Acta Pædiatrica ISSN 0803-5253

REGULAR ARTICLE

COMFORT behaviour scale and skin conductance activity: what are they
really measuring?
Rosana Maria Trist~ao ([email protected]; [email protected])1, Naiara Viudes Martins Garcia1, Jose Alfredo Lacerda de Jesus1, Carlos Tomaz2
1.Faculty of Medicine, Campus Darcy Ribeiro, University of Brasilia, DF, Brazil
2.Institute of Biology, Campus Darcy Ribeiro, University of Brasilia, DF, Brazil

Keywords ABSTRACT
COMFORT behaviour scale, Neonate, Pain, Aim: To assess how efficiently the COMFORT behaviour scale measures acute pain in
Phenomenological approach, Skin conductance
neonates, in comparison with skin conductance activity, a validated measure of pain and
Correspondence stress.
R. Trist~ao, Area of Medicine for Children and
Teenagers, Faculty of Medicine, University of Brasılia, Methods: Images of 36 newborns were analysed before, during and after painful heel
Campus Darcy Ribeiro, DF 70910-900, Brazil. pricks to measure glucose levels and compared with skin conductance activity variables.
Tel:+55 (61) 31071879 |
Fax: +55 (61) 3107.1907 |
Results: Scale indicators and skin conductance variables were sensitive to changes in the
Emails: [email protected]; [email protected] periods ‘during-before’ and ‘during-after’ (Wilcoxon’s test, p < 0.01). Significant values
were found between all scale indicators and number of waves for Kendall’s coefficient
Received
25 February 2013; revised 5 June 2013; (p < 0.05), although responses differed when it came to how long they took to increase
accepted 13 June 2013. and correlations varied from fair to moderate (r < 0.6). Facial tension was more closely
DOI:10.1111/apa.12325 related to 15 sec after the painful event, while crying and calmness were more closely
related to the later intervals (30 and 180 sec).
Conclusion: All scale indicators were related to skin conductance activity in all periods,
indicating pain perception. Facial tension was the most efficient indicator, while others
varied in performance after painful events and possibly indicated stress after trauma. These
results are discussed from a phenomenological approach and in an anxiety paradigm.

INTRODUCTION define pain, especially when the subjects are unable to

Pain perception evaluation is fundamentally related to provide verbal feedback.
clinical skill and it relies on feedback from the patient, Several unidimensional or multidimensional scales have
behavioural observation and physiological measures based been introduced to assess pain in infants (4,5) including the
on physical examination. Although the claim that newborns COMFORT scale (6,7), a multidimensional scale used in
are able to perceive pain was controversial for many mechanically ventilated patients in ICU. The scale uses
decades, there is now enough evidence to show that they physiological measurements such as heart rate variability
are capable of perceiving and reporting pain. This capacity (HRV), blood pressure (BP) and oxygen saturation vari-
relates both to their medical condition and to the clinical ability (OSV). It also uses behavioural responses, including
procedures they undergo, such as blood tests, endotracheal alertness, calmness–agitation, respiratory response (only in
suction, surgery or other invasive procedures, especially in mechanically ventilated children), crying (only in children
an intensive care unit (ICU) (1). Pain is defined by the breathing spontaneously), physical movement, muscle tone
International Association for the Study of Pain (IASP) as and facial expression (6). Van Dijk adapted and modified
‘an unpleasant sensory and emotional experience associ- this scale, cutting out HRV, OSV and BP because they
ated with actual or potential tissue damage, or described in
terms of such damage’ (2). The IASP also considers that the
‘pain is always subjective and is always a psychological Key notes
state’ (3). These definitions have often been used in pain  Pain measurements in newborns involve physiological
studies with different theoretical models and subjects, but and behavioural assessment, although it is not always
their worldwide use cannot disguise how difficult it is to clear whether they are measuring pain or distress.
 This study compared two measurements, the COM-
FORT behaviour scale and skin conductance activity.
 The results will inform discussions on the weight of the
Abbreviations COMFORT scale’s behaviour categories and if they are
AUC, Area under the curve of waves; Cb, COMFORT behaviour measuring other dimensions of human behaviour, such
scale; NWps, Number of waves per second; SCA, Skin conduc-
as anxiety, instead of pain.
tance activity.

