Incremental Bias in Finapres Estimation of Baseline Blood Pressure Levels Over Time
Abstract Finapres finger blood pressure monitoring appears to provide a reliable alternative to intra-arterial blood pressure measurement under many circumstances. However, few studies have focused on the limitations of Finapres assessment. In a previous pilot investigation, we observed that Finapres pressure following mental stressors failed to return to initial resting levels. Our objectives in the present study were to (1) replicate earlier findings, (2) examine whether local changes in the measured finger were responsible for the observed drift, and (3) test a method to facilitate the return of pressure to systemic baseline levels. We studied two groups of healthy subjects who underwent a protocol consisting of two mental stressors preceded and followed by baseline periods. In the control group, the Finapres continuously monitored pressure on a single finger for the entire protocol. The intervention group periodically had the Finapres cuff removed and the measured finger exercised to prevent local changes that might influence Finapres estimation of blood pressure. Comparisons indicated a group×baseline interaction effect for systolic and diastolic pressures (P<.0004 and P<.003, respectively). The group with the exercise intervention showed much greater recovery during the final baseline than the control group. Recovery of pressures in the control group but not the intervention group was inversely related to the stress level of blood pressure (r=.86, P<.0002), indicating a relationship between blood pressure rise and the degree of distortion of subsequent baseline values. On the basis of our results, we propose that in prolonged protocols, the measurement finger be exercised to facilitate accurate measurements of finger pressure with the Finapres.
The Finapres blood pressure (BP)–monitoring device provides continuous, noninvasive beat-by-beat measurements of BP. For researchers investigating the mechanisms underlying hypertension, the Finapres is a useful alternative to intra-arterial measurements, as a strong correspondence between Finapres and intra-arterial measures has been documented in different populations.1 2 3 4 5 6 Because the Finapres is noninvasive, it can be used over a range of conditions for which intra-arterial measurement would be inappropriate. An ambulatory version of the Finapres, the Portapres, additionally permits BP monitoring in the natural environment.7 Analysis of the Finapres and Portapres signals can also be used to derive valid estimations of cardiac output, total peripheral resistance (TPR), baroreflex sensitivity, and sympathetic indexes of vasomotor tone.8 9 10 11
The versatility of Finapres measurement has led to widespread applications of the method in basic and applied research studies. However, very few investigations have focused on the constraints of Finapres monitoring. In this article, we note one problem we have encountered with Finapres BP monitoring that may be significant, and we provide a possible solution. Although the Finapres accurately reflects arterial BP under various conditions, it is unclear whether the Finapres reliably tracks systemic BP during protocols in which BP is elevated for some time. Preliminary findings from our laboratory indicated that in protocols 3 to 4 hours long, Finapres pressure levels following mental stressors failed to return to initial resting levels. Problems of this nature, should they reflect a real artifact, could lead to gross misestimation not only of BP levels but also of other measures derived from the Finapres signal, including stroke volume, cardiac output, and TPR.
We first observed this phenomenon in a group of 46 individuals with coronary artery disease participating in an ongoing study involving two mental stressors preceded and followed by baseline resting periods. In the baselines following both tasks, Finapres pressure failed to recover to initial baseline levels and in fact increased progressively with each subsequent baseline period. We hypothesized that a hemodynamic phenomenon localized in the instrumented finger prevented BP from recovering to systemic levels and therefore began an investigation to explore this phenomenon systematically. This article presents the results of our subsequent investigation. We also propose a method to promote recovery of Finapres pressure so that it corresponds to recovery of systemic pressure.
The study protocol was approved by the Western Institutional Review Board, and the experiment was conducted in accordance with their guidelines. All subjects gave informed consent before participation. The experimental protocol lasted 3 to 4 hours and consisted of an initial 10-minute baseline, 12 minutes of paced breathing (for autonomic assessment), 10 minutes of mental arithmetic followed by a second 10-minute baseline, and a 7-minute speech stressor followed by a third 10-minute baseline (Fig 1⇓). There were several pauses between phases, used for completion of psychological questionnaires, instructions, and training. During each phase of the protocol, we collected a one-lead electrocardiogram and finger BP measurement. Arm pressures were taken at the beginning of the protocol with a digital BP monitor (model HEM-705CP, Omron Healthcare).
