor stroke (undefined in this population). Carotid angiography was performed by intraarterial digital subtraction technique (DSA) through the femoral artery, and biplanar images were obtained for all ICAs. Both the European and American carotid trials had previously developed linear methods to determine the degree of stenosis (1 - [the diameter of the residual lumen divided by an estimation of the "normal" vessel lumen] x 100). However, these authors recalculated carotid stenosis by angiography using the area luminal reduction, essentially the same formula but squaring the diameter measurements (1 - [d2/n2] x 100), where d equals the diameter of the residual lumen and n is the diameter of the normal vessel, and correlated the linear and area measurements. Peak systolic velocity was measured, and a conversion curve (linear to area stenosis) was used to calculate the degree of stenosis by duplex. The surgical plaque samples were preserved in formalin, photographed, decalified, and rephotographed, and planimetric tracings were taken of the narrowest lumen and the carotid bulb. Both area cross-sections were calculated by computer and percents of stenosis were derived (1 - [area of the residual volume divided by the area of the carotid bulb]). The mean percents of stenoses (± SD) by angiography were calculated: NASCET (linear) 63 ± 18, ECST (linear) 73 ± 16, eyeballing 76 ± 19, NASCET (area) 82 ± 17, ECST (area) 90 ± 12. These angiographic measurements were significantly different from duplex (P < .01) except for the NASCET area and eyeballing measurements. There was also a significant (P = .006) difference between NASCET and ECST measurements. For example, a patient estimated to have 55 stenosis calculated by the NASCET linear method would be classified as a 70 stenosis using the ECST linear, 78 by NASCET area, and 88 by ECST area measurements. Stenosis of 80 by DU correlated to an 80 stenosis by NASCET area. Having established these differences, they compared the planimetric measurements of the 15 surgical specimens to angiographic and ultrasonographic results and found that NASCET consistently underestimated ECST, and both underestimated the planar measurements of the surgical specimens, whereas DU and the two areal measurements were not significantly different from each other.

Barnett et al,(24) in discussing the preliminary results of NASCET involving patients with "less than very severe stenosis," stated that ultrasound studies were not used to determine the qualifying degree of stenosis in the NASCET (this was done through CA) and further stated "no results of surgical benefit in a series dependent on the use of ultrasound measurements have been published except as uncontrolled case studies. Therefore, their use must be regarded as a screening procedure with unproven usefulness in deciding on surgical benefit and, thus, in deciding on appropriateness for surgery. We regard their use . . . as a step along the way in determining the presence or absence of disease in individual patients who merit further study." McCann(78) has stated that "despite improvements in ultrasonography, [its] accuracy remains a question, particularly the quantification of degree of stenosis. Although few surgeons will operate on the basis of noninvasive examination alone, most require contrast angiography to facilitate operative planning."

The safety of this group of diagnostic instruments has not been of great concern, although there was some question that OPG examination may expose patients to ischemia or pain when pressures of 300 mm Hg or 500 mm Hg were applied to the globe. Gee and others(43,47) reported rare occurrences of temporary scleral ecchymoses that resolves spontaneously in several days. AbuRahma(38) reported that patients with certain eye diseases should not undergo OPG but did not present supporting documentation. There have been sporadic instances of transient blindness that lasted several seconds, but no subsequent or long-term adverse effects were reported.

Overall, the clinical studies using OPG are deficient in internal validity (patient selection, exclusion criteria, randomization, unmasked observers). Many authors redefined validity measures (sensitivity and specificity) and confused posttest probabilities (PPVs and NPVs) with them.

Plethysmography for Peripheral Vascular Disease

Noninvasive plethysmographic techniques such as IPG, SGP, PPG, and APG have been explored to aid in the diagnosis of both venous and arterial disease and to help reduce exposure to iatrogenic morbidity (pain, small calf thrombi, and common or external iliac vein thrombi) associated with CV and CA.(79)

Electrical plethysmography (impedance, strain gauge) is based on two concepts: 1) a change in voltage directly corresponds to a change in blood flow, and 2) the voltage change is proportional to a change in regional blood volume. Mechanical (air or water displacement) plethysmography measures the volume changes directly by reading the displacement volume of air or water. Mechanical plethysmographs have been historically cumbersome to work with and difficult to calibrate. Today, these chambered plethysmographs are used primarily in clinical research, although a recent air plethysmograph has been in use in medical practice. Electrical plethysmographic techniques measure changes indirectly induced by volume changes within a compartment or segment with a variety of different transducers: impedance (IPG), mercury or silastic strain gauge (SGP), photoelectric (PPG), or phleborheography (PRG).

It has been estimated that 800,000 cases of DVT occur annually and over 200,000 deaths per year are attributed to the sequelae of deep venous thrombosis (DVT),(80) estimated to occur in about 20 million Americans each year,(81) of which 66 (13 million) remain silent or present with nonspecific symptoms. Pulmonary embolism (PE), a potentially fatal disorder, may occur as a result of the development of DVT, primarily through proximal thrombi formation. Although its true incidence remains unknown, Sigel et al(82) calculated postoperative PE rates ranging from 5.8 to 11.8/1,000 in patients who had no preoperative evidence of venous occlusive disease by clinical presentation or Doppler ultrasound. Mortality rates vary according to detection and treatment; untreated, PE death rates have been estimated to be as high as 30, whereas detection and treatment reduces the estimated rate to about 8.(83) In a recent meta-analysis by Collins et al(84) of orthopedic, urologic, and general surgical patients, it was demonstrated that use of prophylactic subcutaneous heparin can prevent about 50 of all PE and about 60 of all DVTs.

The detection of DVT remains elusive;(85) the clinical signs and symptoms have widely variable accuracy and predictive value (Table 10).(86)

The true incidence and prevalence of DVT are unknown. The 1985 mean age-adjusted incidence rate of phlebitis or thrombophlebitis (ICD9CM code 451) in hospitalized patients in the United States was 79/100,000. The incidence rates for ages 15-44, 45-64, and 65 and older were 35, 143, and 289/100,000, respectively. (87) Factors other than age that have been associated with an increased risk of DVT are: race, surgical operations, pregnancy, vein trauma, prolonged immobility, malignant disease, varicose veins, obesity, and cardiac disease. (88)

All plethysmographic techniques are inaccurate in detecting DVT in those vessels in which the thrombogenic material has not significantly impeded venous outflow. (89) Although investigators have attempted to develop a predictive index for DVT risk,(90) an unacceptably high false-positive rate has precluded widespread use.

Originating most frequently within the deep venous system of the lower extremity in the soleal veins, thrombi usually form at the insertion of the valve cusps within the vein wall. Once propagated, successive layers of fibrinous materials collect and enlarge the clot and bind it to the vessel wall, leading to a decrease in the internal diameter of the vein. Small fragments may embolize proximally or lead to thrombosis and partially or completely obstruct the vein segment.

Clinically, acute DVT may manifest itself by a cascade of symptoms, ranging from mild pain, swelling, and increased skin warmth to redness, mottled cyanosis, cording, severe pain, arterial obstruction, and motor weakness. The presentation of symptoms depends on the severity of the outflow obstruction, competence of the venous valves, and the ability of the venous system to develop collateral venous pathways. If embolization does not occur, venous thrombosis can resolve either partially or completely by a combination of canalization, organization, and lysis.

Deep venous reflux is the cause of chronic venous insufficiency (CVI) in 80 to 90 of patients and results from postthrombotic damage, i.e., venous valvular incompetence, venous outflow obstruction, or a combination of both.(91-93) Chronic venous insufficiency, also known as postthrombotic syndrome, is associated with pain on prolonged standing, thigh pain after strenuous exercise, edema, hyperpigmentation, and ulceration. (94) It can be treated in its milder form with compression stockings and proper skin care, but those individuals with advanced disease may require surgical interventions such as venous transplantation or transposition and subfascial ligation. Those patients with primary vascular incompetence (nonthrombotic) may be candidates for valvuloplasty. The majority of noninvasive methods for evaluating CVI are aimed at assessing hemodynamic events, such as venous blood flow, pressure or volume, valvular closing times, and the presence of venous reflux in large vessels.

