Prediction of intradialytic hypotension based on oxygen saturation variations

Abstract

Hemodialysis sessions are often complicated by the occurrence of intradialytic hypotension (IDH), mainly due to the decrease of central blood volume. Blood oxygen content may play a role in hypotension onset and can reflect changes in cardiac output and tissue perfusion. Currently, there is an increasing interest in intradialytic monitoring, through the development of biofeedback based technologies aimed to IDH prevention. Blood Oxygen saturation (SO2) is a patient parameter easy to monitor during hemodialysis thanks to optical noninvasive sensors placed on blood line. The aim of this study was to analyze SO2 variations in relationship with IDH in ten hypotension prone patients with central venous vascular access. A higher decrease of SO2 was found in sessions complicated by hypotension, in comparison with stable sessions (p

Full text

Prediction of Intradialytic Hypotension Based
on Oxygen Saturation Variations
C Perazzini1, PG Bolasco2, L Corazza3, M Tramonti1, E Mancini4, A Santoro4, S Severi1,5
1Health Sciences and Technologies Interdepartmental Center for Industrial Research, University of
Bologna, Cesena, Italy
2Territorial Department of Nephrology and Dialysis, ASL 8, Cagliari, Italy
3Scientific Affairs Bellco srl, Mirandola, Italy
4Nephrology, Dialysis, Hypertension, Policlinico S.Orsola-Malpighi, Bologna, Italy
5Department of Electrical, Electronic and Information Engineering "Guglielmo Marconi",
University of Bologna, Cesena, Italy
Abstract
Hemodialysis sessions are often complicated by the
occurrence of intradialytic hypotension (IDH), mainly
due to the decrease of central blood volume. Blood
oxygen content may play a role in hypotension onset and
can reflect changes in cardiac output and tissue
perfusion. Currently, there is an increasing interest in
intradialytic monitoring, through the development of
biofeedback based technologies aimed to IDH prevention.
Blood Oxygen saturation (SO2) is a patient parameter
easy to monitor during hemodialysis thanks to optical
noninvasive sensors placed on blood line. The aim of this
study was to analyze SO2 variations in relationship with
IDH in ten hypotension prone patients with central
venous vascular access. A higher decrease of SO2 was
found in sessions complicated by hypotension, in
comparison with stable sessions (p<0.001). The existence
of predictive information in SO2 variations in relationship
with IDH was evidenced through ROC analysis
(AUC=0.631, CI95% 0.564-0.690).
Our results suggest that SO2 decrease can
characterize sessions complicated with hypotension and
monitoring SO2 could be useful in clinical practice to
prevent IDH occurrence.
1. Introduction
Hemodialysis (HD) is a well-established treatment of
end stage renal disease (ESRD) patients, able to improve
the living conditions of patients, but it is also associated
with critical side effects. Intradialytic hypotension (IDH)
is the most important and frequent short-term
complication of HD, and affects twenty to fifty percent of
patients with a 20% incidence of all treatment [1,2]. IDH
is the critical consequence of multifactorial causes, but
mainly it is due to the decrease of central blood volume
resulting from the insufficient refilling of fluid from the
interstitial to the vascular compartment. Furthermore,
IDH presents a series of symptoms, as dizziness,
weakness, nausea, cramps etc. that complicate dialysis
sessions with the risk of limiting effectiveness of
treatment and determining nursing interventions.
The evolution of dialysis therapy is increasingly direct
towards the development of technologies answering to the
specific needs of each patient during dialysis sessions
with the possibility of IDH prediction and/or prevention.
In this context biofeedback system seems able to adapt
dialytic therapy to patients status starting from on-line
monitoring of patient parameters, such as relative blood
volume [3].
Blood Oxygen saturation (SO2) is a patient parameter
that can be continuously and non invasively monitored
during dialysis through optical sensors placed on blood
lines. A relationship between SO2 variations and
hemodynamic instability is documented in literature [4-6]
and it is based on the assumption that blood Oxygen
content variations can reflect changes in cardiac output
and tissue perfusion in agreement with Fick’s principle.
The aim of the study was to analyze SO2 variations in
relationship with hypotension occurrence in dialysis
patients with central vascular access.
2. Methods
2.1. Patients
Ten ESRD patients in renal replacement therapy with
thrice-weekly standard bicarbonate haemodialysis
afferent to different Italian dialysis centres were selected.
Patients were eligible if aged between 18 and 85 years old