e402 ©2013 Foundation Acta Pædiatrica. Published by John Wiley & Sons Ltd 2013 102, pp. e402–e406
Trist~ao et al. COMFORT behavioural scale and skin conductance activity

depend on other factors such as weather conditions and (mean = 38.9;  1.4 weeks). All had been born at the
cardiorespiratory state (1,5,8–11), that may generate non- Maternity Unit of the University of Brasılia Hospital within
specific signs of stress, increasing ambiguity and hindering the previous 48 h. They all underwent routine heel pricks to
pain assessment (12). He created the COMFORT behaviour assess their glucose levels. Before the procedure took place,
(Cb) scale, which just uses the behavioural and phenome- all infants were kept in a comfortable thermal environment,
nological dimensions of the COMFORT scale (7). were kept clean, were fed and left undisturbed for at least
The phenomenological dimension is linked to the behavio- 1 h. Infants were given 25% oral glucose in line with the
ural dimension and consists of two subdimensions, medical protocol of the intensive care unit. Parents were
affective-emotional and sensory-discriminative. The affec- consulted and informed about the purpose of the study and
tive-emotional subdimension measures the subject’s spon- signed the consent form. This research was conducted in
taneous cognitive/conative reactions to personal pain accordance with Declaration of Helsinki principles and was
experiences conceived during the formation of conative approved by the Ethics Committee of the University of
propositional attitudes (13). The sensory-discriminative Brasilia. Clinical information about the newborn infants
aspect of pain is representational and covers tissue damage. was extracted from their medical records, including: type of
Defining the phenomenological aspect of pain is very anaesthesia during delivery, gestational age, Apgar 5th min,
difficult. Feeling pain involves perception, but that does need for resuscitation, postnatal age, neonatal disorders,
not fully describe its nature. It is also an affective/emotional previous painful procedures, breastfeeding or oral glucose
experience that can be explained in terms of the functional in the last hour before heel prick. Infants were excluded if
role of pain’s sensory/representational content. they had: a postnatal age of < 24 h, Apgar score of less than
The skin conductance activity (SCA) (14,15) has been seven, intracranial haemorrhage in the third or fourth
used as a faster and more efficient way of measuring pain in degree (19); metabolic, respiratory, circulatory, congenital
neonates, especially when they are under sedation or disorders; drugs use that interfered with pain perception;
anaesthetic, because researchers discovered that the concept opioid use during pregnancy except to half an hour before
of emotional sweating applied to newborns (5,8,16–18). the delivery or general anaesthesia.
Sweating occurs as a result of the response of the cerebral
cortex when the sympathetic nervous system (SNS) is Procedure
activated changing skin conductance by releasing acetyl- The neonates were submitted to a daily routine heel prick
choline at postganglionic synapses of the muscarinic for glucose monitoring. This was carried out in the morning,
receptors on the sweat glands. This occurs regardless of while they were in their cot and by the same examiner. They
cardio-respiratory conditions (1,9,17). A previous study has were videotaped (DCR-SR 47, Sony) and a trained observer
already compared the results obtained from SCA and the assessed the films, scoring muscle tone, facial expression,
COMFORT and Cb scales (17) during endotracheal suc- alertness, calmness/agitation, respiratory pattern and phys-
tioning. A better correlation was observed with the increase ical movement. Each of the six Cb behavioural indicators
in the COMFORT scale score than with variations in heart has a range of one to five points, with six points reflecting
rate and BP. One possible question arising from these results the lowest degree of discomfort and 30 points reflecting the
is whether some indicators of the Cb are more sensitive and highest level (7). SCA was measured by the skin conduc-
faster in revealing infant pain perception. If so, the Cb scale tance measurement system (SCMSâ, Medstorm Innova-
could be used in emergencies to indicate that further tion) through number of waves per second (NWps) and the
evaluation using other pain scales or SCA is necessary. area under the waves curve (AUC) (8,9). To capture the
We hypothesized that some of the Cb scale factors would SCA, three electrodes were attached to the infant’s left foot
be more efficient than others when it came to measuring a 10 min before the observation period started. SCA was
neonate’s response to pain. If so, this could prove useful for taken in three time periods: ‘before’ (3 min before the
future studies on acute pain in infants and for clinical and painful event), ‘during’ (3 min after the painful event) and
research purposes. To verify the efficiency of each indicator, ‘after’ (a further 3 min). This provided us with 9 min for
this study aimed to compare Cb scale indicators to SCA at classical analysis of pain perception, allowing us to verify if
three different intervals after painful events and to assess the infant had reacted to, and recovered from, the heel
whether they could be used independently as a pain prick. To fulfil the aim of this study, the ‘during’ period was
measurement predictor in newborns. We also aimed to subdivided and analysed at 15, 30 and 180 sec after the heel
analyse the performance of each factor over specific time prick. Both the SCA and Cb measurements took place at
periods to verify if they changed with time and whether the same time.
some responses were more related to the painful event than
others. Statistical analyses
Cb scores were compared between the two pair-related
sample periods (during-before and during-after) to verify
METHODS the infant’s reaction and recovery to the painful event. The
Participants Wilcoxon nonparametric-paired test was used to analyse
Thirty-six newborns were selected (19 males and 17 any significant differences between the scores over these
females) with gestational ages of between 37 and 41 weeks periods. GLM two-way ANCOVA analyses were used to