The main hydraulic force of the venous system of the leg is the deep calf muscles (calf pump), primarily the gastrocnemius muscle. As the calf muscle contracts, very high venous pressures (up to 100 mm Hg)(95) are generated, and blood is propelled rapidly. Concurrently, the unidirectional valves of the perforating veins (those veins connecting the deep and superficial venous systems) and the superficial veins close, allowing blood to flow proximally through the deep venous system without reverse flow or reflux into the superficial system. During relaxation, the deep veins dilate, causing negative pressure to draw blood from the superficial compartment through the perforating veins and main junctions into the deep system. During prolonged standing without muscle contraction, venous pressure builds because of the force of gravity. The points of connection between the deep and superficial systems (perforators) are the usual sites of leaks or reverse flow (reflux) that, when incompetent, transmit high pressure to superficial veins. This high-pressure reverse flow causes dilatation (varicosities) at the connecting points. The major sites where these varicosities occur are at the greater saphenous-femoral vein junction, lesser saphenous-popliteal junction, and multiple major perforating veins that are greater in number around and below the knee (50-100) than above the knee (10-20).(96)

The reference method in the evaluation of patients with suspected DVT is CV,(97) whereas AVP measurement (also known as phlebodynamometry) has been used as the reference method for assessing venous muscle pump function.(6-9) The value of AVP has been called into question because single-pressure measurements do not allow discriminant separation of diseased and nondiseased patients.(54)

Impedance Plethysmography

In IPG, a weak alternating current is passed through the leg. The frequency is so high as to be incapable of stimulating the heart and is imperceptible to the patient. Electrical current is held constant, and voltage changes are directly related to the resistance of the leg, based on Ohm's law (I = E/ R).(98,99) When a constant, weak, high-frequency current is applied to the outer pair of electrodes, voltage changes are measured by the second pair of circumferential electrodes. Because blood is a good electrical conductor, the change in resistance during venous occlusion is inversely proportional to the increasing volume of blood within the segment being tested.(98,100,101)

The impedance plethysmograph has several components: a thigh tourniquet, two pairs of circumferential calf electrodes placed about 10 cm apart -- one pair that discharges a weak, constant alternating current, and a second pair that measures the change in voltage over time -- a current source, and strip-chart recorder. Newer models have been adapted to be used in a personal computer environment.

The IPG technique is simple and straightforward to execute. After placement of the thigh cuff and electrodes, the limb is raised so that the calf is about 15-20 cm above the level of the heart, externally rotated, and rested on a pillow to maximize patient comfort and relaxation. This is necessary to minimize false-positive results due to extrinsic muscle tension or venous compression. The tourniquet and electrodes are connected to the IPG instrument and calibrated. The thigh tourniquet is inflated to a value lower than arterial diastolic pressure and higher than venous pressure, usually between 40 and 60 mm Hg, to occlude the deep venous circulation while allowing normal arterial inflow. The tourniquet inflation time is long enough for the maximum venous volume (VV) to collect in the segment, demonstrated on the strip-chart recorder as a plateau in the voltage. Usually, the occlusion time is about 45 seconds to 2 minutes, although the time to reach this plateau is less important than achieving complete venous filling. Once the plateau has been reached, the tourniquets are rapidly deflated and the change in voltage seen at 3 seconds postdeflation (VO3) is expressed as a function of the plateau filling voltage (VF). In subjects without venous obstruction, the voltage drops quickly as the limb segment empties and returns to its "preinflation" resting volume. In subjects with a thrombosis or other obstruction to flow, there is a delay in venous emptying of the segment. Hull et al(101) examined the relationship between repeated venous filling and emptying cycles and measured the sensitivity and specificity of IPG in suspected acute proximal DVT. Their successful method of repeated exams (five exams lasting 45-120 seconds each) increased its accuracy and has become the standard method of testing. Individual test results are plotted on a predetermined discriminant line graph, a "stop line," where the VF is plotted against the VO3. When any test result falls above the stop line, the test is stopped and the result is consiered normal. When any result falls below the stop line, the test is repeated and the recording that demonstrates the greatest VO3/VF ratio is used. Testing can and should be repeated up to five times to minimize false-positive results and to allow for maximum improvement in venous capacitance.

Table 11 lists the clinical studies that measured the accuracy of IPG in diagnosing DVT. The accuracy of IPG in the clinical studies reviewed ranged from 78 to 90. Impedance plethysmography is very specific in the case of serially negative IPG results, serving as the basis for withholding anticoagulant treatment.(102)

The sensitivity of IPG is markedly reduced in distal disease.(99) There has been much debate about the value or necessity of treating distal disease. Although some clinicians believe the detection of a calf thrombus is an early manifestation of iliofemoral thrombophlebitis, others consider calf thrombus a separate disease and not likely to contribute to more proximal venous disease or pulmonary embolus.(110)

Impedance plethysmography has been shown to be less sensitive in the presence of recurrent disease. Patients with chronic or recurrent DVT pose additional diagnostic challenges because of persistent venous outflow obstruction or recanalization. Hull et al(111) studied 270 patients with clinically suspected acute, recurrent disease. Based on serial IPG and fibrinogen leg scans alone, long-term outcome was significantly different in those who initially tested positive on IPG or leg scan (89 patients) from the 181 patients who tested negative. Two percent (3/181) of patients who had anticoagulants withheld because of negative noninvasive tests developed a recurrence and died, whereas 18 of 89 patients (20) who tested positive and were treated with anticoagulants had a recurrence. Four of those 18 died of massive pulmonary emboli. Based on these observations and long-term followup, Hull et al recommend withholding anticoagulant therapy in those patients who test negative on IPG and leg scan.

Impedance plethysmography has been used as a diagnostic tool in at-risk hospital-based populations,(98) outpatient community settings,(112) and in symptomatic pregnant women.(113) The clinical studies have demonstrated very high sensitivity and specificity in acute disease of the popliteal area. Many false-positive results were due to inadequate relaxation of the leg or improper positioning during testing. Other more serious comorbid conditions such as severe arterial insufficiency, systemic shock, respiratory failure requiring positive pressure ventilation, venousoccluding tumors of the pelvis, and ascitic compression have also been associated with false-positive results presumably due to compromised venous outflow.(86) Impedance plethysmography should not be administered to patients who have severe neuropathies, burns, leg casts, or any condition that impedes relaxation of the extremity.

Strain Gauge Plethysmography

Strain gauge plethysmography has been used to evaluate the degree of compromise in the venous system by noninvasively quantifying retrograde outflow in the deep venous system through incompetent venous valves. It has also been used as an adjunct in the diagnosis of peripheral arterial disease (PAD). The SGP transducer consists of a flexible tube filled with a conductive medium (mercury, gallium) that fits snugly around an extremity and is connected to a source of electricity. Changes in the circumference alter resistance in the column and generate an analog signal.

Strain gauge plethysmography was originally introduced into clinical practice by Whitney in the early 1950s.(114) It has been used in the past to establish the patency of deep veins and to determine the competence of deep and superficial venous valves.(115) In those cases where treatment for acute DVT is delayed or absent, disease progression may lead to chronic edema, deep vein and/or perforator vein incompetence, ambulatory venous hypertension, and a cascade of secondary skin changes to the lower limbs, as well as persistent occlusion or recanalization.(116) Development of CVI, a postphlebitic syndrome, depends on the degree to which residual occlusion decreases the efficiency of venous emptying and on the severity of the damage to venous valves.

Using SGP to study venous disease, the practitioner positions the patient so that the legs are elevated above the level of the right atrium without undue compression of the calves. A transducer containing a mercury (or alloy of gallium and indium)(14) column in silastic tubing is placed around the thickest part of the calf, while a pneumatic cuff is applied to the thigh and inflated to a subdiastolic pressure by a calibrated pressure-regulated air source.(117) Both the cuff and transducer are connected to the plethysmograph and recorder. The electrical resistance in the mercury column changes as the silastic tube expands or contracts with the changing blood volume of the limb. Two assumptions exist with use of this device: 1) flow is linearly related to the change in circumference of the limb, and 2) volume expansion is equal throughout the segment and indicative of that throughout the limb.(80) Therefore, the tissue volume calculated is truly segmental and restricted to the tissue under the gauge.

Test results are used to approximate the following parameters: maximum venous outflow (MVO), venous volume (VV), maximum venous incremental volume (MVIV), maximum venous reflux volume (MVRV), and maximum venous reflux flow (MVRF). Maximum venous outflow reflects the functional capacity of the venous system to drain blood collected in the lower limb during periods of venous occlusion. It is calculated from the slope of the emptying curve during the first second after the release of the thigh cuff:

MVO (mL/min/100 mL tissue) = - (12,000/C)(dc/dt),

where C is the calf circumference in centimeters and dc/dt is the rate of change of the calf circumference in centimeters per second. If MVO is abnormal, a pneumatic cuff is placed below the knee to mimic the normal response of venous valves during exercise or emptying and the test is repeated. If MVO is then within the normal range, superficial venous incompetence is suspected. If the MVO remains abnormal, deep venous insufficiency is presumed.

The values for MVO in acute thrombophlebitis by SGP have been shown to be substantially reduced when compared with voluntary controls.(118) The measurement of MVO has not been demonstrated to be of any diagnostic utility when postphlebitic patients were compared with control patients or postsurgical patients.(7) In the absence of venous occlusion, MVO values are often indistinguishable in patients with normal limbs, those with superficial venous incompetence with competent deep venous valves, and those with deep venous incompetence.(115)

Venous volume (VV), the maximum increase in the calf volume, is calculated from the recording of the calf volume during occlusion and expressed in mL/100 mL of tissue. The ambulatory volume change (AVC) is the maximum volume change during exercise, also expressed in mL/100 mL tissue. The AVC, measured over the period of exercise, typically about 30 seconds, yields less discriminant information than the VV and makes little distinction between limbs with or without superficial venous occlusion, but can discriminate between normal limbs and those with deep venous occlusion.

