Biobehavioral Measures for Pain in the Pediatric Patient

Mamoona Arif-Rahu, RN, PhD, CCRN; Deborah Fisher, RN, MS, CS, CPON; Yui Matsuda, RN, BSN


Pain Manag Nurs. 2012;13(3):157-168. 

In This Article

Assessing Pain in Pediatric Patients

Neuromuscular Activity of Pain

The quantification of pain involves the incorporation of many imprecise factors, so measurements tend to be very subjective, relying heavily on the expression of the patient as well as the interpretation of the caregiver. An objective pain measurement tool would provide those in the clinical arena with a standardized form of measurement that would not vary with patient or caregiver. With this as a goal, various researchers have evaluated the use of electroencephalography (EEG), electromyography (EMG), and brain imaging techniques as the basis for such objective measurements of pain (Davis, Taylor, Crawley, Wood, & Mikulis, 1997; Dowman, Rissacher, & Schuckers, 2008; Henderson, Gandevia, & Macefield, 2008; Jancke, Vogt, Musial, Lutz, & Theodor, 1996; Raux, Ray, Prella, Duguet, Demoule, & Similowski, 2007).

EEG is readily available in the hospital setting for use on a wide range of patients, which makes it a tool of interest in research applications. It provides a noninvasive measurement of cortical brain activity. This activity can be measured by application of electrodes to the scalp and measurements taken at various band frequencies. It has been shown that the somatosensory cortex, the anterior cingulate cortex, and the somatosensory association areas located in the parietal operculum and insula are activated by painful stimuli (Apkarian, Bushnell, Treede, & Zubieta, 2005). Dowman et al., (2008) studied the pain-related cortical activity of the brain by using EEG during tonic experimental pain stimulus and control conditions. They observed pain-related changes in the EEG at specific electrode locations, such as the contralateral central, frontocentral, and temporal scalp regions. Their results indicated that pain-related changes in the EEG showed a decrease in alpha over the contralateral temporal scalp (Newman–Keuls test: p < .05).

Raux et al., (2007) used EEG to demonstrate a premotor cortical activation during a situation mimicking one form of patient-ventilator asynchrony in noninvasive ventilated patients. The basis of the Raux et al., (2007) study was the underlying association between specific brain regions and the control of breathing by use of EEG. The results indicated that mean inspiratory flow increased during "discomfort" compared with the two "comfort" periods (p = .0003). Furthermore, the EEG results indicated that motor potentials were significantly more frequent during "discomfort" than during any of the "comfort" conditions (Fisher exact test: p < .001) and thus are associated with an activation of the premotor cerebral cortex. Although EEG may be a valid tool to measure brain activities associated with pain response, it requires trained technician time to apply electrodes and translate results, therefore making it impractical for clinical settings.

EMG has been widely used in the assessment of pain in various regions of the body. EMG registers muscle activity with surface electrodes. Several studies have used EMG to study facial expressions by using intensity of muscular response in patients suffering from back pain, headache, and neck and shoulder pain (Cram & Steger, 1983; Ong, Nicholoson, & Gramling, 2003; Sonnby-Borgstrom, 2002; Weyers, Muhlberger, Hefele, & Pauli, 2006; Wolf, Raedler, Henke, Kiefer, Mass, Quante, & Wiedemann, 2005). Wolf et al., (2005) validated a facial EMG method for the measurement of facial pain expression in ten healthy subjects induced with a laser system pain stimulus. They studied nine muscle groups of the face. The results indicated two groups of muscles corresponding to pain expression. The first muscle group identified is assembled around the orbicularis oculi muscle, which initiates staring. The second group consists of the mentalis and depressor anguli oris muscles, both of which trigger mouth movements. The investigators recommended further studies with psychometric measurements, a larger sample size, and a female test group. Another study, by Jancke et al. (1996), examined the changes in facial EMG to auditory stimuli. They found that when high-intensity stimuli were administered to subjects, a strong upper-face EMG recording was observed. Similarly, in response to the crying of a baby, muscle activity in the mouth region increased.

So although EMG can be used to measure facial muscular activity, which can be used as an indicator of pain, EMG on its own should not be used as a direct absolute measurement of pain, in either a research setting or a clinical setting. One of the main drawbacks in the use of objective measurements such as EEG, EMG, and other brain imaging techniques is that there is a lack of research on the associated interpretation of the emotional response of patients (Jancke et al., 1996). In addition, intersubject reliability can be influenced by lead placement, possibly leading to inconsistent results.