We used the Ohmeda 2300 Finapres (series FAX) to measure beat-to-beat BP for the duration of the protocol. There was no drift in the baseline electrical signal over 4 hours. At the beginning of the study, the Finapres cuff was placed on the middle finger of the subject’s nondominant hand. Occasionally, we needed to reposition the finger cuff when the morphology of the BP wave was not optimal. Repositioning the cuff involved turning off the Finapres, removing the cuff, placing it either on the same finger or on an adjacent finger, and restarting the Finapres. Once we had an acceptable BP signal from a particular finger, the Finapres remained on that finger only. In all cases, the Finapres was allowed to run continuously for 5 minutes before data acquisition was begun.
During the protocol, the subject was seated comfortably in an easy chair with the Finapres hand resting on a firm pillow placed on the arm of the chair. This posture typically elevated the subject’s measurement finger approximately to heart level. Since we were interested in relative changes in BP, not absolute levels, the precise location of the finger was not essential as long as its vertical displacement from the heart was constant for the duration of the study. We noted periodically the position of the subject’s hand to verify that it remained at the original level.
Our major comparisons were between two groups who underwent the above protocol. One group (control) completed the entire protocol with no special intervention. The other group underwent the same protocol with an intervention periodically used to enhance finger circulation. The two groups were similar with respect to age, body mass index, systolic and diastolic finger pressures for baseline 1, RR interval for baseline 1, and arm systolic and diastolic pressures before the protocol (Table⇓). Data presented are means±SE throughout this report. All subjects were healthy and not on medications as determined by physical examination or medical history provided by the subject’s primary care physician.
For the no-intervention, control group (n=13), the Finapres cuff was placed on the finger at the beginning of the experiment and remained on the finger for the duration of the protocol. At the end of the paced breathing, the Finapres was turned off for approximately 5 minutes while the finger cuff remained on the subject’s hand.
For the intervention group (n=16), we systematically removed the finger cuff and periodically exercised the subjects’ fingers in a controlled manner over the course of each experiment. Our objectives were (1) to prevent local vascular or fluid changes that might distort Finapres estimation of systemic pressure, and (2) to examine whether the intervention would facilitate the return of finger pressure to initial baseline levels. The procedure involved stopping the Finapres and removing the finger cuff four times during the protocol. The cuff remained off of the finger while the subject alternately clenched and relaxed the hand 10 times, completed a psychological questionnaire, and then clenched and relaxed the hand 10 more times. The only differences between the control and intervention protocols were the finger exercises, among intervention subjects, before and after the psychological questionnaire, along with brief instruction describing the exercise technique. After the finger was exercised, we replaced the cuff, using the infrared LED and photocell indentations on the finger to guide the cuff back into its original position, and restarted the Finapres. Additionally, cuff tightness was carefully maintained at the same level throughout each session. This exercise procedure was integrated into the protocol so as to add only 30 seconds to 1 minute of additional time per intervention, thus negligibly lengthening the entire protocol for the intervention group by only about 3 minutes compared with the control group.
In additional analyses, we tested a third group of six healthy subjects in the same protocol. For this group, we used a Portapres instrument (TNO Instrumentation) to measure BP. The Portapres is a portable BP-measuring device that uses the same principles and algorithms as the Finapres. It includes two measurement channels for two different fingers and allows the experimenter to switch between the two fingers from a remote location. This finger-switching option permitted us to examine whether lack of recovery was specific to the finger that was continuously measured or generalized to other (unloaded) fingers of the same hand. Use of the Portapres also allowed us to demonstrate that the apparent drift phenomenon we observed was not merely instrument specific. Therefore, at the beginning of the study, we placed two BP finger cuffs on the subject’s middle and ring fingers of the nondominant hand. If the pressures in the two fingers were more than 10 mm Hg apart, we adjusted and repositioned the cuffs in the manner described for the Finapres until we had two fingers whose pressures were similar. The subjects were seated in an easy chair with the Portapres arm resting on a firm pillow placed on the arm of the chair. We measured on one finger for the entire study through minute 5 of baseline 3, when we switched to the other, unloaded finger of the same hand. Automatic finger switching during baseline 3 resulted in the loss of 2 minutes of data. Because all previous data indicated only a modest and very gradual decline in BP across these minutes, we assumed that an abrupt, large reduction in BP with finger switching would indicate that specific local factors probably accounted for the upward drift of baseline BP we describe.