The calculation of MVRF and maximum venous reflux volume (MVRV) have yielded substantially different results when patients with chronic phlebitis were compared with patients with normal limbs.(117) In this technique, a proximal thigh tourniquet is inflated above the arterial systolic pressure (300 mm Hg) while the distal cuff is rapidly inflated to 50 mm Hg. The underlying thigh venous blood is then directed distally at a rate and magnitude proportional to the degree of valvular incompetence of the veins. Maximum venous reflux flow is calculated as:

MVRF (mL/min/100 mL tissue) = (12,000/C)(dc/dt),

where C is the calf circumference, in centimeters and dc/dt is the rate of increase of the calf circumference, in centimeters per second. Maximum venous reflux volume is calculated as:

MVRV (mL/100 mL) = (200 Delta C)/C,

where Delta C is the change in calf circumference in centimeters, at 10 seconds, and C is the calf circumference in centimeters at time zero.

Strain gauge plethysmography has not demonstrated reliably accurate results in the diagnosis of DVT or CVI. There has been one published report(119) that compared SGP with the reference method (CV) upon which this conclusion is based. The review of other available data (see below) involved published studies comparing SGP with clinical symptoms or other noninvasive techniques.

AbuRahma et al(119) studied 752 patients referred to their vascular lab for evaluation of suspected DVT. Venous duplex imaging (real-time B-mode ultrasound with Doppler) was performed on 713 patients, and 743 patients underwent SPG, whereas 602 patients underwent both noninvasive tests. However, only 68 (9) underwent ascending venography within 48 hours of the noninvasive testing and serve as the basis for accuracy calculations. Examiners were blinded to venography results. The criteria specified for an abnormal duplex scan were visualization of acute thrombus with noncompressiblity of the vein or absence of flow on Doppler examination of the noncompressible vein. Three criteria were used to differentiate acute thrombi from chronic thrombi: echogenicity (older thrombi are more echogenic), texture (mature thrombi have a more heterogenous appearance), and the presence of collaterals (which suggests a chronic process). The SGP measurements (not stated directly, but incorporated by reference) were also performed, but no attempt was made to determine the age of the thrombi with this method. The reference method for patient classification was ascending venography, as follows: acute DVT was diagnosed if obvious intraluminal thrombus was observed, if deep veins were not visualized in the presence of adequate venous filling, or if eccentric luminal filling defects were seen in more than one projection. Chronic DVT was determined by irregular, narrow, tortuous (recanalized) vessels, corkscrew collaterals around an area of nonvisualization, or incompetent perforators. The results for SGP were: sensitivity = 82, specificity = 70, PPV = 56, NPV = 89, accuracy = 74. Duplex imaging was more reliable in all measures except NPV: sensitivity = 94, specificity = 80, PPV = 82, NPV = 84, accuracy = 87. The false-positive rate for duplex imaging was 10 and for SGP was 21. In reviewing misclassification by SGP, calf thrombi were missed by SGP in 75 of the cases (three of four false-negative results occurred with nonocclusive thrombi and thrombosis of the calf vein with extensive collateral circulation). Of the 14 false-positive results, concurrent, contributory conditions included pregnancy (two), leg edema (two), paresis of the extremity (two), congestive heart failure (two), chronic DVT with inadequate collaterals (three), severe peripheral vascular disease (two), and one case of extrinsic iliac vein compression. In addition to the data presented on their own patients, these authors reported the accuracy of duplex imaging in acute lower extremity DVT from 10 earlier studies involving a total of 1,365 limbs. Sensitivity ranged from 89-100, and specificity from 78-100. Overall accuracy ranged from 89-100. This study was the only recent report available that measured SGP against venography and duplex imaging in the diagnosis of venous thrombotic disease. The authors differentiated acute and chronic disease in a post-hoc analysis; this is of limited practical value to clinicians who may be unable to determine the likelihood of an acute or chronic process in their patients before testing.

Fernandes et al(115) measured SGP, Doppler ultrasound, ascending venography, and AVPs in patients to determine whether SGP could distinguish among normal limbs, those with superficial valvular incompetence only, and those with deep valvular incompetence with and without occlusion.(115) They also wanted to validate quantitative measures of venous incompetence by measuring venous pressure. They demonstrated a linear relationship between ambulatory volume and pressure measurements, even in the presence of deep venous insufficiency and occlusion.(115) However, this report was methodologically flawed. Patients were studied with DU or CV (with AVP and plethysmography). The results were not reported in a way to directly compare plethysmography with AVP or CV.

Cramer,(120) in an attempt to diminish the variance in SGP test interpretations, developed the venous outflow/capacitance ratio as the diagnostic criterion. (121) The overall variability of SGP has been reported to be between 6 and 28.(122-124)

Struckmann(7) prospectively evaluated long-term outcomes (reflux, venous emptying) in surgical patients with lower extremity varicosities. Despite the author's claim that "strain gauge plethysmography may serve as a method to control the effect of different treatments in patients with venous valvular incompetence," SPG was not assessed against venography. Comparison of SPG results preoperatively and at 3 months demonstrated significant reduction in mean reflux, from 30 seconds to 45 seconds (P < .001), although degradation of that response was noted at 60 months' followup (42 seconds).

In a 1992 study, Struckmann et al(9) attempted to examine blood volume changes measured by ambulatory blood volume scintimetry and SGP in six patients with CVI and eight controls (26 limbs). Strain gauge plethysmography measurements (previously described) included venous return time (VRT) and ejection volume (EV). Radioactive labeling of red blood cells was performed in vivo by injecting intravenous stannous fluoride followed by injection of technetium-99m. After calibration, EV and VRT were measured using both methods. Venous return time measured by scintimetry at both the ankle and calf closely approximated SGP: 52, 50, and 58 seconds, respectively, in the controls; and 17, 10, and 15 seconds, respectively, in the CVI patients. Ejection volume, however, showed more variability. Scintimetry at the ankle and calf measured 33 mL/100 mL and 23 mL/100 mL, respectively, whereas SGP recorded an EV of 1.9 mL/ 100 mL. The measurements in patients with CVI showed the same variation: the ankle and calf measurements were 19 mL/100 mL and 20 mL/100 mL, whereas EV measured by SGP was .8 mL/100 mL. The differences in the SGP measurements vs. the scintimetry measurements are thought to reflect the fact that SGP measures total leg volume, whereas scintimetry measures actual blood displacement. The authors concluded that scintimetry does not give selective information about deep venous reflux or volume displacement but may be of value as a research tool to study segmental participation of venous pump activity.

Arterial flow by SGP is measured by placing a large pneumatic cuff on the thigh proximal to the strain gauge and inflating the cuff to a pressure that prevents venous outflow but permits arterial inflow.(14) The parameter measured is the slope of the initial increase in volume of the limb. In a resting limb (nondynamic studies), SGP is known to be insensitive to all but extremely occlusive disease. Investigators have studied the pulse reappearance time in the toe using a strain gauge transducer following proximal occlusion as an index of severity of PAD.(14)

Steer et al(125) evaluated symptomatic patients with PAD by SGP and angiography and compared those results with the patients' symptomatology. In the first part of the study (preoperative comparison of SGP and clinical symptoms), 73 limbs were studied by SGP. Mean rest flow was higher in limbs with distal occlusive disease than in normal limbs or those with proximal disease (9.4, 6.7, and 5.1, respectively). Peak flow was highest in nondiseased limbs and lower but not significantly different in limbs with distal or proximal arterial disease. Velocity studies (time to peak flow) were significantly different in nondiseased limbs compared with limbs with femoropopliteal disease but not with those with aortoiliac disease. Using uniplanar anteroposterior films, 46 limbs underwent angiography. The authors reported an unstated number of limbs with both aortoiliac and femoropopliteal artery occlusive disease who were subsequently eliminated from this second part of the study. Twenty-six limbs with isolated femoropopliteal disease and 11 limbs with aortoiliac disease only and 9 limbs without significant radiologic disease (<50 vessel lumen narrowing on uniplanar films) underwent SGP. The authors found a significant difference in rest flow between the group of limbs with distal disease and the group of limbs with proximal disease (P < .05) but no difference between either of these groups and the nondiseased group. Thirty-one limbs were used for post-hoc analysis of preoperative plethysmography after reconstructive surgery. Only twenty-six limbs underwent SGP after surgery, of which 23 were also studied preoperatively. By stratifying limbs according to proximal or distal disease, the authors were able to show a modest increase in resting arterial flow in diseased limbs. They further claimed that SGP can differentiate limbs with PAD from normals, particularly based on velocity measures (P < .05). However, methodologic problems with this study should make any conclusions suspect. For example, patient inclusion and exclusion crieria were not reported. Although the investigators did have angiographic information available on all patients (except the 21 voluntary controls), they chose not to report any measures of validity (sensitivity, specificity, or predictive values). In reporting their postoperative results, where a comparison of pre- and postoperative plethysmographic results would have been helpful, particularly resting and peak flows, they report the correlation between the postoperative pressure index (systolic ankle pressure by Doppler divided by brachial artery systolic pressure) and postoperative peak flow as significant (P < .01). Comparing diagnostic techniques (Doppler and plethysmography) without a comparison to the available reference method, particularly when that information was collected, is not an acceptable analysis of a diagnostic technology.

Strain gauge plethysmography does not provide reliable quantitative data in partially occluded vessels in which flow is marginally reduced. It does provide an indication of the qualitative function of the extremity in venous insufficiency.