©2013 Foundation Acta Pædiatrica. Published by John Wiley & Sons Ltd 2013 102, pp. e402–e406 e403
COMFORT behavioural scale and skin conductance activity Trist~ao et al.

verify if the clinical or demographic variables were related

Table 2 Differences during-before and during-after painful event of Cb total and
to the scores. In multiple comparisons, the alpha level was
partial scores and SCA variables in 36 newborn infants.
adjusted using the Bonferroni procedure. Two different
During-before During-after
analyses were used to verify the relationship between the Cb
Indicators Mean (range) Mean (range)
and SCA scores: Spearman’s bivariate correlation was used
to compare the Cb total and partial scores, and Kendall’s Cb 23 (14–30)** 9 (6–25)**
coefficient of concordance was used to analyse agreement Alertness 3 (1–5)** 2 (1–5)**
between all Cb scores and skin conductance variables for Calmness 3 (1–5)** 2 (1–5)**
the time intervals of 15, 30 and 180 sec in the ‘during’ Crying 4 (1–5)** 1 (1–5)**
period. Kolmogorov–Smirnov’s normality test was run for Physical movement 4 (1–5)** 2 (1–5)**
Muscle tone 3 (2–5)** 2 (1–5)**
all data sets (p > 0.05), and Levene’s test of homogeneous
Facial tension 4 (3–5)** 2 (1–5)**
variances was not significant for all analyses (p > 0.05). The
NWps 0.12 ( 0.2–0.53)** 0.11 ( 0.86–0.4)**
minimum sample size was established at 26 participants to AUC 0.42 ( 4.94–8.26) 0.74 ( 9.77–8.07)*
permit an alpha error of 5% and a beta error of 20% for a
statistical power of 80%. Wilcoxon test *p < 0.05, **p < 0.01.

RESULTS
The mean gestational age of the infants was 38.9 weeks, Table 3 Spearman’s correlation between Cb and SCA indicators, 15, 30 and 180 sec
68.3% were delivered by caesarean section, 39% were large after pain
for their gestational age and 41.5% were small, 5.1% were Indicators 15 sec 30 sec 180 sec
born to diabetic mothers and 75% were breastfed 1 h before
Cb–NWps 0.424** 0.383* 0.504**
the procedure (Table 1). The paired-sample test of SCA
Cb–AUC 0.057 0.128 0.118
variable NWps and each indicator of Cb showed signifi-
Alertness–NWps 0.576** 0.468** 0.444**
cance (p < 0.01) for all differences ‘during-before’ and Alertness–AUC 0.431** 0.058 0.078
‘during-after’ (Table 2). AUC was only significant for Calmness–NWps 0.449** 0.405** 0.333*
‘during-after’ (p < 0.05). Cb indicators were significantly Calmness–AUC 0.316* 0.228 0.090
related to NWps after the procedure (15, 30 and 180 sec) Crying–NWps 0.328* 0.348* 0.437**
(p < 0.05), but showed fair to moderate correlations Crying–AUC 0.140 0.164 0.259
(r < 0.6), while Spearman’s correlation showed that crying Physical movement–NWps 0.560** 0.533** 0.609**
and calmness were the weakest factors for all time windows Physical movement–AUC 0.424** 0.314* 0.302
(Table 3). Physical movement and facial expression showed Muscle tone–NWps 0.383* 0.404** 0.413**
Muscle tone–AUC 0.264 0.208 0.206
a significant correlation to AUC 15 sec and 30 sec after the
Facial tension–NWps 0.418** 0.391* 0.417**
procedure, together with fair to moderate correlations
Facial tension–AUC 0.327* 0.154 0.150
(r < 0.6). The Cb total score was only statistically significant
(p < 0.005) after heel prick for NWps at all intervals, but *p < 0.05.
the highest correlation was to NWps at 180 sec (r = 0.504, **p < 0.01.
p < 0.001). No relation was found for clinical and demo-
graphic variables by GLM two-way ANCOVA analyses.
No Kendall’s coefficient of concordance was found
between total score of Cb and NWps and AUC at 15 sec,
Table 1 Sample clinical characteristics (N = 36) with marginally significant values at 30 sec. A highly
Clinical characteristics Values
significant Kendall tau-b value (0.313, p = 0.006) was
found at 180 sec. When the indicators and the three
Postnatal age in hours (M*  SD†) 18.2  10.1 intervals of SCA were examined, agreement was found
Number of previous painful events (M  SD) 2.7  1.2 between facial tension and NWps at 15 sec ( 0.303,
Number of previous heel pricks (M  SD) 2.4  1.0 p = 0.023). When it came to NWps at 180 sec, there was
Apgar 5º (M  SD) 9.2  0.7
agreement with facial tension (0.363, p = 0.001), and also
Resuscitation [N‡, (%§yes)] 3 (8.3)
muscular tone (0.270, p = 0.021), physical movement
Use of 25% oral glucose prior to heel prick [N(%yes)] 4 (11.1)
(0.299, p = 0.006) and alertness (0.321, p = 0.002).
Anaesthesia during delivery [N(%)]
None 2 (5.5)
Local 11 (30.5)
Regional 23 (63.8) DISCUSSION
Almost every factor of Cb was statistically significant
*
Mean;% (p < 0.005) when correlated with the NWps at 15, 30 and
†
Standard deviation.
‡
180 sec. Despite the statistical significance between the
Number.
§ scale indicators and the SCA, the overall score showed a
Percentage.
weak correlation (r < 0.50). This result may suggest that,