Although a myriad of brain imaging techniques exist, the main one used in the assessment of pain is magnetic resonance imaging (MRI). This is largely due to the fact that MRI is noninvasive and has a combination of high temporal and spatial resolution. A study done by Henderson et al., (2008) demonstrated gender differences in functional MRI measurements made during muscle and cutaneous pain, 2008. They examined 22 healthy adult subjects (11 men, 11 women; aged 19–49 years). An increase change in signal intensity was used as their form of measurement, and they observed that there were gender-based differences in the hippocampus, cerebella cortex, midcingulate cortex, and dorsolateral prefrontal cortex. Women showed increased signal intensity which occurred in the cingulate, insular, primary somatosensory, secondary somatosensory, and cerebellar cortices during both muscle and cutaneous pain compared with men. Another study performed on 14 adult subjects (aged 14–40 years), by Owen, Bureau, Thomas, Prato, and Lawrence (2008), used perfusion MRI to examine pain-induced changes in cerebral blood flow. The induced pain consisted of a thermal stimulus on the left hand. They found that this method was effective for the measurement of chronic pain and observed changes in the insula, secondary somatosensory, and cingulate cortexes. Functional MRI has great utility but has limited applicability for real-time assessment of pain owing to the time involved in making measurements as well as the limitation that the machine itself poses to the mobility of the patient.

Chemical Biomarkers of Pain

Several chemical biomarkers for pain have been implicated in the literature (Table 1). Endogenous opioid neuropeptides (enkephalins, endorphins, dynorphins) have the ability to dampen the perception of pain (Rittner, Brack, & Stein, 2008). These actions include opiate-like activity that is involved in regulation of tolerance to pain within the central nervous system. β-Endorphin is a neuropeptide produced by the pituitary starting as early as 22 weeks' gestation. It can be found in amniotic fluid and fetal blood. Other substances implicated in the down-regulation of pain include epinephrine, norepinephrine, serotonin, and γ-aminobutyric acid. Some chemical mediators dually implicated in pain and inflammations include prostaglandins, histamine, serotonin, cytokines, and chemokines (Abbadie, 2005). Chemokines function in activating or modifiying nociceptive transmission. Increasing levels of interleukin-8, one of the first chemokines studied, are associated with increasing levels of hyperalgesia in rat models (Cunha, 1991). Studies in humans have found a positive relationship between spinal fluid levels of interleukin-8 and back pain (Brisby, 2002). Further research is warranted to delineate possible confounding variables affecting the ultimate pain response.

β-Endorphins have been associated with pain and stress response (Rittner et al., 2008). In addition, amniotic fluid β-endorphin (AFBE) levels were studied as a possible prognostic predictor of degree of intestinal damage in fetuses with gastroschisis, which causes severe pain response (Mahieu-Caputo et al., 2002). The intent was to find a relationship between the level of β-endorphin and postnatal morbidity. Postnatal morbidity was higher when AFBE exceeded 10 μg/L. That is, higher levels of AFBE were associated with more severe cases of gastroschisis. Researchers believe that the endorphin production is a result of prenatal stress and/or pain from bowel injury (Mahieu-Caputo et al., 2002). Limitations of this study include the relatively small sample size (13 infants with gastroschisis versus 33 infants without gastroschisis). One must weigh the limited benefit versus the risk of potentially causing more fetal harm with repeated invasive studies throughout the pregnancy. In summary, an increase in β-endorphin may be associated with increased pain response caused by injury, such as gastroschisis. Further research is indicated to better delineate the optimal utility of testing amniotic fluid β-endorphin levels and the impact on clinical outcomes after delivery.

A double-blind randomized trial compared the analgesic effectiveness of continuous infusion opioid versus intermittent bolus infusion of opioid in the ≤36-month-old postsurgical population (n = 204) (Bouwmeester, Anand, van Dijk, Hop, Boomsma, & Tibboel, 2001). Plasma concentrations of biologic variants (lactate, glucose, insulin, norepinephrine, and epinephrine) known to be elevated in pain response were measured at five time points (before surgery, at end of surgery, and 6, 12, and 24 hours after surgery). Presurgical levels of epinephrine were found to be higher than postsurgical levels of epinephrine. Elevation in biologic variables (lactate, glucose, insulin, norepinephrine, and epinephrine) was compared with pain assessment scales (visual analog scale and comfort). The two opioid treatment groups showed no significant difference in pain response as measured by the pain scales and the biologic variants. Further conclusions included difference in surgical stress response in neonates compared with older age groups. After surgery, neonates showed higher glucose levels, lower mean increases in epinephrine and norepinephrine, and higher insulin levels compared with older age groups. Implications for further research include opportunities to compare and determine efficacy of other analgesic interventions to assist in determining best practice. Further study is warranted to examine the variations in pain responses related to varying chronologic age.