Despite the lack of validation studies, many clinicians continue to use SGP to measure segmental venous and arterial pressures in the lower extremity to determine venous flow measurements and the approximate level of arterial occlusion.

Photoelectric Plethysmography

Photoplethysmography has been applied by medical and surgical practitioners since 1937 to count the heart rate, evaluate peripheral arterial pressure, oxygen saturation, and peripheral microcirculation after skin grafting, drug ingestion, burns, or revascularization. It has also been used to evaluate peripheral CVI by indirectly calculating changes in skin blood volume during static and postexercise states.(91)

Photoplethysmography is not strictly a plethysmographic technique.(15) Rather, its operation is based on the principles of light densitometry and photon diffusion theory.(127) Currently, two configurations of the PPG light sensor are possible; one, in which the tissue to be evaluated, i. e., finger or toe, is placed between a photoelectric cell (receptor) and the LED, or the second design, in which the LED source is placed side-by-side with the photoelectric cell, e.g., 10 cm above the medial malleolus, on the surface of the tissue of interest. Other components of the system include an amplifier and oscilloscope, calibrated recording device, and electrocardiogram and may include digital readouts and timed and dated recording equipment. Photoplethysmography is used in the alternating current (AC) mode for arterial pulse detection and in the direct current (DC) mode for venous evaluations.(127)

The light source is near infrared to minimize interference from other light sources. The receptor is a phototransistor with a light response over a wide range of intensities. Light absorption by hemoglobin, the principle chromopore in the skin, is largely dependent on the volume of blood in the superficial dermal venous plexus (1.5-2.0 mm).(54)

The reflected light technique uses three infrared LEDs that emit a constant light level in the 600- to 700-nm wavelength range. The photoresistor (transducer) detects the degree of attenuation of backscatter from the superficial layers of skin (1.5-2.0 mm).(128) The amount of reflected light varies with the number of red blood cells in the cutaneous microcirculation. Slight dilatation and contraction of arterioles and capillaries during each cardiac cycle attenuate the light reflection. The transducer is connected to a recorder which shows the relative change in skin blood content over time, varying in response to the cardiac cycle or active or passive calf muscle exercise, depending on the PPG application. The three main applications for PPG are: pulse waveform analysis, segmental arterial systolic pressures, and pulse volume changes secondary to compression maneuvers.(129) Pulse waveform analysis allows qualitative evaluation of the systolic upstroke, peak pressure, dicrotic notch (aortic valve closure), and downslope. When the PPG sensor is applied distal to an arterial obstruction, the slope of the upstroke is delayed, the peak is rounded, and the dicrotic notch may be absent (damped). In arterial vasospastic disease, the peak may show abnormally high amplitude and an elevation of the dicrotic notch on the downslope. With the use of very small pneumatic cuffs, PPG has been used to measure systolic digit pressure in patients with diabetes or thromboangiitis obliterans for preamputation evaluation. Pressures are evaluated on an absolute basis (digit pressures <50 mm Hg are unlikely to support healing) and on a comparative basis with wrist or ankle pressures. Compression tests, such as the Allen test of the continuity of the palmar arch, are used to evaluate the contribution of each artery (ulnar and radius) before indwelling arterial catheters are inserted for continuous monitoring. They can be recorded with a PPG sensor-attached distal to the compression site and sense sequential pulsatile flow during the test. Oter compression applications include the evaluation of thoracic outlet syndrome(130) and SPPG, as previously described.

Photoplethysmography also permits assessment of altered cutaneous hemodynamics in patients with postphlebitic syndrome. Response abnormalities during exercise are assumed to be related to the presence and severity of postphlebitic dermatitis and stasis ulceration. When incompetent perforators are present, the PPG usually remains abnormal in patients with previous DVT. It is possible to have venous valvular incompetence demonstrated by Doppler and have a normal PPG. In CVI, recovery is markedly decreased. Qualitative evaluation with DU remains the simplest and most rapid method to detect venous reflux.(131)

Photoplethysmography has also been used as an adjunct in the diagnosis of arterial disease. Ando et al(132) used PPG to demonstrate changes in the arterial pressure/volume relationship (arterial elasticity) in healthy volunteers and compared them with patients with demonstrated coronary atherosclerosis. They reported an association between a reduction in arterial elasticity and atherosclerosis, although their conclu-sions were somewhat obscured by the interaction of age and atherosclerosis.

Photoplethysmography may also be used as a pulse detector, particularly to measure ankle systolic blood pressure in suspected arterial disease. It has been postulated that the resting ankle pressure index (ankle pressure divided by brachial pressure) is the best indicator of arterial occlusive disease of the lower extremity.(129) It has also been used in conjunction with thigh occlusion cuffs to evaluate peripheral arterial flow (toe pulse) during treadmill exercises and reactive hyperemia. (133,134)

Light-reflection rheography, also known as quantitative PPG, developed by Wienert and Blazek in West Germany,(135) is a PPG method used to detect changes in blood volume. Both PPG and LRR have been used in the detection and diagnosis of varicose veins, venous insufficiency, valvular insufficiency, phlebothrombosis, postthrombotic syndrome, and venous decompensation during pregnancy.

The major difference between LRR and PPG is in the construction of the measuring probe. The sensor (probe) is composed of three gallium-arsenide LEDs with integrated lenses to narrow the output, a detector (silicon phototransmitter) to measure the backscatter of infrared light, a thermocouple to measure skin temperature, and a strip-chart recorder. The light is absorbed by hemoglobin, scattered or reflected; thus, the integrated lenses are necessary to increase detector efficiency. Other factors that influence light reflection are the density of the skin and the volume of blood in the skin capillaries. The amount of reflected light is inversely proportional to the blood volume and indirectly quantifies changes in the dermal venous plexus so that the intensity of the reflective signal is reduced in the presence of increased blood content. Venous patency and valvular competence are tested under active conditions of exercise. The test is alleged to be valid only when skin temperature is between 25 and 33 degrees C. The addition of a Doppler probe enables arterial blood flow evaluation.

In evaluating the upper extremity venous circulation, three measurements are made: 1) comparative changes in blood flow and venous recovery time after timed, repetitive flexion/extension exercises in both limbs; 2) comparative changes in blood flow during compression of each limb with a blood pressure cuff (60 mm Hg for 2 minutes) to mimic venous emptying; and 3) comparative changes in respiratory variations in the supine position.

Lower extremity dynamic venous function (venous emptying) is evaluated with LRR by placing the patient in the sitting position (and subsequently in the supine and reverse Trendelenberg positions) with knees and hips flexed and the sensor placed 2-3 inches above the medial malleolus of the leg to be tested.(137) The light reflectance is measured at baseline (R₀). Ten dorsiflexions of the ankle are performed. As venous emptying occurs and venous pressure decreases, the volume of erythrocytes decreases with a resultant increase in light reflectance. The end of the dorsiflexion exercises (peak venous emptying) corresponds to maximum reflection (R_max). The measured change in light reflection at R_max is called Delta R (R_max-R₀), and the rate of change (slope) of light reflection is (R_max-R₀/time) or Delta R/time.

Delta R = R_max - R₀

The exercise period is followed by one of motionless venous recovery (venous refill time, VRT) with blood reentering the extremity and subsequent diminution of light reflectance until a new baseline is reached (R₁). If R₁ is >20 of baseline R₀, these exercises are repeated several times in each limb. Three parameters measured during the test are: 1) the difference in light reflectance at the end of venous emptying, compared with the initial baseline (R_max-R₀); 2) the rate of change in reflectance (R_max-R₀/time), which is dependent on the rate of dorsiflexion by each patient; and 3) VRT, the time required to return to a new baseline (R₁) reflectance (R₁-R_max/time).(138,139)

Differentiation between superficial and deep venous insufficiency is achieved through the use of leg tourniquets. Incremental insufflation of the tourniquets can estimate the level of superficial venous insufficiency. When venous refill remains slow in the presence of complete insufflation of the occluding tourniquet, deep venous insufficiency (DVI) is suspected. The following clinical studies were performed to validate the accuracy of LRR (Table 12).

In a retrospective review by McEnroe et al,(143) the records of 386 consecutive patients (720 limbs) with CVI were reviewed to calculate: 1) the proportion of patients with DVI and superficial venous insufficiency who present with CVI; 2) the proportion of patients with advanced CVI (stage III) in this population and the prevalence of DVI in this subset; 3) the prevalence of obstructive vs. valvular incompetence in patients with advanced CVI; and 4) how the noninvasive assessment of CVI by LRR correlated with the clinical severity of CVI. All LRR measurements were taken with patients in the sitting position. When the initial VRT was abnormal (<25 seconds), sequential tourniquets were employed to further differentiate the involvement of the deep and superficial venous systems. There were no other additional confirmatory tests, i.e., volumetry, venography or ultrasound, undertaken in this population.