e404 ©2013 Foundation Acta Pædiatrica. Published by John Wiley & Sons Ltd 2013 102, pp. e402–e406
Trist~ao et al. COMFORT behavioural scale and skin conductance activity

although the scale is validated for pain in newborns, it in infants. Due to this variation, and analysing scale factors
works better for chronic pain, when the analysis time is separately, it can be seen that each one has an influence on
longer than 3 min. Added to this, the AUC showed poorer the total score and, therefore, can be used to predict pain in
sensitivity and only correlated with the 15 sec analyses for newborns. Despite this, its use in cases of acute pain should
physical movement and facial expression. This finding does be avoided because none of its factors, except facial tension,
not corroborate previous study findings that suggested that can truly produce the desired effect. Based on the findings
the AUC was the most sensitive and specific variable in of our study, our advice to health professions dealing with
measuring pain levels (8). the critical care of neonatals is to use complementary
Kendall’s coefficient of concordance analyses indicated parameters, such as the skin conductance activity, to
that the Cb total score bears more relation to SCA at late measure pain instead of relying solely on a multidimen-
intervals (180 sec), longer after the pain event (15 sec). sional scale.
Facial tension seems to be the gold standard factor for this
scale, as it remained at high levels throughout the whole
observation time. In contrast, crying and calmness proved FINANCIAL DISCLOSURE/CONFLICT OF INTEREST
to be poor indicators of acute pain, while muscle tone, The authors have no financial relationships or conflict of
physical movement and alertness only agreed with SCA at interest to disclose.
180 sec. Nevertheless, all agreements were only for NWps
and all reached a fair level of agreement. This result was
expected as the SCA measures pain much faster than ACKNOWLEDGEMENTS
behavioural scales. This research was sponsored by grants from CNPq and
Pain perception should be evaluated in three dimensions: FAHUB and also received the support of the staff from the
physiological, phenomenological and behavioural (6). Neonatal Intensive Care Unit, University of Brasilia Hos-
Using this theoretical approach, it is reasonable to assume pital, Brazil. Special thanks go to the mothers and babies
that facial tension and SCA responses are more related to and to Marcos Vinıcius Melo de Oliveira for video analysis
the physiological level, while alertness, physical movement of the COMFORT behaviour scale.
and muscle tone are more related to the behavioural and
phenomenological dimensions. The phenomenological
References
component, described as the personal account of pain and
anxiety, among others, is the most difficult to assess in 1. Harrison D, Boyce S, Loughnan P, Dargaville P, Storm H,
newborns as they are unable to speak. Anxiety can exist Johnston L. Skin conductance as a measure of pain and stress
even in the absence of pain. However, pain is related to in hospitalised infants. Early Hum Dev 2006; 82: 603–8.
2. Merskey H, Bogduk N, Part III. Pain terms, a current list with
external noxious stimulus that are able to activate ascending
definitions and notes on usage. In: Merskey H, Bogduk N,
cortical pathways for perception and interpretation of pain editors. Classification of chronic pain, IASP task force on
and descending pathways for the development of a response taxonomy, 2nd ed. Seattle, WA: IASP Press, 1994.
(6,20). Hence, it is necessary to differentiate between these 3. Anand KJ, Craig KD. New perspectives on the definition of
phenomenological components and to consider what can pain. Pain 1996; 67: 3–6.
definitely be interpreted as pain or just a sign of distress. 4. Bellieni CV. Pain assessment in human fetus and infants. AAPS
We also found that the behavioural variables of the Cb J 2012; 14: 456–60.
5. de Oliveira MVM, de Jesus JAL, Trista ~o RM. Psychophysical
showed significant differences, but a weak correlation, with
parameters of a multidimensional pain scale in newborns.
SCA variables. From this, we can assume that the variation Physiol Meas 2012; 33: 39–49.
in the score of each indicator follows the variation in the 6. Ambuel B, Hamlett KW, Marx CM, Blumer JL. Assessing
infant’s perception/response to pain. However, the data can distress in pediatric intensive care environments: the
also suggest that responses like increased physical move- COMFORT scale. J Pediatr Psychol 1992; 17: 95–109.
ments, muscle tone and alertness are more likely to be 7. van Dijk M, Peters WB, van Deventer P, Tibboel D. The
COMFORT behavior scale: a tool for assessing pain and
signals of anxiety than pain, or the cognitive/conative
sedation in infants. Am J Nurs 2005; 105: 33–6.
reaction, strengthening the affective-emotional dimension 8. Eriksson M, Storm H, Fremming A, Schollin J. Skin
hypothesis of the infants’ behaviour (13). To study this conductance compared to a combined behavioral and
premise in future studies, it could be suggested that, instead physiological pain measure in newborn infants. Acta Paediatr
of considering the whole 2-min interval, researchers should 2008; 97: 27–30.
use varieties of behavioural observation strategies such as 9. Storm H. Changes in skin conductance as a tool to monitor
time sampling, splitting the time window into smaller nociceptive stimulation and pain. Curr Opin in Anaesthesiol
2008; 21: 796–804.
intervals to identify any rise and fall of responses. This
10. van Dijk M, de Boer JB, Koot HM, Tibboel D, Passchier J,
might help to define which responses are more likely to be Duivenvoorden HJ. The reliability and validity of the
related to the past painful event and which are related to the COMFORT scale as a postoperative pain instrument in 0 to
fear of future pain. 3-year-old infants. Pain 2000; 84: 367–77.
In conclusion, this study observed that all factors of the 11. Storm H. The development of a software program for analyzing
Cb scale showed moderate correlation with the skin skin conductance changes in preterm infants. Clin
conductance measurement when it came to assessing pain Neurophysiol 2001; 112: 1562–8.

