Intradialytic hypotension (IDH), occurring in 20% to 50% of dialysis sessions, 1 is caused primarily by the decrease in blood volume induced by excessive ultrafiltration, 2 lack of compensatory vasoconstriction, 3 and to autonomic insufficiency 4 or a combination of other issues. 5,6 IDH can cause undesirable symptoms such as nausea, vomiting, and cramps, as well as life-threatening conditions such as acute coronary syndrome, transient ischemic attack, cerebrovascular accident, and nonocclusive mesenteric ischemia. 3,7,8 In addition, IDH can result in dialysis under delivery secondary to saline infusion or lowering of blood flow during hypotensive periods. 3
Standard cuff blood pressure (BP) measurements are usually performed at a predetermined schedule during the dialysis session and repeated when demanded by patient’s symptoms. Thus, early recognition of a hypotensive episode during dialysis may prevent patient discomfort and morbidity as well as reduce nursing interventions. Online relative blood volume monitoring has been used to help prevent volume related hypotensive episodes during hemodialysis. 9 However, it has been shown that a predictable individual blood volume threshold for hypotension exists in only 70% of the patients. 10 There is, also, considerable interindividual and intra-individual variability in the hematocrit–BP relationship during dialysis. 11 In addition, a hypotensive episode may occur during hemodialysis without a sudden decrease in plasma volume. 3 Prevention and early recognition of IDH still remain an important challenge to the dialysis physician.
The Harmonized Alert Sensing Technology (HASTE) device is a new noninvasive, continuous monitoring technology developed by COLIN Corporation (Komaki-City, Japan). By constantly monitoring delay time (DT) between electrocardiogram (ECG) and photoplethysmogram (PPG) pulse wave, the system estimates the systolic BP of the patient and visually displays the estimated systolic BP (ESYS) trend on a beat to beat basis. During the session, a more intensely monitored patient with the HASTE should allow for a more focused response to patient hemodynamic changes earlier than with the pre-established scheduled pressure determination approach. The HASTE system may allow for more responsive and focused patient care activities and yet at the same time diminish the caregiver’s workload. This is the first prospective evaluation of both concept and technology associated with the HASTE system on chronic dialysis patients compared with BPs obtained using standard sphygmomanometer readings.
Patients and Methods
Seventeen (10 males and 7 females) patients with end-stage renal disease undergoing hemodialysis treatments three times a week at the Cleveland Clinic Foundation were studied. Their mean age was 62.9 ± 15.4 (range 33–85) years, and their mean hemodialysis duration was 27.2 ± 25.5 (range 1–84) months. Seven (41%) patients had a peripheral access, arteriovenous fistulae, or graft; the remaining 10 (59%) patients had a central double lumen venous access (permanent or temporary). Each patient was studied on three sequential hemodialysis sessions. The entire study was conducted within a period of 1 week.
Treatments were excluded if patients had cardiac arrhythmia, excessive extremity movement during the study period, or accurate PPG measurements could not be obtained. Treatments were also excluded for incomplete data collection. The protocol was approved by the institutional review board, and all participants gave informed consent.
Hemodialysis was initiated and prescribed at the discretion of the patients’ attending physician and was performed with the Althin system 1000 (Baxter Medical, McGaw Park, IL) machine using single use F70 dialyzers (Fresenius, Lexington, MA). Dialysate temperature was set at 36°C to 38°C. Dialysate composition varied according to clinical needs, but the default dialysate was composed of 35 mmol/L of bicarbonate, 138 mmol/L of sodium, 1.5 mmol/L of calcium, and 2 mmol/L of potassium. The individual dialysis session duration ranged from 3 to 4.5 hours and remained constant for each patient.
Readings for ESYS were continuously obtained by combining continuous ECG and PPG pulse wave information with periodic cuff systolic blood pressure (NIBP) measurements. ECG and PPG pulse wave information are used to calculate DT. PPG can detect the changes in the volume of the arterial blood in the peripheral vascular bed and provides a continuous, relative measurement of the pulsatile blood volume. PPG data were obtained by Nellcor pulse oximetry (Nellcor, Pleasanton, CA). As shown in Figure 1, DT is measured from ECG generated R-waves to the maximum slope the PPG pulse wave obtained with pulse oximetry.
The systolic BP was then estimated from DT using the linear equation ESYS = αDT + β, where ESYS is the estimated systolic BP (mm Hg), α is the calibration coefficient relating pressure to DT (mm Hg/ms), β is the calibration constant (mm Hg), and DT is the pulse delay time (ms).