Based on LRR results alone, there were 346 limbs (48) with abnormal VRT (<25 seconds) and 375 limbs (52) with normal VRT. Of the 48 of limbs with abnormal VRT, 65 of these (225 limbs) had deep venous insufficiency, 21 (72 limbs) had superficial insufficiency, and 14 (49 limbs) had a combination of both. The predominant cause of DVI was valvular incompetence (95), whereas deep venous obstruction was observed in 5 of limbs. Advanced CVI was diagnosed by clinical symptomatology in 142 limbs (41), another 40 had stage I or II, and 20 were asymptomatic. Regardless of the level of venous disease (deep or superficial), patients with ulcers had VRTs between 0 and 24 seconds with the majority between 5 and 14 seconds. Patients with less severe symptoms (swelling or varicosity alone) had VRT from 0 to 24 seconds, the majority between 15 and 24 seconds. The authors state that the identification of those patients with pure superficial venous insufficiency in stage III is necessary because, rather than benefiting from deep venous reconstruction, they are candidates for simple saphenectomy, perforating vein ligation, or both.

In a recent study by the same investigative group, Welch et al,(93) compared LRR, DU, and APG with descending CV in 4 volunteers and 17 patients with mild (8) or severe (9) venous reflux. The volunteers did not undergo ascending or descending venography. Light-reflection rheography enabled differentiation between patients with normal limbs from those with venous reflux but was unable to distinguish the severity of the reflux. To make that distinction, the group developed receiver operating curves (ROCs) for the noninvasive APG tests (ejection fraction [EF], VFI, residual volume fraction) and the DU tests. The latter comprised valve closure time assessed in the superficial femoral vein and the popliteal vein; those two values together represented the total valve closure time (TVCT). They found that TVCT, reflecting femoropopliteal venous segment function, was the most discriminant of the noninvasive tests able to separate mild and severe venous reflux. They concluded that TVCT by DU and APG gave the best noninvasive assessment of the severity of venous reflux in patients with severe CVI who may be candidates for venous reconstructive procedures. They further concluded that patients with a TVCT >4 seconds should be referred for contrast venography.

In a study by Mitrani et al,(137) 69 patients with clinical symptoms suggestive of DVT were referred for study (patient selection criteria unstated). Twenty-two patients (32) had proximal DVT, two patients (3) had isolated calf (distal) DVT, and 45 patients (65) had no evidence of DVT, as demonstrated by venography. Light-reflection rheography was performed on patients in the sitting position within 48 hours of CV. These investigators constructed ROC analyses to maximize the sensitivity and specificity of LRR. When using a slope of <.17 mm/sec as an indication of DVT, sensitivity was 83 and specificity was 89. When the definition of a positive test was broadened to a slope of <= .31 mm/sec, sensitivity was 96 with a corresponding decrease in specificity to 78. Also, when R_max-R₀ values of 6 mm or less vs. 3 mm or less were used for a positive test, sensitivity decreased from 96 to 83 and specificity increased from 71 to 89. These authors recommend using LRR as an "excellent noninvasive alternative in screening patients" for occlusive DVT.

Arora et al(141) attempted to determine the accuracy of LRR in 61 patients who were referred to them for CV for suspected DVT (by clinical examination). The authors state that DU is now the accepted reference method for diagnosing DVT, but DU is not widely used because of its expense, required operator expertise, and the fact that it is time consuming to perform. It is interesting to note that patients referred for CV were included in the study, but those patients referred for DU were not. The authors do not explain this difference. We could not determine whether this study was prospective or not. Contrast venography was used as the reference method in this study. Examiners were masked to the results of both LRR and CV until both were completed (within 24 hours of each other). A rate of venous emptying <=.35 mm/sec on the LRR curve was considered positive for DVT. There was a 10 false-positive rate and a <2 false-negative rate (in a patient with proximal DVT).

A prospective study by Thomas et al(142) enrolled 124 consecutive patients with variable symptomatology who were considered at risk for DVT. Light-reflection rheography was performed on the same day as the contrast venogram, and the examiners (radiologists, vascular technicians) reported results without referral to the clinical notes or other diagnostic examinations performed. These investigators performed two tests on each limb with the method described previously, and, where superficial venous abnormalities (varicose veins) were present or where test results were equivocal, LRR was repeated "at higher sensitivity" with the use of thigh tourniquets. Of the 124 patients, venography was not performed in four, the records and venogram were lost in one, and 12 patients had bilateral venograms, so the study population was 131 limbs. It is interesting to note that the authors reported the accuracy of CV as 47 (the reference used was clinical symptomatology), when in fact they were actually reporting the prevalence of DVT in the limbs that they studied (number of positive venograms out of the total number of limbs tested, 61/131). The accuracy of LRR, which was the index test under study, was 88. The authors noted that the sensitivity of LRR was no different among calf vein thrombosis, isolated thrombosis of the proximal veins, or both. Light-reflection rheography testing in this population elicited an 8 false-positive rate (11/131) and a 4 false-negative rate (5/131). The false-positive results were in patients who were reported to be older and had concurrent medical conditions which may have interfered with the LRR performance, e.g., congestive heart failure with leg edema, four patients who were unable to dorsiflex their ankles, and one patient with a previous history of DVT. The five patients who did have thrombosis not detected by LRR (false negatives) were reviewed; one patient was determined to have been misclassified (really a true positive), two had been started on heparin before LRR testing, and the remaining two had partially obstructing thrombi that were missed by LRR.

Kuhlmann et al(140) prospectively enrolled 75 emergency room patients who presented with symptoms suggestive of DVT in a study to validate LRR in the diagnosis of acute DVT. Unfortunately, they used either of two tests (DU or CV) as the reference method to determine the accuracy of LRR. Only 16 out of the 75 patients had CV, DU, and LRR. In these patients, CV and DU were in agreement in 14/16 patients, but the authors neglected to report the results of the LRR in these 16 patients specifically. Eighteen patients had positive CV or DU for DVT (24 prevalence), and 57 patients had no evidence of DVT on either. Light-reflection rheography was positive in 17/18 patients (95 sensitivity) using either CV or DU as the reference method. Despite this curious methodologic design, the PPV was 49 because there were 18 false positives (24) reported with LRR. The authors did not discuss the false-positive rate. They further stated that "an abnormal LRR mandates further testing; a normal LRR may not eliminate the need for further evaluation, and in our study there was one false-negative LRR."

Recently, Somjen et al(139) used LRR to differentiate varying reflux patterns in 54 patients with varicose veins. After duplex (Doppler and B-mode ultrasound) scan with manual compression of the calf, patients were divided into three groups according to the severity of their venous reflux: 1) short saphenous vein, 2) popliteal vein with distal popliteal competence, and 3) superficial femoral vein with distal popliteal competence. Light-reflection rheography was then performed to determine the VRT, with and without thigh and leg tourniquets. The results showed that patients in group 1 (19 legs: 10 asymptomatic, 9 symptomatic with aching or discomfort) had a mean VRT of 22.7 seconds (± 12.9), whereas group 2 (22 legs: 8 asymptomatic, 14 with varying degrees of aching, leg swelling, hyperpigmentation, ulceration) had a VRT of 12.7 seconds (± 7.1) and group 3 (13 legs: 2 asymptomatic, 11 with varying degrees of severe edema, hyperpigmentation, and ulceration) had a mean VRT of 12.3 seconds (6.6 SD). No ranges were given on the VRT in any of the groups. However, the authors report that very rapid VRT (less than 10 seconds) was noted in 3 legs in group 1, 12 legs in group 2, and 7 legs in group 3. When tourniquets were applied below the knee and the patients were retested with LRR, 10 out of 19 legs in group 1 showed "significant improvement" (not quantified), indicating short saphenous vein incompetence; group 2 demonstrated in all but one leg VRT >25 seconds (a normal response in that laboratory); and, in group 3, 10 legs "improved" and 1 leg displayed normal VRT; the results in the remaining 2 legs were unreported. Prolonged VRT was seen during compression with a thigh tourniquet in 6 of 54 legs. When a below-the-knee tourniquet was applied in these patients, improvement in VRT was seen, indicating a contribution to the reflux from the long saphenous vein. In 7 of the 54 legs, no improvement was seen with tourniquet applications, and the authors speculated that it could have been due to technical difficulties or reflux in the perforator veins. The authors conclude that LRR provided useful information (reduced VRT) to the duplex scan information in assessing venous reflux in the popliteal fossa.

Neumann and Boersma(144) believe that LRR does not measure blood volume but changes in the reflection of the skin. They state further that a noninvasive method, such as LRR, can never give an exact value of the fall in venous pressure, because qualitative volume measures are made with LRR and, in the venous system, no linear relationship exists between pressure and volume; they conclude that this method is not suitable for pressure measurements. Further, positional changes in the patient during testing (sitting or standing) make precise pressure calibrations (relative to the right atrium) difficult. Because of the placement of the sensor (probe) medially, insufficiency of the long saphenous vein is easier to detect than incompetence or insufficiency in the short saphenous vein, and must always be verified with another method such as Doppler ultrasound.

In the study by Mitrani et al,(137) the use of LRR as a PPG method to detect delayed venous filling had a sensitivity of 95 and a specificity of 77 using an LRR slope of .31 mm/sec. The 30 false-positive rate in patients with coexisting conditions such as knee abscess, effusion, pregnancy, pelvic mass, and phlebitis leads to cautious recommendation for its use in these patients. Arora et al(141) defined a positive LRR as any test with a slope <= .35 mm/sec. They were able to demonstrate high sensitivity (96), a higher specificity than the Mitrani group (83, 77, respectively), and a lower false-positive rate (10).