An analog-to-digital converter sampled the Finapres analog signal at 200 Hz. Customized computer programs written in Matlab programming language (Mathworks) removed artifacts and estimated systolic and diastolic pressures on a beat-by-beat basis. All data were carefully controlled for artifacts by interactively comparing systolic and diastolic values with peaks and troughs of the actual BP waves for every subject. Additional verification of the accuracy of the BP waveforms for typical examples of baseline drift was made by a separate laboratory (personal communication, K. Wesseling, TNO Instrumentation).
One-minute BP averages were calculated for all phases of the protocol. We also calculated cardiac output and TPR from the BP waveform using the FAST pulse-contour analysis software (TNO Instrumentation).8 9 Correlation analyses and repeated measures MANOVA were performed to compare the two Finapres groups. ANCOVA was used when appropriate to control for reactions to stressors. Statistical significance was set at a value of P<.05.
When we initially detected the Finapres drift in our previous study of individuals with coronary artery disease, finger systolic pressure remained elevated by 20 mm Hg from minute 10 of baseline 1 to minute 10 of baseline 3 (P<.00001 for systolic and P<.00001 for diastolic, across baselines; Fig 2⇓, top). We observed a similar trend in the control group of the present study tested with the Finapres. For this group, Finapres systolic pressure was elevated by 26 mm Hg from minute 10 of baseline 1 to minute 10 of baseline 3 (P<.0001, across baselines; Fig 2⇓, bottom). Finapres diastolic pressure also became progressively elevated across baselines (P<.002). Unpublished observations from our laboratory (1996) suggest that participants in this experimental protocol show increases in systolic and diastolic arm pressures of only 6 and 3 mm Hg, respectively, from the start of baseline 1 to the end of baseline 3 (n=82).
In the group tested with the Portapres, systolic pressure measured on the initial finger for baselines 1 and 2 and the first half of baseline 3 showed the same upward trend seen with the Finapres subjects. In contrast, switching to the unloaded finger during the last 3 minutes of baseline 3 was associated with much lower systolic pressure, approaching initial baseline levels (Fig 3⇓). Diastolic pressure exhibited a similar trend. Both systolic and diastolic pressures differed between the loaded finger (minute 5, baseline 3) and unloaded finger (minute 8, baseline 3; P<.002 for systolic and P<.02 for diastolic). In comparison, the same minutes did not differ from each other in either Finapres control or exercise groups. Therefore, the sudden decrease in BP immediately after finger switching supported our hypothesis that local changes in the loaded finger were occurring during the protocol and were responsible for the lack of recovery we had seen previously in both Finapres and Portapres subjects.
Results of the Finapres comparisons indicated a group×baseline interaction effect (P<.0004 and P<.003 for systolic and diastolic pressures, respectively; Fig 4⇓). The Finapres group with exercised fingers showed much greater recovery during baseline 3 than the unexercised group (Fig 5⇓). We found that finger systolic and diastolic pressures for the intervention group recovered to within 7 and 5 mm Hg, respectively. Comparable values for the control group were 30 and 18 mm Hg, respectively. More specifically, all subjects in the control group showed an increase in systolic BP of greater than 14 mm Hg from baseline 1 to baseline 3, with the range extending to 64 mm Hg; 64% of subjects in the control group showed an increase of greater than 10 mm Hg from baseline 1 to baseline 2. In contrast, only 25% of subjects in the intervention group showed systolic BP increases of greater than 14 mm Hg from baseline 1 to baseline 3 (range, −14 to 29 mm Hg), and only 7% showed an increase of greater than 10 mm Hg from baseline 1 to baseline 2.