At the start of monitoring, a population-based calibration coefficient, α, is used for estimating systolic BP derived from 50 patients not on dialysis in whom BP was measured invasively. Because ESYS produced using the population-based coefficients have a high error rate, the calibration coefficient and constant are adjusted after each NIBP calibration measurement. The HASTE system uses a method for incrementally adjusting the calibration coefficient followed by recomputation of the calibration constant. With this method, the error of ESYS, ΔP, is determined by subtracting the newest NIBP measurement from the ESYS at the time of the NIBP measurement. On the basis of the magnitude of the error, the regression coefficient is adjusted by a predetermined amount. After adjustment, the calibration constant, β, is recomputed using the new calibration coefficient, NIBP, and DT by the equation β = PNIBP − α × DT, where PNIBP is last cuff BP value, DT is delay time, and α is the calibration coefficient.
The entire dialysis session was monitored. Because height differences between two arms cause hydrostatic pressure differences, all patients were required to keep both arms at heart level during all measurements. To allow some patient movement, a BP cuff of appropriate size was wrapped on the upper arm without a functioning arteriovenous fistulae, and a pulse oximetry probe was placed on the fingertip of the opposite arm. Three electrodes for ECG were placed on the chest.
BP cuff measurements were performed using a device based on standard oscillometric methodology and was incorporated into the HASTE system both as the comparison entity for HASTE calibration as well as the stand alone standard for BP measurement. Although this method was not compared with any other “standard,” its use in dialysis is well established. Furthermore, using the same method for the entire trial allows us to be consistent in our comparisons and thus allows for statistical review. We initially performed NIBP readings every 10 minutes to calibrate the HASTE system. After 40 minutes (calibration period), the system was internally set to take NIBP measurements every 20 minutes for calibration purposes and at every 5 minute interval for comparison in the measurement period. Although it is true that the continued use of the NIBP to “recalibrate” the HASTE introduced a bias in favor of the HASTE methodology, the results continued to be compared with the unchanging NIBP methodology. Systolic, diastolic, and mean arterial pressure values obtained from the cuff measurement and beat to beat HASTE ESYS values were recorded on a laptop computer attached to the HASTE system. During the measurement period, the HASTE system continuously recorded presumed changes in BP beat to beat as ESYS was visually displayed. Patient movement, cardiac arrhythmia, and inaccurate PPG pulse wave existence were noted.
Comparison between NIBP and HASTE ESYS Values
The study protocol required NIBP measurements to be taken every 5 minutes. The HASTE ESYSs were generated on a beat to beat basis. We used the average of the HASTE ESYS values recorded from the beginning of cuff inflation to the end of cuff deflation of each cuff measurement to compare NIBP values.
Hypothesis tests were regarded as statistically significant if p < 0.05 (two sided). Values were described by means and standard deviations. The degree to which the ESYS and NIBP measures agreed were characterized using Pearson correlation coefficients as was the median difference from perfect agreement (to characterize bias) and the median of the absolute value of the difference from perfect agreement (to characterize variability). The mean and standard deviation of the difference from perfect agreement were also computed, and Bland-Altman plots were produced for further illustration. All analyses were conducted using SAS software (Unix version 8.0, Cary, NC).
Seventeen patients consented to the study. There were 42 eligible sessions for the 15 patients. Nine additional sessions of five patients were excluded because of a variety of reasons that include arrhythmia, extreme extremity motions, dialysis stop, insufficient PPG pulse wave information, and incomplete data collection.
An average of 32 (range 16–51) simultaneous cuff and HASTE ESYS measurements were performed per dialysis session, and a total of 1,370 measurements for 15 patients were compared. The consistency of HASTE was tested according to accepted standards of the British Hypertension Society (BHS) as well as the standards established by the Association for the Advancement of Medical Instrumentation (AAMI). These standards are shown in Table 1. An overall mean difference between HASTE ESYS and cuff measurements was 1.41 ± 16.90 mm Hg. This is beyond the limits of agreement required by the AAMI. For HASTE measurements compared with cuff measurements, 31% of systolic readings fell within a 5 mm Hg difference range, 57% of systolic readings fell within a 10 mm Hg range, and 73% of systolic readings fell within a 15 mm Hg band. The HASTE results revealed a consistently poor correlation with cuff measurements (i.e., grade “D” according to BHS Protocol and “fail” according to AAMI standards).
As exemplified in Figure 2, the agreement analysis between two measurement values showed high correlation (Pearson correlation coefficient 0.80) and low bias (median change from identity line 1.07 mm Hg). The degree of dispersion, however, was troublesome (median of absolute differences from identity line 8.49 mm Hg).