The mean (nonweighed) sensitivity, specificity, and predictive values for LRR in the four studies that used CV as a reference method were .955, .823, .755, and .943, respectively. Mean accuracy was 85 and involved 225 patients. Light-reflection rheography is a safe, noninvasive, semiquantitative test that reliably diagnoses venous insufficiency in the absence of comorbid conditions.

There does not appear to be any difference in clinical utility between PPG and LRR. They both are semiquantitative methods that have been used to study CVI and successfully accomplish that with reasonable accuracy.

Water Plethysmography

Water plethysmography (WPG), also known as foot volumetry, is one of the oldest plethysmographic methods in use. A limb is placed through a self-sealing chamber, so that a segment from below the elbow or knee and above the wrist or ankle is enclosed. Body-temperature water (37 degrees C) is then inserted into the chamber. An occlusion cuff is placed on the limb above the WPG chamber and inflated to a subdiastolic pressure (approximately 60 mm Hg) for about 5 seconds and then released. The changes in blood volume increase the diameter and length of the limb and displace water into a measuring cylinder. Repeated measures at longer occlusion times are performed. When the limb is removed from the plethysmograph, the segment under study is measured for changes in circumference and length, so that volume changes can be assessed. Absolute volume measurement cannot be evaluated, but relative, semiquantitative changes can be inferred.

There have been no studies published to date that compare WPG with a reference method. Bradbury et al(145) prospectively studied 43 patients with foot volumetry and DU before and after superficial and perforating vein ligations. Nine patients developed recurrent venous ulcerative disease postoperatively (six with femoral and three with popliteal vein incompetence), and the remaining 34 patients did not develop any recurrent venous ulcers during the period of followup (range, 18-144 months). In the nine patients that developed recurrence, the mean ejection (expulsion) fraction (EF) was .8 (range, .6 to 2.3), whereas in the 34 patients who remained ulcer-free, the EF was 1.5 (range, .4-2.9). This finding was not significant (P = .25). The median half-refill time in the recurrent ulcer group was 1.5 seconds (range, .5-5.5 seconds) and 5.0 seconds (range, .5 - 23.0 seconds) in the ulcer-free group (P = .01). These investigators believe that foot volumetry gives useful quantitative information to evaluate postoperative patients.

In a later report, the same authors(146) measured foot volumetry preoperatively, immediately after surgery, and at regular intervals during followup (median, 60 months; range, 3-144 months) in the 43 patients previously described plus an additional 10 patients with severe CVI (recurrent leg ulcers). Foot volumetry studies included total foot volume and expelled volume after exercise (20 knee bends). Expulsion fraction (EF) was defined as expelled volume/total foot volume (), and half-refilling time (HRT) as the time taken for the foot to return halfway to its preexercise volume. A below-the-knee narrow-gauge tourniquet was applied at a pressure between 100 and 150 mm Hg to occlude the superficial venous system, so that any changes observed would indicate deep venous system function. All 53 patients underwent subfascial ligation of perforating veins (saphenofemoral or saphenopopliteal) with or without saphenous vein ligation. Fourteen patients (26) had recurrent ulceration (median time to recurrence, 48 months; range, 10-72 months) during the followup period (median = 60 months; range, 36-144 months). The other 39 patients remained ulcer-free during the followup period (median, 60 months; range, 3-144 months).

Because WPG has not been compared with CV or AVP, it is difficult to judge in which patients it may be of benefit. Even in the study by Bradbury, in which venous ejection time and HRT were prospectively measured, there was no significant difference in either parameter in patients with or without recurrence of their venous ulcer disease. At this time, WPG does not offer unique or reliable information upon which to base current or future treatment in patients with venous insufficiency.

Air Plethysmography

Air plethysmography (APG) or pneumoplethysmography, is a noninvasive PPG method used to aid in the detection of CVI, as well as occlusive PAD.(147,148) Technically speaking, two of the oculoplethysmographic methods previously discussed, OPG-G and OPG-Z, are also air plethysmographs.

Early investigational studies attempted to demonstrate this method of plethysmography as efficacious in measuring forearm blood flow and quantifying pulse volumes, but technical difficulties related to pressurized air as the transmission fluid precluded its use.(149) Newer APG devices consist of a tubular, polyurethane, malleable air chamber with proximal and distal inflatable cuffs that surround the whole lower leg from knee to ankle, along with a pressure transducer, amplifier, and strip-chart recorder. A calibrated volume of air is injected into the cuff to achieve a prescribed pressure. The tightness with which the cuff is wrapped is the variable that can be adjusted to keep the pressure/volume relationship constant.(14 Changes in pressure also occur as the cuff air is warmed by the limb, so final calibration is done after a short waiting period. Pressure changes noted after the pressure/volume relationship stabilizes reflect the changes in segmental blood flow under the cuff. These changes are recorded in an analog system and are converted to a pulse contour for analysis at different segmental locations along the limb. For example, a 1-liter soft plastic calibration bag is placed on the leg inside the air chamber. The patient's leg is elevated to 45 degrees (with heel support) to empty the veins. The operating pressure within the air chamber is increased to 6 mm Hg, the plethysmograph is calibrated, and a resting arterial inflow at a stable temperature is achieved over a 5- to 10-minute period. After a baseline recording, 200 cc of water is injected into the smaller bag in 50 mL increments, and the corresponding pressure changes in the air chamber are recorded. After removing the water, when the pressure returns to baseline, the patient is asked to stand weight-bearing on the opposite leg. The plethysmograph records this change in leg volume until a plateau tracing is achieved (functional venous volume [VV]). Single and repeated tip-toe maneuvers are performed with the patient in standing position to calculate ejectin volume (EV, a measure of efficiency), residual volume (RV, thought to correlate with AVP),(150,151) and the venous filling index (VFI, the mean filling rate of the veins) of the calf pump. The VFI is a function of both venous reflux and arterial inflow.

There have been a few case series that have validated the effectiveness of APG against the reference method of AVP measurements, or CV.(6 -9)

Recently, Gillespie et al(152) reported the use of APG to evaluate venous surgical techniques. Volume changes in the extremity were measured using an air-filled polyurethane cuff applied just below the knee to just above the ankle. Functional VV and VFI (mL/sec) were measured. Normally, the VFI should be relatively slow (<5 mL/sec) due to arterial inflow. However, in the presence of incompetent venous valves, the VFI can be as fast as 25 mL/sec.

While standing on both legs, patients performed a single tip-toe maneuver with the index leg; the reduction in VV represents contraction EV of the muscle. The EF equals the EV divided by the functional VV. After a new plateau is reached, 10 repeated tip-toe movements are performed and the residual volume (RV) is measured. These authors reported a direct correlation between residual volume fraction (RVF = RV/VV) and venous pressure, although this conclusion has been challenged(153) because of the known nonlinearity of the venous pressure/volume relationship. Maximal venous outflow was also assessed with the use of a concurrent thigh cuff inflated to 80 mm Hg pressure and volume outflow calculated at 1 second. In this study, there were 15 extremities with primary varicose veins and 10 extremities with severe lipodermatosclerosis (undefined) with ulceration. In the 21 extremities that underwent greater saphenous vein stripping and ligation (from groin to ankle), mean VFI improved from 6. 7 mL/sec preoperatively to 1.8 mL/sec in the immediate postoperative period (P = .0001). Mean VV decreased from 177 mL to 140 mL after surgery (P = .0008). Less improvement was seen in this group in mean EF (45. 8 to 50.8, P = .07) and mean RVF was not improved at all (45 to 42, P = .4) after surgery. The remaining four extremities underwent autologous popliteal vein valve transplantation, using axillary veins. One patient had immediate postoperative thrombosis. In the other three patients, mean VFI decreased from 14.7 mL/sec to 5.3 mL/sec (P = .2), and mean VV was reduced from 166 mL to 128 mL (P = .4). Ejection fraction and RVF worsened immediately postoperatively and did not show improvement at 3-month followup. Gillespie et al(152) suggested that APG may be too painful to perform correctly postoperatively, and changes in EF or RVF may not occur or may be too difficult to detect.

Van Rijn et al(154) evaluated 50 limbs of patients with suspected DVT by measuring venous outflow by APG and pulse-volume recorder (PVR). Five repeated measures of the 3-second pressure drop after the thigh cuff release were averaged and compared with a discriminant analysis from a previous group of 25 normal limbs and 25 limbs with DVT. The latter group findings were confirmed with radionuclide venography and ascending venography. The APG had a sensitivity of 94 for proximal occlusive disease but very poor sensitivity for proximal nonocclusive disease and calf thrombosis, 33 and 17, respectively.

Welkie et al(150) used APG to evaluate 36 ulcerated limbs (30 patients) and 80 asymptomatic extremities (54 patients). Only 13 asymptomatic and 16 ulcerated limbs underwent evaluation with the reference method (AVP). These authors were able to demonstrate a linear relationship between AVP and RVF in both asymptomatic and ulcerated extremities (r = . 87).