and after provided informed consent, were monitored
along three months without changes of the usual
prescription.
Only subjects with central vascular access were
selected, and included in the study if hypotension prone,
meaning that they were affected by acute intradialytic
hypotensive episodes in at least 20% of the last month
sessions before the study started.
IDH was defined according the following criteria [7]:
1) if predialysis systolic arterial pressure (SAP) is greater
than 100 mmHg, then any episode with SAP less than 90
mmHg, even without complaints; 2) if predialysis SAP is
less than 100 mmHg, then any SAP reduction by at least
10% associated with complaints; 3) any SAP reduction of
25% or more of the predialysis value with the typical
symptoms requiring specific intervention.
2.2. Hemodialysis equipment and data
measurements
All patients were treated with the dialysis monitor
Formula Therapy (Bellco srl, Italy) equipped with the
optical sensor Hemox (Datamed srl, Italy) placed on the
arterial blood line. SO2 was acquired during the treatment
with a frequency of 1 sample per 5 seconds (fc = 0.2 Hz)
by means of Hemox.
Systolic and diastolic arterial blood pressures were
measured in pre-dialysis phase and every 30 minutes
during the treatment by an automatic oscillometric
sphygmomanometer (SPHYGMO, Bellco srl, Italy).
The occurrence of typical low blood pressure
symptoms (muscular cramps, headache, dizziness,
vomiting, nausea, sweating) and time of appearance were
also recorded, as well as every fluid infusion (saline,
plasma expanders, sodium).
All data downloaded by the monitors were stored and
analyzed in a personal computer (HP Pavilion dv6000)
equipped with Matlab 7.8 and NCSS 2007.
2.3. Data analysis
Dialysis sessions were classified according to the
presence or absence of hypotensive event (Hypo or
NoHypo, respectively).
To analyze temporal variation of SO2, mean values for
30-min interval were calculated. A delta value of Oxygen
saturation (∆SO2) was calculated as difference between
the SO2 mean value in the last 30 minutes before the end
of treatment (in NoHypo sessions), or before a
hypotensive episode (in Hypo sessions) and SO2 mean
value in the first 30 minutes.
According to Cordtz et al [4], a preliminary
investigation of SO2 variation predictive power in
relationship with IDH occurrence, was done considering
the ∆SO2 corresponding to the first 150 minutes (the
average time at which IDH occurs [8]). Then, a more
rigorous analysis was carried out setting a 30-min length
temporal window, 1 minute shifting, from the beginning
of the session to the hypotensive episode in Hypo
sessions or until 30 minutes before the end of session in
NoHypo sessions.
2.4. Statistics
Values are expressed as mean ± standard deviation
(SD) or standard error (SE).
Intergroup comparison were made using the unpaired t
test, while a paired t test was performed for in intragroup
comparisons. ROC curves analysis was run to identify a
threshold leading to a good compromise between
sensitivity and specificity. Statistical analysis was
performed with NCSS 2007 and Matlab Statistic
Toolbox.
3. Results
The sessions analyzed were 365: 124 sessions were
classified as Hypo due to the presence of a hypotensive
event and 241 were classified as NoHypo. In Figure 1 are
shown systolic and diastolic blood pressure (top panels)
and SO2 signal (bottom panels) of exemplificative
sessions with and without hypotension (left and right
panels, respectively).
Figure 1. Blood pressure and SO2 signal during
exemplificative Hypo and NoHypo session. The red
arrow indicates the time of hypotension occurrence. In
SO2 panels the dashed line is the mean value of SO2 in the
first 30 minutes.
Table 1 summarize treatment characteristics of Hypo
and NoHypo sessions: only postdialysis systolic and

