tracking algorithm☆
Falk von Dincklage⁎,1, Mark H
Jan Harald Baars, Benno Rehb
Charité—Universitätsmedizin Berlin—Departm
B R A I N R E S E A R C H 1 2 6 0 ( 2 0 0 9 ) 2 4 – 2 9
ava i l ab l e a t www.sc i enced i r ec t . com
ominterchangeable conclusions drawn from data obtained with both methods.
© 2009 Elsevier B.V. All rights reserved.
1. Introduction
The nociceptive flexion reflex (NFR) is a polysynaptic spinal
activity is monitored using an electromyogram (EMG) during
the application of electrocutaneous stimuli to the ipsilateral
sural nerve. Based on the observed EMG response, thealgorithm: SD = 4.44 mA). Discussion: The continual algorithm can be used for a continuous
estimation of the reflex threshold at the 50% probability of reflex-occurrence. No evidence of
habituation was detected. This allows for investigations of the time courses of
pharmacological and physiological influences on the reflex threshold by using this
algorithm. The lower variability between the continual algorithm and the standard
algorithm compared to the variability between tests and retests of the methods allows forPain measurementwithdrawal reflex that is elicited thro
nociceptive A-delta afferents. To assess
nociception-specific RIII component, bice
☆ Supported by Deutsche Forschungsgemei
⁎ Corresponding author. Charité—Universitäts
10117 Berlin, Germany. Fax: +49 30 450 53192
E-mail address: falk.von-dincklage@chari
1 Both authors contributed to this publicati
0006-8993/$ – see front matter © 2009 Elsevi
doi:10.1016/j.brainres.2009.01.001determined by the different algorithms (test: SD = 2.50 mA, retest: SD = 1.80 mA) was lower
than the variability between test and retest (standard algorithm: SD = 4.32 mA, continualNociceptive flexion reflex
NFR
Threshold tracking
Reflex habituationackbarth1, Martin Schneider,
erg
ent of Anesthesiology, Germany
A B S T R A C T
Introduction: The RIII reflex is used in fundamental and clinical pain research. Here we
introduce a continual reflex threshold tracking algorithm to facilitate investigations of the
time courses of influences on the threshold. Methods: First we investigated the probability
of reflex occurrence at the threshold estimated by the continual algorithm and the changes
of the threshold over the time during continual recordings of 100 min duration in 10
subjects. Secondly we compared the threshold estimates of the continual algorithm with
those of a standard algorithm of threshold estimation in 52 subjects and compared the
differences between the two methods with the test–retest-variability of each method.
Results: The average probability of reflex occurrence at the threshold estimated by the
continual algorithmwas 48.7% (SD = 3.2%). Changes of the RIII reflex threshold over the time
were not significant (Friedman test, pN0.05). The variability between the thresholdsA R T I C L E I N F O
Article history:
Accepted 5 January 2009
Available online 10 January 2009
Keywords:
RIII reflex
RIII reflex thresholdIntroduction of a continual RIII reflex thresholdResearch Report
www.e l sev i e r. cugh activation of
the reflex and the
ps femoris muscle
nschaft, Bonn, Germany;
medizin Berlin Klinik für
7.
te.de (F. von Dincklage).
on equally.
er B.V. All rights reserved/ l oca te /b ra in resstimulus intensity required to elicit the reflex is used as an
objective measure of the individual nociceptive threshold
(Willer 1977; Chan and Dallaire 1989). Since the method of
grant numbers RE1534/2-4 and BA2686/2.
Anästhesiologie und Intensivmedizin, CampusMitte Charitéplatz 1
.

McIntyre et al., 2007; Emery et al., 2008; Ring et al., 2008) as well
2.3. Variability between the thresholds estimated by the
continual algorithm and the thresholds estimated by the
standard algorithm compared to the test–retest-variability
of each algorithm
For the second part of the study, all 52 subjects completed the
first session (Test), but only 35 subjects participated in the
second session (Retest), mainly due to scheduling problems.
The groups for Test and Retest did not differ in any of the
parameters age, height, weight, BMI or Trait-STAI scores
(student t-tests with Bonferroni corrections for each para-
meter, pN0.3). Also, the State-STAI scores for Test and Retest
did not differ for the individual subjects (repeated measures
student t-test, pN0.05).
The average reflex threshold (data for Test and Retest
condition combined) for the standard algorithm amounted to
9.18 mA (SE=0.89 mA) for the male and to 7.23 mA
(SE=0.73 mA) for the female subjects. For the continual
algorithm the average reflex thresholds amounted to
8.99 mA (SE=0.91 mA) for the male and to 7.76 mA
(SE=0.82 mA) for the female subjects. For both algorithms,
the difference of the thresholds between male and female
subjects was statistically not significant (student t-test with
Bonferroni corrections for each, pN0.05).