One deterrent to using biomarker levels as a unidimensional tool to assess pain is their concurrent relationship with inflammation and/or stress response. Mousa, Straub, Schafer, and Stein (2007) found that the amount of opioid receptors and endogenous opioid agonists (β-endorphin and metenkephalin) correlated with level of inflammation associated with arthritis. Synovial fluid samples were subjected to double immunohistochemical analysis of opioid peptides with immune cell markers. The researchers found that β-endorphin and metenkephalin were expressed by macrophage-like cells within synovial lining layers. Overall, β-endorphin and metenkephalin were more prevalent in patients with rheumatoid arthritis than in patients with osteoarthritis or joint trauma. This study further supports the relationship between inflammation and pain response.

The potential role of cytokine measurement in pain assessment is limited by the potential difficulty in collecting samples without causing serious harm to the patient. Many of the cytokines are located within the dorsal root ganglia or spinal cord or within the tissues surrounding nerves and the site of injury (Zhang & An, 2007). The potential benefit of assessing chemical biomarkers of pain appears to be in furthering our knowledge of the neurobiology of the pain response. An understanding of the pathophysiologic responses of the human body to painful stimuli may allow for the development of improved multidimensional pain assessment tools.

Behavioral Measures of Pain

The ASPMN guidelines (Herr et al., 2006) recommend several clinical tools for use on "infants and preverbal toddlers" and "pediatric intubated and/or unconscious persons." Table 2 presents these scales for the nonverbal behavioral assessment tools for pediatric patients. Furthermore, the Pediatric Initiative on Methods, Measurement, and Pain Assessment in Clinical Trials identified measures to use in pediatric pain clinical research trials (Cohen, Lemanek, Blount, Dahlquist, Lim, Palermo, Mckenna, & Weiss, 2008). Those authors assessed the validity of some clinical pain assessment tools and suggested to select pain tools according to the purpose and context of the setting.

The three approaches to measure pain include self-report, behavioral, and physiologic. Self-report is considered to be the gold standard of pain assessment (Schiavenato & Craig, 2010). In one study comparing nurses' pain ratings using the Face, Legs, Activity, Cry, and Consolability (FLACC) scale (Merkel, Shayevitz, Voepeol-Lewis, Malviya, 1997) with those of patients' pain rating using the Wong-Baker FACES scale, no correlation between the two scales (r = 0.254; p = .381) were found in patients <5 years old (Willis, Merkel, Voepel-Lewis, & Malviya, 2003). However, there was a significant and positive correlation in children 5–7 years old (r = 0.830; p = .0001) demonstrating that self-report should be used to validate pain if the target population is developmentally capable of rating its own pain (Willis et al., 2003).

Unfortunately, pain assessment is difficult in nonverbal pediatric patients such as neonates, infants, preverbal toddlers, and intubated and/or unconscious cognitively impaired patients. Nonverbal pediatric patients are unable to communicate pain or discomfort, because of language, cognitive, developmental, or physiologic issues (Breau, Camfield, McGrath, & Finley, 2004; Cohen et al., 2008; Herr et al., 2006; Johnston, 1993; McGrath, Rosmus, Camfield, Campbell, & Hennigar, 1998). In these contexts, observational methods that focus on nonverbal behavior, including indices of facial, vocal, and motor behaviors and physiologic measures, have been identified.

Facial Expressions

Facial expression is considered to be a more stable and valid component for assessing acute and short-term pain (Breau, McGrath, Craig, Santor, Cassidy, & Reid, 2001; Craig, Hadjistavropoulos, Grunau, & Whitfield, 1994; Grunau & Craig, 1987). There are a number of behavioral assessment tools developed to capture facial expression (Table 3). The Facial Action Coding System (FACS), developed by Ekman and Friesen (1978), identifies facial expressions specific to pain that have shown evidence for validity and reliability in adults and children (Craig, 1998, Craig et al., 1994, Craig & Patrick, 1985; Oberlander, Gilbert, Chambers, O'Donnell, & Craig, 1999). The FACS describes 44 specific facial muscle movements or action units (AUs). The facial action units are typically identified through the use of slow-motion stop-frame feedback video to determine the presence and absence of discrete facial actions. In a study by Larochette, Chambers, and Craig (2006), the investigators used the FACS to identify genuine, suppressed, and faked facial expressions of pain in 50 healthy 8–12-year-old children. They identified 18 AUs related to pain response (Table 3). The interrater reliabilities for overall frequency and intensity of AUs were 0.95 and 0.79, respectively. The results indicated that more frequent and more intense lip corner puller (AU12), cheek raiser (AU6), and lid tightened (AU7) were displayed when exposed to cold water as a pain stimulus.

The Neonatal Facial Coding System (NFCS) developed by Grunau and Craig (1987) identifies specific pain facial actions among newborns undergoing heel lance procedures. Like the FACS, the NFCS assesses discrete facial actions using video played back in real time with stop-frame capability. An expert coder identified total facial activity and cluster-specific facial features (brow bulge, eye squeeze, nasolabial furrow, and open mouth) that have been shown to be significantly associated with acute and postoperative pain in infants (Craig, 1998; Craig, Whitfield, Grunau, Linton, Hadjistavropoulos, 1993; Grunau and Craig, 1990). Figure 1 illustrates some of the common facial actions corresponding to pain in infants.