In a 1982 report, Schroeder and Dunn(155) studied a total of 320 patients over a 2-year period with APG. Only 83 of those studied underwent simultaneous evaluation by the reference method, CV (one patient underwent a fibrinogen leg scanning). Sensitivity was 60 (21/35), specificity was 77 (37/48), PPV was 64 (21/33), NPV was 72 (37/51), false-positive and false-negative rates were 13 and 17, respectively. Overall accuracy was 70; the prevalence of venous disease in the study population was 42.

Neglen and Raju,(6) in a 1993 report, tried to validate AVP measurements, including venous refilling time (VRT), the Valsalva test, and APG against the clinical stage of CVI, and to present a logical strategy for investigation of reflux or venous insufficiency. Their findings were correlated to standardized quantification of reflux with erect DU scanning. The morphologic distribution of venous incompetence (erect duplex and descending venography), AVP measurements, VRT, the Valsalva maneuver, APG (venous refill index, VRI), and clinical severity were described in 118 patients. Fifty-one of these patients had descending venography performed. The results of the venograms were not reported. It appears that measures of accuracy for the hemodynamic studies (flow, pressure) were compared to duplex scanning or clinical severity rather than venography. Patients were placed in two clinical severity grades, mild or severe, and morphologic and hemodynamic studies were compared.

The hemodynamic studies (AVP measurements, VFT, the Valsalva maneuver, VFI, and venous RVF) all demonstrated significant abnormal results in patients with severe disease when compared with patients with mild disease. The authors further reported accuracy values for ultrasound, AVP, Valsalva, and VFI, but never made clear what reference method was used to calculate them. Although they developed a segmental ultrasound scoring system for the presence of reflux in a segment (superficial femoral, long saphenous, popliteal, short saphenous, and proximal and distal posterior tibial veins), deep veins were not visualized consistently. In those cases, the authors recommended adding a hemodynamic test to increase diagnostic accuracy.

Cordts et al(156) compared hemodynamic changes using APG in patients with primary varicose veins without skin changes (39 extremities) with those in patients with severe CVI (83 extremities). The control group consisted of 59 undiseased (by physical examination and venous duplex imaging) limbs with no history of DVT. Control and case inclusion and exclusion criteria were not stated. The 39 cases came from extremities with primary varicose veins, of which 48 had greater saphenous vein incompetence by duplex scan. None of these 39 limbs had hyperpigmentation, induration, or eczematous changes suggestive of postphlebitic syndrome. The second group of cases included 83 extremities, each exhibiting frank ulceration. All groups were studied with high-resolution duplex venous imaging and APG. Results showed that MVO was identical in the normal and varicose vein groups (mean, 54.4), and slightly lower in patients with ulceration (51.3). Venous filling index was significantly faster in the cases than in the controls (5.4 mL/ sec, 6.8 mL/sec, and 1.0 mL/sec, respectively). Venous volumes ranged from 38-178 mL, with little distinction between patients and controls. All control limbs had VFI values <2.5 mL/sec. The extremities with ulceration and varicose veins varied from no reflux to severe reflux (>10 mL/sec), but 46 of ulcer limbs had mild or no reflux. The RVF in 63 (37/59) of control limbs was <35, whereas in limbs with varicosities 69 (27/39) had an RVF >35, and 54 (45/83) of the limbs with CVI had an RVF <50. The authors state that the best APG parameter to distinguish CVI patients from those with varicose veins is the RVF, a direct estimate of AVP (no citation or data supplied).

Air plethysmography, PPG, and DU (quantitative valve closure time, TVCT) were used by Welch et al(93) to differentiate severe venous reflux from normal venous segments. In evaluating 20 limbs (17 patients) with mild to severe venous reflux, they demonstrated that the VFI increased as the severity of reflux increased. Mean VFI ranged from 1.92 ± 0.24 mL/sec in normal limbs (controls) to 6.44 ± 1.98 mL/sec in patients with mild reflux to 10.26 ± 1.75 mL/sec in patients with severe reflux, but they were unable to distinguish mild from severe reflux based on EFC. Patients with severe reflux had over twice the RVF as did those with mild reflux (P < .05). Part of the difficulty in interpreting the results of this case series is the fact that the patients were grouped according to their physical findings, rather than discrete (separable) clinical entities. For example, heaviness, skin pigmentation, severe swelling, and liposclerosis were nonquantified variables used to differentiate the severity of disease, and four of the eight patients in the mild reflux group demonstrated symptoms equivalent to those nine patients in the severe venous reflux group. Residual volume fraction was significantly different between normal veins and those with severe reflux (P < .05). Light-reflection rheography was able to distinctly separate normal limbs from those with reflux but was unable to differentiate the degree of reflux.

Some investigators have used APG to evaluate chronic peripheral arterial insufficiency. (157) The PVR has been used since the late 1970s to comparatively study pressure amplitude changes in patients with PAD. It is a segmental air plethysmograph and assumes that, at a given pressure, the volume in the cuff surrounding the limb is constant from reading to reading.(158) This operating principle has been challenged by Neumann et al(144) as contrary to the known nonlinear relationship of pressure to volume in the vascular system. The PVR operating principle is similar to that of a sphygmomanometer. A series of sphygmomanometric blood pressure cuffs are applied to the thigh, calf, and ankle; a known volume of air is injected into the cuffs; and the pressure amplitude degradation (from proximal to distal segments) is plotted and measured and the level of arterial obstruction estimated. In an early study,(148) PVR was used intraoperatively to measure pressure changes before and after arterial reconstructive surgery of the lower extremities. In this prospective study by Baird et al,(148) 83 arterial reconstructions were performed: 31 distal grafts to the popliteal artery (18 femoropopliteal, 13 iliofemoral) and 52 proximal grafts (31 aortoiliac/femoral, 14 aneurysmectomies with sleeve grafts, three transpubic [femoro-femoral] and four axillo-bifemoral). Twenty-two control limbs were also studied. Lower limb ischemia was evaluated by segmental PVR, by Doppler blood pressure measurements at the thigh, calf, and ankle, and by contrast arteriography. These authors were interested in demonstrating the utility of PVR to predict the postoperative patency of these vessels by measuring PVR and Doppler systolic pressure intraoperatively immediately after the bypass graft was in place and arterial circulation was reestablished to the lower limbs. Three PVR measurements were used: upstroke (mL/sec, the time interval between the foot of the upstroke and the point of maximum deflection), transit time (from the R wave of the electrocardiogram tothe peaks of the PVR and Doppler waveforms), and amplitude. Mean PVR amplitude in 30 vessels with aortoiliac disease (with a patent superficial femoral artery) was 41 mm (range, 29-56 mm; 95 confidence interval [CI]) preoperatively and almost identical postoperatively (mean, 42 mm; range, 31-62 mm; 95 CI). When the superficial femoral artery was not patent (22 vessels), mean amplitude was 11 mm (range, 6-14 mm; 95 CI) preoperatively and 8 mm (range, 6-15 mm; 95 CI) postoperatively. Mean PVR amplitudes in patients with popliteal artery grafts (31 vessels) were 4 mm (range, 1-6 mm; 95 CI) and improved immediately to 18 mm (range, 15-50 mm; 95 CI) after intraoperative graft insertion. It is interesting to note that in the 22 vessels that were used as controls (contralateral limbs used during reconstruction) amplitude of the PVR waveform fell from 38 mm (range, 28-53 mm; 95 CI) during reconstruction but returned to normal in the immediate postoperative period. Clearly, measuring the amplitude of the PVR was not helpful in determining future patency or graft survival. Mean transit time measured in patients with combined proximal and distal disease was prolonged preoperatively at 577 msec (range, 542-633 msec; 95 CI), compared with patients with patent distal systems (455 msec; range, 430-503 msec; 95 CI). After reconstruction, filling times were reported to return to normal in the patent distal group, although no quantitative data were given. In patients with proximal reconstruction in the presence of distal disease, transit time after 5 minutes was shortened to 481 msec (range, 467-495 msec; 95 CI; P = .05). In this series of patients, when transit time was delayed (showed no improvement) on initial declamping of the artery, immediate operative revision of the graft was undertaken. The authors suggested that if there was confirmation of a "good, technical result [at 30 minutes after reconstruction], the probability of graft patency is high." Unfortunately, the authors did not define a "good, technical resul" and further noted that there was no pedal pulse (measured by palpation or Doppler) in any of the limbs at any stage, and ankle Dopplers were absent in 23 and undecipherable in an additional 18 postreconstruction. All grafts remained patent in the immediate postoperative period.

As in other plethysmographic methods, pulse wave analysis with PVR, which includes visual assessment of the contour (upstroke, peak, downstroke) and amplitude (height) of the arterial waveform, gives a qualitative assessment of the degree of vessel occlusion.(159) Kempczinski(160) studied 568 patients over a 5-year period with segmental limb systolic blood pressure measurements and segmental volume plethysmography to assess lower extremity arterial occlusive disease. Only 105 of these patients underwent biplanar arteriograms and serve as the basis for comparison for sensitivity, specificity, and predictive values. In this group of patients, the best noninvasive index of aortoiliac disease was a >25 mm Hg arterial pressure differential between the brachial and femoral systolic pressure. A differential >20 mm Hg between thigh and calf gave the best accuracy in detecting superficial femoral artery occlusion. When these parameters were compared with arteriography, segmental limb pressure accuracy in aortoiliac vessels was 53, with a 33 false-positive rate and a 14 false-negative rate, whereas in superficial femoral vessels, accuracy was 62, with a 10 false-positive rate and a 28 false-negative rate. The lesion most responsible for the inaccuracies was total occlusion of the superficial femoral artery.