As a result of the Bland–Altman-analyses to compare the
1 2as individual differences in nociceptive processing in partici-
pants with or without pain disorders (Willer et al., 1986; Dahl
et al., 1992; Sandrini et al., 1993; Desmeules et al., 2003;
Bouhassira et al., 2003; Banic et al., 2004; Campbell et al., 2008).
Especially under conditions where subjective ratings of
nociception cannot be obtained, such as during anesthesia,
an objective measure such as the nociceptive flexion reflex is
of particular value.
However, up to now the estimation of the reflex threshold
is usually limited to singular events, following differing
algorithms (Sandrini et al., 2005). Here we propose a reflex
threshold tracking algorithm based on the objective criteria
suggested by Rhudy (Rhudy and France 2007) that allows for a
continual estimation of the reflex threshold, facilitating
investigations of the time course of threshold changes for
pharmacokinetic–pharmacodynamic modelling or other time
dependent mechanisms.
In the first part of this study we assessed the reliability of
the threshold as estimated by the continual algorithm, at the
50% probability of occurrence of the RIII reflex. This was done
for serial measurements over a period of 100 min. During
these, we also investigated the stability of the reflex threshold
over the time to investigate a possible drift of the RIII reflex
threshold due to a possible habituation or wind-up of the RIII
reflex. Both effects have been reported for the RIII reflex at
different stimulation frequencies (Arendt-Nielsen et al., 1994;
Dimitrijevic and Nathan 1970). A systematic drift of the RIII
reflex threshold due to interaction of the continually applied
stimuli would not allow a useful continual threshold
estimation.
In the second part of the study we compared the threshold
estimates of the continual algorithm with estimates obtained
by application of a standard algorithm of threshold estimation
and compared the variability between the two algorithmswith
the test–retest-variability of each algorithm.
2. Results
2.1. Probability of RIII reflex occurrence at the threshold
estimated by the continual algorithm
For the first part of the study, all 10 subjects completed the
experimental procedure successfully. The median number of
control stimuli at the RIII reflex threshold estimated by the
continual algorithm amounted to 50 for all subjects (range: 39–
54). The average probability of reflex occurrence for the controlinvestigation has been standardized (Skljarevski and Rama-
dan 2002; Sandrini et al., 2005), the NFR has been used
increasingly in fundamental and clinical pain research to
investigate pharmacological modulation of nociception
(Willer 1985; Piletta et al., 1990; Sandrini et al., 1992; Arendt-
Nielsen et al., 1995; Remy-Neris et al., 1999; Paradiso et al.,
2002; Bossard et al., 2002; France et al., 2007; Escher et al., 2007;
Arnold et al., 2008), spinal and supraspinal influences on
nociception (Garcia-Larrea et al., 1993; Arendt-Nielsen et al.,
1994; Leroux et al., 1995; Rossi et al., 1996; Rhudy et al., 2007;
B R A I N R E S E A R C Hstimuli amounted to 48.7% (SD=3.2%, 95% confidence inter-
val=42.4%–54.9%) (Table 1).2.2. Stability of the RIII reflex threshold over time at a
constant level of distraction
The changes of the threshold over time for the 10 individuals
tested were not significant (Friedman test, pN0.05). The
absolute mean deviation of the threshold from its mean over
the time amounted to 1.04 mA±0.16 mA (mean±SE).
Table 1 – Individual probabilities of RIII reflex occurrence
at the estimated thresholds
Subject RIII occurrence at the
control stimuli
Total number of
control stimuli
Total
number
Percentage (%)
A (male) 25 50.0 50
B (male) 24 44.4 54
C (male) 25 50.0 50
D (male) 20 51.3 39
E (male) 22 45.8 48
F (female) 23 46.0 50
G (female) 26 52.0 50
H (female) 22 47.8 46
I (female) 27 54.0 50
K (female) 23 46.0 50
Mean 23.7 48.7 48.7
SD 2.1 3.2 3.9
Each individual that participated in the first part of the study
received approximately 600 stimuli during the course of 100min. At
randomized intervals of 10 to 14 stimuli, a control stimulus was
delivered at the estimated threshold. Here shown are the total
number of control stimuli for each individual and the total
numberandpercentageof control stimuli onwhichaRIII reflexoccurred.
256 0 ( 2 0 0 9 ) 2 4 – 2 9thresholds estimated by the standard algorithm with those
estimated by the continual algorithm, the average differences