diastolic blood pressure were found significantly different
(p<0.001).
Table 1. Values obtained during sessions with or without
hypotensive event, Hypo or NoHypo, respectively.
Treatment
characteristics
Hypo
NoHypo
p
Systolic BP (mmHg)
Predialysis
Postdialysis
Diastolic BP (mmHg)
Predialysis
Postdialysis
Duration (min)
SO2, first 30 min (%)
126 ± 40
110 ± 29
68 ± 18
61 ± 17
217 ± 24
60 ± 10
127 ± 27
132 ± 29
70 ± 14
73 ± 14
222 ± 24
60 ±11
ns
0.001
ns
0.001
ns
ns
Values are mean ± SD
BP = blood pressure; ns = nonsignificant; SO2 = Oxygen
saturation
In Table 2 are reported delta blood pressure values
measured at the end of treatment (NoHypo sessions) or
before a hypotensive event (Hypo sessions) and ∆SO2
from the last 30 minutes of treatment or before the
hypotensive event. The different distribution of ∆SO2 in
Hypo and NoHypo sessions is shown in Figure 2.
Results of ROC analysis are shown in Figure 3.
Dashed grey curve was obtained considering ∆SO2 during
the first 150 minutes: the AUC was 0.720 (CI95% 0.657-
0.773). Setting a threshold value, e.g. equal to -4.77,
Sensitivity reached 60% (CI95% 0.508-0.682) and
Specificity was 75% (CI95% 0.691-0.802). Solid blue
curve was obtained through ∆SO2 computed by a 30-min-
length shifting window: AUC was 0.631 (CI95% 0.564-
0.690); choosing a threshold value equal to -7.80
Sensitivity was 62% (CI95% 0.537-0.709) and Specificity
was 60% (CI95% 0.535-0.659).
4. Discussion and conclusions
In our study, we have found that sessions complicated
with hypotensive episode are characterized by a higher
decrease of SO2 with respect to hemodynamically stable
sessions. This result is in agreement with what
demonstrated by Cordtz et al [4]. In fact they found that
hypotension prone patients exhibited a SO2 decrease
during dialysis session, while this decline was not
observed in hypotension resistant patients. The most
interesting and innovative aspect of our study, is that
patients selected were all affected by acute intradialytic
hypotensive episodes in at least 20% of sessions. Even in
NoHypo sessions we found a negative mean ∆SO2 but
with an absolute value significantly minor than ∆SO2 of
Hypo sessions (p<0.001).
Based on these results, distinguishing Hypo e NoHypo
sessions could be realistic and more relevant than
Table 2. Mean values measured at the end of treatment or
before a hypotensive event.
Hypo
NoHypo
p
∆BPsys (mmHg)
-46.7 ± 2.8‡
5.8 ± 1.6
0.001
∆BPdias (mmHg)
-15.5 ± 1.4‡
3.0 ± 0.9
0.001
∆SO2 (%)
-7.5 ± 0.6‡
-3.6 ± 0.4
0.001
Values are mean ± SE
‡ Value significantly different (p<0.001) from the predialysis BP
value and SO2 in the first 30 minutes
∆BP = changes in systolic and diastolic blood pressure; ns =
nonsignificant; ∆SO2 = changes of Oxygen saturation
Figure 2. ∆SO2 distribution in Hypo and NoHypo
sessions.
Figure 3. ROC curves obtained considering ∆SO2 during
the first 150 minutes (dashed grey curve) and ∆SO2
computed in temporal window of 30-min length, 1 minute
shifting, from the beginning of the session to the
hypotensive episode in Hypo sessions or until 30 minutes
before the end of session in NoHypo sessions (solid blue
curve). Red circles indicate exemplificative threshold
values.

recognizing HP and HR patients as shown by the ROC
analysis.
In conclusion, our study suggests that SO2 decrease
can reflect hemodynamic instability and characterize
sessions complicated by hypotension. Monitoring SO2
variations could be useful in clinical practice in order to
prevent IDH occurrence. Larger studies are needed to
evaluate and plan the possible use of SO2 also in
combination with other parameter to potentiate the
effectiveness of intradialytic monitoring.
Acknowledgements
This work was supported by the grants of the project
Optimization of dialysis with artificial kidney (POR-
FESR - Regione Emilia Romagna).
References
[1] Leunissen KM, Kooman JP, van der Sande FM, van Kuijk
WH. Hypotension and ultrafiltration physiology in
dialysis. Blood Purif. 2000; 18:251–254.
[2] Zucchelli P, Santoro A. Dialysis-induced hypotension: a
fresh look at pathophysiology. Blood Purif. 1993;11(2):85-
98.
[3] Santoro A, Mancini E, Basile C, Amoroso L, Di Giulio S,
Usberti M, Colasanti G, Verzetti G, Rocco A, Imbasciati E,
Panzetta G, Bolzani R, Grandi F, Polacchini M. Blood
volume controlled hemodialysis in hypotension-prone
patients: a randomized, multicenter controlled trial. Kidney
Int. 2002 Sep;62(3):1034-45.
[4] Cordtz J, Solem K, Ladefoged SD, Central venous oxygen
saturation and thoracic admittance during dialysis: New
approaches to hemodynamic monitoring, Hemodial. Int.,
2008 vol. 12, 369–377.
[5] Mancini E, Corazza L, Cannarile DC, Soverini ML,
Cavalcanti S, Cavani S, Fiorenzi A, Santoro A, Short term
variability of oxygen saturation during hemodialysis is a
warning parameter for hypotension appearance, in Proc.
Comput. Cardiol., 2008, vol. 35, 881–883.
[6] Solem K, Olde B, Sörnmo. Prediction of intradialytic
hypotension using photoplethysmography. L. IEEE Trans
Biomed Eng. 2010 Jul;57(7): 1611-9.
[7] Sörnmo L, Sandberg F, Gil E, Solem K. Noninvasive
techniques for prevention of intradialytic hypotension.IEEE
Rev Biomed Eng. 2012;5:45-59.
[8] Nette RW, van den Dorpel MA, Krepel HP. Hypotension
during hemodialysis results from an impairment of
arteriolar tone and left ventricular function. Clin Nephrol.
2005; 63:276–283.
Address for correspondence.
Claudia Perazzini
Health Sciences and Technologies Interdepartmental Center for
Industrial Research
University of Bologna - Cesena
Via Venezia 52, 47521 Cesena (Italy)
claudia.perazzini2@unibo.it