Because intervention and control groups varied in BP responses to mental stressors, we used ANCOVA to determine whether group×baseline interaction effects would remain significant after controlling for differences in stress reaction (defined as mean pressure level during stress minus mean pressure level during the preceding baseline). After reactions to stressors were included as covariates, the two groups still differed in BP elevation from baseline 1 to baseline 3 (P<.0001 and P<.02, systolic and diastolic, respectively). These findings provide consistent evidence that our intervention of exercising the finger significantly influenced the degree of Finapres recovery.
We also found a high correlation between absolute systolic pressure levels during speech stress and the extent of systolic BP recovery in the control group (r=.86, P<.0002) and the coronary artery disease group (r=.58, P<.001) but no correlation between these two variables for the intervention group (r=.07, P<.80). These findings indicate that increased pressure in the finger is related to the lack of recovery and that exercising the finger decreases the pressure buildup, thus promoting recovery of finger pressure concordant with the recovery of systemic pressure.
Finally, we examined the effects of the Finapres drift on cardiac output and TPR, as artifactual drift may influence Finapres pulse-contour–derived measures. A group×baseline comparison revealed that the control and intervention groups did not differ in cardiac output across baselines. However, we found a group×baseline interaction effect for TPR (P<.03), with the control group showing a greater increase than the intervention group in TPR from baseline 1 to baseline 3. This finding suggests that pulse-contour estimates of TPR recovery may be biased when BP drift occurs.
This study has shown that in relatively lengthy protocols involving mental stress–elicited increases in BP, the Finapres device does not provide accurate estimates of posttask BP when the measurement finger is not periodically relieved of the pressure load. Discrepancies in recovery found between intervention and control groups could not be explained by differences in systemic BP arising from the very brief and occasional interruptions of the intervention group during finger exercise (30 to 60 seconds, four times). This procedure added only about 3 minutes to the entire protocol of several hours. Figs 4⇑ and 5⇑ document the minute-to-minute stability of baseline levels among groups, clearly indicating that a brief intervention should not have had any significant effect.
Additionally, the Portapres data on finger switching are consistent with the conclusion of a local phenomenon being responsible for the drift across baselines (see Fig 3⇑). During the last half of the baseline period, merely switching measurement from the chronically loaded finger to an adjacent finger (already instrumented with a previously uninflated cuff) produced a rapid and profound lowering of BP.
Intra-arterial comparisons with Finapres data, although not possible in the present study, would provide the best evidence that the observed drift is artifactual. However, the close correspondence between brachial arm cuff recovery in a large, independent sample (6 and 3 mm Hg, systolic and diastolic pressures, respectively; n=82) and Finapres recovery for the intervention group (7 and 5 mm Hg, systolic and diastolic pressures, respectively) strengthens the evidence that the Finapres drift does not represent true systemic pressure changes.
Drift occurred for both Finapres and Portapres systems, indicating that this phenomenon is not instrument specific. However, differences in the magnitude of mean stress responses derived from Portapres and Finapres instruments may suggest that the two instruments provide dissimilar BP values, although the small sample size of the Portapres group precludes definitive conclusions.
Our data therefore suggest that peripheral changes at the finger caused by constant cuff pressure are responsible for artifactual elevations of resting BP after prolonged measurement. Future studies of repeated measurements across sessions may determine whether individual differences in drift are reproducible, thus biasing measurement for some subjects (and possibly clinical groups) more than for others. Possible mechanisms influencing drift may include (1) interstitial fluid loss from the finger tissue under the cuff, (2) altered blood flow, (3) edema, and/or (4) changes secondary to altered finger temperature at the measurement site. It is beyond the scope of the present investigation to determine the cause. Other lengthy studies comparing the Finapres with intra-arterial BP measurement make no mention of this lack of recovery.4 However, these studies focus on the correlation between the two methods with respect to absolute levels and variability within a given condition and therefore may not attend to issues concerning reactivity or recovery.