First described by Cranley(161) in 1973, PRG is an APG technique that involves measurement of extremity blood volume changes and respiratory changes by means of recording cuffs placed at the midthorax and at four limb levels -- midthigh, and upper, mid-, and lower calf. These multiple segmental limb pneumatic cuffs are used as air plethysmographs to measure volumetric responses to respiration and limb compression.(159) The lower calf cuff and a foot cuff are also inflatable compression cuffs, used to mimic the action of the calf venous pump during exercise. Each recording cuff measures two phenomena: 1) phasic changes in venous pressure reflecting transmission of respiratory fluctuations in intraabdominal pressure (respiratory waves), and 2) increase or decrease of overall blood volume in the leg manifested by the rise or fall of the baseline tracing after compression. The presence of obstructive venous thrombi will block the transmission of the respiratory waves, whereas repeated compression of the foot or calf in the presence of an obstruction will propel venous capillary blood into the deep venous system, increasing the venous volume of the limb and eventually raising the baseline tracing. These two changes are thought to be diagnostic for acute DVT. The presence of a baseline rise in the absence of any diminution of respiratory waves is diagnostic for chronic venous obstruction. The absence of a change in the amplitude of the respiratory waves is due to venous recanalization.

Bynum et al(162) studied 75 patients who were referred for CV because of suspected DVT. In a blinded comparison between PRG and CV, overall accuracy was 81, with a false-negative rate of 15. The acute DVT false-negative rate was 31.

Classen et al(163) reported slightly improved accuracy (93), with a lower false-negative rate (4.4) and a false-positive rate of 2. This study was retrospective and involved 700 patients (1,758 lower limbs), only 90 of whom were tested with the reference method, CV. Patients were excluded if venography was not obtained. Of those tested, 8 of the results were uninterpretable or equivocal and were excluded from analysis.

Sullivan et al(164) retrospectively reported 40 patients (90 limbs) over a 10-year period who underwent PRG of the upper extremity for suspected DVT. Sixteen patients (22 limbs) also underwent venography. The sole criterion for a positive test in these patients was a rise in the baseline of the PRG. It was reported in previous case studies that the respiratory wave does not disappear in upper extremity venous occlusions. There was a 14 false-negative rate.

Ouriel et al(165) studied 137 patients (216 limbs) of 802 patients (17) with venous Doppler, PRG, and venography. All patients had been referred to them over a 4-year period with suspected DVT. The authors stated that referral for venography in the 137 patients was an attempt to get additional information in patients whose noninvasive tests and clinical impressions were disparate. After eliminating equivocal test results, the accuracy of PRG alone was computed as 92, with sensitivity and specificity 81 and 96, respectively. By combining the results of PRG and Doppler, a positive result on either test increased sensitivity, specificity, and accuracy to 88, 99, and 95, respectively. Both tests combined still-missed small thrombi in the calf and popliteal area.

Sottiurrai et al(166) retrospectively reported the PRG results of 23 out of 42 (54) patients with clinically suspected DVT of the upper extremity. The authors did not explain how the 23 patients differed from the total study population or why they were referred for venography. The authors recommend using PRG along with Doppler ultrasound to improve accuracy (from 82 to 91).

Comerota et al(167) reviewed their 10-year experience with PRG. Those results are listed in Table 13. Of note, isolated calf thrombus and popliteal vein thrombosis accounted for most of the false-negative results. Overall, there was a 3 false-positive and 3 false-negative rate over the 10-year period.

Stallworth et al,(168) in a retrospective case series, reported the results of PRG examinations in 1,076 patients (2,152 limbs) that had been examined because of clinical evidence of DVT or a demonstrated PE on lung scan. Of these patients, 118 (11) were diagnosed with acute DVT, 120 (11) demonstrated partial obstruction, and 838 (78) had negative results. There were 458 outpatients (43) and 618 inpatients (57). Thirty-nine of the 618 patients (6.3) underwent CV. Twenty-five of the 39 patients (64) had a negative CV, and 14 patients tested positive (36). Phleborheography was negative in 24 and positive in 13 patients (sensitivity, 0.93; specificity, 0.96). Of the 458 outpatients, 392 had negative PRG examinations (86).

In all, over 1,100 limbs were evaluated with PRG and CV in six retrospective studies. All patients were described as either having PE or suspected DVT, although prevalence of disease (confirmed by venography) varied from 26 to 83. Nonocclusive distal disease attributed to the majority of false-negative examinations.

Oculoplethysmography

Plethysmography achieves its greatest predictive value in those patients that have at least a 60 stenosis of their extracranial vessels and flow has been reduced by 90. Non-flow-reducing ulcerative plaque and bilateral occlusive disease is difficult to diagnose with OPG. Operable stenotic disease cannot be differentiated from inoperable occlusive disease with OPG.(171) Relying on a nonimaging method such as OPG is probably inappropriate because of the inaccuracies of the different methods, and the very high risk of future stroke or thrombotic events due to misclassification (false negatives, false positives) in asymptomatic or symptomatic patients. Although not a plethysmographic method, we evaluated recent studies involving the use of CPA and believe that it should not be used as a diagnostic tool in carotid disease.(56) Whether used with OPG-K or as a stand-alone diagnostic test, CPA was nonspecific (PPV ranged from 38 to 88) in detecting the presence of carotid artery disease.

Impedance Plethysmography

The clinical studies reviewed support the accuracy and efficacy of IPG for noninvasive testing in acute symptomatic DVT.(172-174) Thrombotic or embolic proximal disease that reduces flow can be detected by IPG (sensitivity, 94; specificity, 95). Symptomatic distal (calf) thrombosis and asymptomatic proximal disease are more difficult to detect by IPG (sensitivity ranged from 12 to 58 and specificity ranged from 95 to 98). Impedance plethysmography is not accurate in recurrent thrombophlebitis in the presence of patent collaterals or recanalized vessels. B-mode ultrasound and DU (B-mode with Doppler) are more sensitive than IPG for detecting nonocclusive thrombi.(19,174-181) Serial IPG has been validated in outpatients in detecting thromboembolic disease and has served as the basis for instituting successful anticoagulant treatment for the prevention of PE. Impedance plethysmography is easy to perform by a technician, and the results can be objectively read without on-site physician interpretation.

Photoplethysmography

The current effort to distinguish LRR, also known as quantitative photoplethysmography, from PPG is clinically obscure, technically confusing, and probably being driven by differences in reimbursement rates. In fact, the two technologies are quite similar in operating principle, light reflectance measurements, and test results (semiquantitative). The accuracy of PPG ranged from 74 to 95 in the clinical studies reviewed. Although capable of differentiating between limbs with venous reflux and those without, PPG is unable to distinguish the reflux severity. The sensitivity was similar in proximal as well as calf vein thrombosis. In cases of false-positive examinations, patients usually had comorbid conditions that interfered with performance of the device, i.e., leg edema, poor range of motion. In the study involving 75 emergency room patients,(140) there was a 24 false-positive rate. No analysis of these patients was presented. Analyses of false-negative results showed partial thrombotic lesions were not detected by LRR.

Light-reflection rheography can provide immediate, safe, qualitative information on venous outflow in patients with suspected venous insufficiency. Venous emptying times of less than 25-30 seconds signify venous insufficiency and usually require further testing with an imaging system.

Photoplethysmography is reliable, safe, and easy to administer and may be used as an initial diagnostic technique in both PAD and CVI.

Strain Gauge Plethysmography

Strain gauge plethysmography does not provide reliable quantitative data in partially occluded vessels in which flow is marginally reduced, but it does provide an indication of the qualitative function (slow venous refill) of the extremity in venous insufficiency.

Strain gauge plethysmography is safe, does not require the technical skill of an ultrasound technologist to perform, and is a relatively inexpensive, noninvasive, diagnostic tool.

Air Plethysmography, Phleborheography, and Pulse-Volume Recorder

Several clinical studies have demonstrated the marginal effectiveness of APG in the diagnosis of functional peripheral venous disease and extracranial arterial disease with overall accuracy rates ranging from 53 to 70.(150,152,154,155,160) In the PRG studies, which involved a total of 1,119 limbs, accuracy ranged from 82 to 94. The studies that used segmental pressure/volume measurements by APG were useful in determining the anatomic level of arterial obstruction and in qualifying the reduction in flow in venous and arterial obstructive diseases. It is a valid initial noninvasive test but may need to be followed by an imaging test before surgery is contemplated.

Venous insufficiency may develop after acute episodes of DVT or may develop through an insidious process that remains subclinical until functional impairment, i.e., pain on exercise. For these patients, whose treatment options may be limited beyond palliation, functional information of the venous circulation can be a valuable guide to the clinician in deciding on a course of action.

Water Plethysmography

Based on the reviewed available studies, WPG has not been validated to a reference method as a reliable diagnostic procedure. Although foot volumetry has been used to evaluate CVI, the water plethysmograph remains an unvalidated diagnostic method.