The individual comparison was plotted as a Bland-Altman plot in Figure 3. The Bland-Altman plot demonstrated a considerable variability between the values of the two methods (mean absolute difference 11.8 mm Hg). The individual average differences between cuff measurements and the HASTE systolic BP ranged from 0.2 to 14 mm Hg. Seventeen of 42 sessions for eight different patients revealed a remarkable visual similarity between NIBP and the HASTE BP trend. However, we did not find the same similarity in the rest of the 25 sessions from 14 patients’ NIBP and the HASTE BP trends (Figures 4 and 5 depict two patients’ HASTE and NIBP BP trends as a graph). A 1 mm Hg change in systolic BP changes the DT to 0.085 ms in the opposite direction, but the association between systolic BP and DT was not statistically significant (p = 0.15)
The system was recalibrated every 20 minutes to enable compensation for the effects of variability of the vascular properties that may interfere with pulse arrival time (PAT) that may result in misinterpretation of BP estimates. As a result of this ongoing recalibration, the late HASTE ESYSs were better correlated with NIBP values than were the earlier values, but they never ultimately reached statistical significance (p = 0.090). Furthermore, there were insufficient numbers to perform meaningful statistical comparisons to account for the differing pulse pressures noted among patients.
Current practice requires preset periodic BP monitoring during hemodialysis. Alteration of the frequency of BP monitoring is usually performed in response to clinical expression of hypotension (i.e., nausea, light-headedness, or cramping). However, in some patients, there are no symptoms whatsoever until the BP falls to extremely low levels. There have been many attempts to identify specific patients who may be prone to hypotensive episodes and to trace all patients with devices that measure known causes of hypotension such as online blood volume monitoring. The development of intraarterial pressure monitoring has proven useful in the intensive care unit setting to help identify and reveal effective treatment of hypotension in a variety of patient populations. The invasive nature of this technique makes it impractical and unwarranted in the outpatient chronic dialysis population. The ability to mimic these continuous readings by using a noninvasive methodology would be a valuable and worthwhile contribution toward dialysis hemodynamic stability. The HASTE system was developed to accomplish this goal in patients who have a high possibility of hemodynamic instability in hemodialysis or intensive care units. As part of a developmental process, we tested this system in an outpatient dialysis population where such continuous monitoring might increase patient stability and at the same time decrease caregivers’ workload.
In our study, the overall mean difference between NIBP and HASTE ESYS values was 1.41 ± 16.90 mm Hg. The present study demonstrated that the standard deviation of the mean difference between NIBP measurements and ESYS values was extremely large. On the basis of AAMI, 12 HASTE for estimating systolic BP in chronic hemodialysis patients was determined to be inaccurate and was not appropriate for clinical use. Likewise, according to BHS guideline, 13 the machine was also graded as unacceptable. The HASTE technology cannot replace other BP measuring devices at this time, but inaccuracies may be overcome.
There have been numerous attempts to use pulse wave velocity or PAT to estimate BP noninvasively. 14–16 The PPG technique offers an alternative means of measuring DT and therefore estimating BP. DT is the sum of the preejection period (PEP) and PAT. PEP is the interval during which the contracting myocardium raises the intraventricular pressure sufficiently to open the aortic valve and push blood out of the ventricle. An earlier study has shown that PEP variability is a significant source of variability in the population’s BP–DT relationship. 17 However, it has been shown that PEP can vary with sympathetic nerve system stimulus rather than changes in BP. 18 PAT indicates the time delay from aortic opening to the arrival of the blood pulse wave at the peripheral vessels. An increase in BP in individual subjects tends to reduce arterial compliance and therefore reduce PAT. 19 However, PAT is influenced by peripheral resistance, venous return, and other cardiovascular variables such as respiratory rate (RR) interval and BP. 20 Because of those factors, interdependency between systolic BP and DT does not always remain highly correlated. To solve this difficulty, Chen et al.14 developed a method of combining the PAT intermittent calibration measurements to estimate systolic BP continuously. They showed that the correlation coefficients between estimated values and invasively obtained systolic pressure reached 0.97 ± 0.02, and error remained within ±10% in 97.85% of the monitoring periods.