©2013 Foundation Acta Pædiatrica. Published by John Wiley & Sons Ltd 2013 102, pp. e402–e406 e405
COMFORT behavioural scale and skin conductance activity Trist~ao et al.

12. Padhye NS, Williams AL, Khattak AZ, Lasky RE. Heart rate comparison study of acute pain in Brazilian newborns. Conf
variability in response to pain stimulus in VLBW infants Proc IEEE Eng Med Biol Soc 2011; 30: 1875–9.
followed longitudinally during NICU stay. Dev Psychobiol 17. Gjerstad AC, Wagner K, Henrichsen T, Storm H. Skin
2009; 51: 638–49. conductance versus the modified COMFORT sedation score as
13. Aydede M. Critical and Quasi-Historical essay on theories of a measure of discomfort in artificially ventilated children.
pain. In: Aydede M, editor. Pain: new essays on Its nature and Pediatrics 2008; 122: 848–53.
the methodology of Its study. Massachusetts: Ed. Cambridge, 18. Hullet B, Chambers N, Preuss J, Zamudio I, Lange J, Pascoe E,
MIT Press, 2006. et al. Monitoring Electrical Skin Conductance: A tool for the
14. Storm H. Skin conductance and the stress response from heel assessment of postoperative pain in children? Anesthesiology
stick in preterm infants. Arch Dis Child 2000; 83: F143–7. 2009; 111: 513–7.
15. Gjerstad AC, Storm H, Hagen R, Huiku M, Qvigstad E, Raeder 19. Volpe JJ. Neurology of the Newborn, 5th ed. Philadelphia:
J. Comparison of skin conductance with entropy during Saunders Elsevier, 2008.
intubation, tetanic stimulation and emergence from general 20. Ranger M, Johnston C, Anand KJS. Current controversies
anaesthesia. Acta Anaesthesiol Scand 2007; 51: 8–15. regarding pain assessment in neonates. Semin Perinatol 2007;
16. Jesus JAL, Trista~o RM, Storm H, Rocha A, Campos Junior D. 31: 283–8.
Heart rate, oxygen saturation, and skin conductance: a

e406 ©2013 Foundation Acta Pædiatrica. Published by John Wiley & Sons Ltd 2013 102, pp. e402–e406