In studies concerned with BP reactivity, the stability of the baseline pressure levels is of particular importance, as the calculation of pressure change depends on an accurate baseline. Since we did not take arm pressure measurements during the mental stressors, we are uncertain about how this gradual increase in pressure over time might have affected Finapres levels during task. We expect that the drift from preceding baseline to task (baseline 2 to speech, for example) is small, as the two are close in time, and therefore reactivity can be evaluated presumably without excessive error. However, as we have shown, this problem of drifting baselines can be largely remedied by removing the Finapres cuff approximately every half hour and exercising the subject’s finger.
We recognize that conditions specific to our study may affect the drift observed with the Finapres. In many of the studies comparing the Finapres with intra-arterial pressure measurements, subjects are supine, whereas subjects were sitting in our study.4 This semiupright posture may contribute to the stepwise increase in successive baselines seen with the Finapres. It is also possible that our mental stressors retard Finapres recovery by eliciting greater changes in peripheral vascular resistance than the stressors used in other studies (eg, Valsalva maneuver, orthostatic maneuvers, and cold pressor test).1 2 3 4 5 In any case, we found that absolute systolic pressure level during speech stress was inversely related to the extent of systolic pressure recovery in the control group but not in the intervention group. This suggests that large stress-induced increases in BP may contribute to the lack of recovery.
Because our study groups included some older subjects, we initially questioned whether this phenomenon was age related. However, examination of a small group of young, healthy subjects who participated in the same protocol without finger exercises revealed that systolic pressure for this group was elevated by 20 mm Hg from baseline 1 to baseline 3. This evidence seems to argue against the possibility that lack of recovery was age related.
Although we devised a method to effect almost full recovery of Finapres pressure, we acknowledge potential limitations regarding accurate repositioning of the finger cuff. Nevertheless, we were exceedingly careful to reposition and accurately retighten the cuff each time it was replaced after finger exercise. We believe that the indentations created on the subject’s finger by the infrared LED and photocell provide an adequate marker by which to guide the cuff into its original position.
In conclusion, we consider the Finapres to be a valuable alternative to intra-arterial BP measurements in physiological studies. The Finapres provides a noninvasive method for measurement of beat-to-beat changes in BP, spontaneous measurements of baroreflex sensitivity, and sympathetic indexes of vasomotor tone that otherwise could only be measured intra-arterially.10 11 Nevertheless, care must be taken to evaluate possible upward drift in finger BP over time that appears to be local in nature and a result of the measurement itself. On the basis of our findings, we propose that studies of prolonged duration implement the finger-exercising method described in this article to facilitate accurate measurements of finger pressure made with the Finapres.
- Received March 4, 1996.
- Revision received April 25, 1996.
- Accepted October 23, 1996.
Bos W, Imholz B, van Goudoever J, Wesseling K, van Montfrans G. The reliability of noninvasive continuous finger blood pressure measurement in patients with both hypertension and vascular disease. Am J Hypertens. 1992;5:529-535.
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. 1990;24:214-221.
Rongen G, Bos W, Lenders J, van Montfrans G, van Lier H, van Goudoever J, Wesseling KH, Thien T. Comparison of intrabrachial and finger blood pressure in healthy elderly volunteers. Am J Hypertens. 1995;8:237-248.
Imholz BPM, Langewouters GJ, van Montfrans GA, Parati G, van Goudoever J, Wesseling KH, Wieling W, Mancia G. Feasibility of ambulatory, continuous 24-hour finger arterial pressure recording. Hypertension. 1993;21:65-73.
Wesseling KH, Jansen JR, Settels JJ, Schreuder JJ. Computation of aortic flow from pressure in humans using a nonlinear, three-element model. J Appl Physiol. 1993;74:2566-2573.
Stok WJ, Baisch F, Hillebrecht A, Schulz H, Meyer M, Karemaker J. Noninvasive cardiac output measurement by arterial pulse analysis compared with inert gas rebreathing. J Appl Physiol. 1993;74:2687-2693.
Watkins L, Grossman P, Sherwood A. Noninvasive assessment of baroreflex control in borderline hypertension: comparison with the phenylephrine method. Hypertension. 1996;28:238-243.