In our study, the correlation between DT and systolic BP is not strong enough to serve as a marker for BP changes in hemodialysis patients. The HASTE system uses an α constant to estimate systolic BP from DT. This α constant was obtained from 50 patients not on dialysis, thus making its application to older patients on hemodialysis who had a diseased circulatory system somewhat suspect. Nitzan et al.21 found that the values of PAT decreased as a function of the subject’s age and claimed that the decrease in the PAT parameters is attributed to the direct structural decrease of the arterial compliance with age. Because the two populations had different arterial wall characteristics, using the same constant may have influenced the result of our measurements.
Data generation collection and interpretation were also hampered by anatomic and physiologic differences unique to this population. Arteriovenous fistulae flow, rapid change in blood viscosity, and electrolyte concentration during the dialysis will all interfere with ongoing measurements seen in the hemodialysis population.
Arteriovenous fistulae will diminish distal blood flow to the extremity and can decrease the quality of the PPG pulse wave. Pulse oximetry may fail or give false readings if peripheral blood flow is markedly diminished. 22 In turn, insufficient wave pattern can influence DT and ESYS. Another important consideration might be the high arrhythmia incidence in hemodialysis patients caused predominantly by fluctuation of electrolyte concentration during dialysis. Because HASTE depends upon regular cardiac rhythm, the existence of the arrhythmia can potentially decrease the sensitivity of the HASTE system.
BP measurements were obtained at different sites and arms by different devices on the basis of different methodology. To understand pressure differences between finger and brachial artery, it is necessary to consider the physiologic factors affecting the pressure transmission from the upper arm to the finger. The pressure gradient along the vascular tree causes the finger pressure to be below the mean radial or brachial pressure. 23,24 The decrease in arterial diameter might be expected to cause a lower finger pressure as observed in patients with vascular disease and in patients with vasoconstriction. 25,26 We obtained cuff BP measurements from upper arms, whereas PPG pulse waves were obtained from fingertips. Because the sites at which the pressure is measured are not identical, we cannot expect to measure identical pressure. However, this relationship should be constant. In addition, although the nonarteriovenous fistulae arm is preferred for measuring BP, we attached the pulse oximetry probe to the opposite fistulae arm. BP measurements were obtained from different arms. Up to three quarters of individuals exhibit a difference in BP between the right and left arms that exceeds 10 mm Hg while seated, and supine pressure tends to be higher on the right side. 27,28 Finally, it is also true that increased stiffness in the arterial tree independently decreases PAT and may deteriorate the BP–PAT relationship. Because our test population exhibited severe atherosclerotic disease, it is possible that the above mentioned factors negatively influenced the precision of HASTE for estimating systolic BP.
The HASTE gave information on the cardiac cycles and its resulting BP on a beat to beat basis. Intermittent BP readings cannot properly assess BP variability. Thus, a novel method of hemodynamic monitoring that noninvasively follows BP on a beat to beat basis and is needed. The HASTE system does not seem ready to replace current BP measuring devices, but it may give the physician additional information with regard to BP trends after some corrections. The system also has the ability to trigger cuff BP measurement in response to BP changes. New studies are needed to evaluate the sensitivity and specificity of this HASTE triggering characteristic. In addition, combining use of the HASTE system with online blood volume monitoring devices may increase the sensitivity of the online device and thus prevent hypotensive episodes during dialysis.
The noninvasive hemodynamic monitoring potential offered by the HASTE system is indeed powerful, and the ability to diagnose hypotension much earlier and to respond more aggressively can make the individual dialysis session less symptomatic and more effective. Focused health care staffing can be more appropriately allocated, and correlation between hemodynamic stability and other monitoring systems can be easily established without having to wait for patient symptoms or signs.
Although it is unfortunate that, as currently configured, the HASTE system did not perform adequately enough to be used as a replacement for prescheduled cuff readings, the potential to track trends and trigger readings may prove useful in today’s dialysis experience after adjustments in both conversion factors and site analysis. The resulting accuracy may well improve to the point of clinical utility.
1. Schreiber MJ Jr: Setting the stage. Am J Kidney Dis 38: S1–S10, 2001.
2. De Vries JP, Donker AJ, De Vries PM: Prevention of hypovolemia-induced hypotension during hemodialysis by means of an optical reflection method. Int J Artif Organs 17: 209–214, 1994.
3. Daugirdas JT: Pathophysiology of dialysis hypotension: an update. Am J Kidney Dis 38: S11–S17, 2001.
4. Kersh ES, Kronfield SJ, Unger A, et al: Autonomic insufficiency in uremia as a cause of hemodialysis-induced hypotension. N Engl J Med 290: 650–653, 1974.
5. Sherman RA: Intradialytic hypotension: an overview of recent, unresolved and overlooked issues. Semin Dial 15: 141–143, 2002.
6. Van Der Sande FM, Kooman JP, Leunissen KM: Intradialytic hypotension: new concepts on an old problem. Nephrol Dial Transplant 15: 1746–1748, 2000.
7. John AS, Tuerff SD, Kerstein MD: Nonocclusive mesenteric infarction in hemodialysis patients. J Am Coll Surg 190: 84–88, 2000.
8. Perazella MA: Pharmacologic options available to treat symptomatic intradialytic hypotension. Am J Kidney Dis 38: S26–S36, 2001.
9. De Vries JP, Olthof CG, Visser V, et al: Continuous measurement of blood volume during hemodialysis by an optical method. ASAIO J 38: M181–M185, 1992.
10. Passlick-Deetjen J, Baldamus C, Ries W, et al: Changes in blood volume and blood pressure: an indicator for symptomatic hypotension? Abstract. J Am Soc Nephrol
10: 298A, 1999.
11. Krepel HP, Nette RW, Akcahuseyin E, Weimar W, Zietse R: Variability of relative blood volume during haemodialysis. Nephrol Dial Transplant 15: 673–679, 2000.
12. Association for the Advancement of Medical Instrumentation. Proposed Standard for Electronic or Automated Sphygmomanometers
. Arlington, VA: Association for the Advancement of Medical Instrumentation, 1992.
13. O’Brien E, Petrie J, Littler W, et al: The British Hypertension Society protocol for the evaluation of automated and semi-automated blood pressure measuring devices with special reference to ambulatory systems. J Hypertens 8: 607–619, 1990.
14. Chen W, Kobayashi T, Ichikawa S, Takeuchi Y, Togawa T: Continuous estimation of systolic blood pressure using the pulse arrival time and intermittent calibration. Med Biol Eng Comput 38: 569–574, 2000.
15. Lane JD, Greenstadt L, Shapiro D, Rubinstein E: Pulse transit time and blood pressure: an intensive analysis. Psychophysiology 20: 45–49, 1983.
16. Wippermann CF, Schranz D, Huth RG: Evaluation of the pulse wave arrival time as a marker for blood pressure changes in critically ill infants and children. J Clin Monit 11: 324–328, 1995.
17. Geddes LA, Voelz MH, Babbs CF, Bourland JD, Tacker WA: Pulse transit time as an indicator of arterial blood pressure. Psychophysiology 18: 71–74, 1981.
18. Newlin DB: Relationships of pulse transmission times to pre-ejection period and blood pressure. Psychophysiology 18: 316–321, 1981.
19. Asmar R, Benetos A, Topouchian J, et al: Assessment of arterial distensibility by automatic pulse wave velocity measurement. Validation and clinical application studies. Hypertension
26: 485–490, 1995.
20. Drinnan MJ, Allen J, Murray A: Relation between heart rate and pulse transit time during paced respiration. Physiol Meas 22: 425–432, 2001.
21. Nitzan M, Khanokh B, Slovik Y: The difference in pulse transit time to the toe and finger measured by photoplethysmography. Physiol Meas 23: 85–93, 2002.
22. Kurki TS, Smith NT, Sanford TJ Jr, Head N: Pulse oximetry and finger blood pressure measurement during open-heart surgery. J Clin Monit 5: 221–228, 1989.
23. Imholz BP, Settels JJ, van der Meiracker AH, Wesseling KH, Wieling W: Non-invasive continuous finger blood pressure measurement during orthostatic stress compared to intra-arterial pressure. Cardiovasc Res 24: 214–221, 1990.
24. Imholz BP, Wieling W, Langewouters GJ, van Montfrans GA: Continuous finger arterial pressure: utility in the cardiovascular laboratory. Clin Auton Res 1: 43–53, 1991.
25. Kurki TS, Sanford TJ Jr, Smith NT, Dec-Silver H, Head N: Effects of radial artery cannulation on the function of finger blood pressure and pulse oximeter monitors. Anesthesiology 69: 778–782, 1988.
26. Nielsen PE, Bell G, Lassen NA: Strain gauge studies of distal blood pressure in normal subjects and in patients with peripheral arterial disease. Analysis of normal variation and reproducibility and comparison to intraarterial measurements. Scand J Clin Lab Invest
128: 103–109, 1973.
27. Kristensen BO, Kornerup HJ: Which arm to measure the blood pressure? Acta Med Scand 670: 69–73, 1982.
28. O’Shea JC, Murphy MB: Ambulatory blood pressure monitoring: which arm? J Hum Hypertens 14: 227–230, 2000.