Figure 1.

Facial expression correlated with pain using the Neonatal Facial Coding System.

Guinsburg, de Araújo Peres, Branco de Almeida, de Cássia Xavier Balda, Cássia Berenguel, Tonelotto, and Kopelman (2000) used two pain assessment methods, the NFCS and the Neonatal Infant Pain Scale (NIPS) (Lawrence, Alcock, McGrath, Kay, MacMurray, & Dulberg, 1993). They coded distinct facial activities that measure pain using NFCS, including presence or absence of eight facial movements: brow bulge, eye squeeze, nasolabial furrow deepened, open lips, mouth stretch, lips pursed, taut tongue, and chin quiver (1 point for each with a total score of 0–8 points). In addition, they used the NIPS, which is composed of facial expression (0/1 point), cry (0/1/2 points), breathing pattern (0/1 point), position of arms (0/1 point), position of legs (0/1 point), and state of arousal (0/1 point). The interrater reliability was established between two observers who scored the NFCS and the NIPS using the number of actions agreed upon by both observers divided by the number of actions scored by the two observers. The results indicated interobserver reliability of κ = 0.94 for NFCS and κ = 0.93 for NIPS. Furthermore, there was strong agreement between the two coders for the vast majority of NFCS and NIPS items.

Similarly, the Child Facial Coding System (CFCS) (Chambers, Cassidy, McGrath, Gilbert, & Craig, 1996) was developed to assess acute pain responses in young children aged 1–6 years. The CFCS codes 13 discrete AUs adapted from both the FACS and the NFCS. Because there are less age-related differences in facial activity beyond 1 year of age, CFCS has been used in adolescents to identify pain expressions (Lilley, Craig, & Grunau, 1997). Breau et al., (2001) studied 123 children aged 4–5 years undergoing routine diphtheria, pertussis, tetanus, and polio (DPT) immunization. The purpose of the study was to establish sensitivity of the CFCS during noxious stimuli. For the study, the DPT immunization injection was the noxious stimulus. They identified 13 facial actions, of which six reflected the children's acute pain experience. The results indicated that AU frequency and intensity for brow lower, squint, flared nostril, nose wrinkle, lip corner pull, and vertical mouth stretch occurred more often during immunization injection (the needle phase) than before the injection (preneedle phase; p < .004 [frequency] and p < .005 [intensity]).

Overall, these FACSs provide a comprehensive description of the facial expression during painful stimuli which could be used to develop better behavioral assessment tools. One major limitation for use of the FACS in the clinical setting is the extensive training and complexity of coding that is required. Nonetheless, once the facial expressions are narrowed down, FACS coding could be used to assess facial expression during pain responses related to varying pediatric populations.


There have been cry features that have been extensively studied using spectrographic devices. Short latency to onset of cry, longer duration of the first cry cycle, higher fundamental frequency, and greater intensity in the upper ranges are pain-specific cry features in infants and neonates during painful procedures (Grunau, Johnston, & Craig, 1990; Johnsonton & Strada, 1986; Krechel & Bildner, 1995).

In a study by Runefors, Arnbjörnsson, Elander, and Michelsson (2000), the researchers hypothesized that a newborn infant's cry can be used in conjunction with an instrument to measure pain. They used heel sticks as noxious stimuli to elicit pain response for phenylketonuria screening in 50 healthy newborn infants. Their cries of pain were recorded and analyzed acoustically with the assistance of a computer program especially designed for this purpose (Innomess Elektronik, Berlin, Germany). The sound spectrogram, a well tested instrument, is a visual diagram of the sound signal. Time is recorded on the horizontal scale; frequency is recorded on the vertical. The curve of the lowest harmonic on the spectrogram gives the fundamental frequency, and the upper lines give the harmonic overtone multiples of the fundamental frequency. The analysis showed that the crying sound after the painful stimulus of the heel prick had a significantly higher fundamental frequency and lasted longer at the first cry (2.7 seconds) than at the fifth cry (0.8 seconds; p < .001). The results indicated that the first cry was more like a cry of pain, and the fifth cry more resembled crying for reasons other than pain. They suggested that newborn infants react to pain in a recognizable way. However, they also suggested that other stimuli may cause a similar reaction. The researchers recommended that crying can be used to measure pain in newborn infants only when the cause of crying is known. A significant clinical practice limitation of cry analyses is the need for a specialized spectrographic apparatus that requires advanced training. The expense of the training and the apparatus contributes to increased research costs. In addition, not all infants will cry when in pain, such as patients with endotracheal intubation or physiologic fatigue (Grunau et al., 1990, Johnston and Strada, 1986, van Dijk et al., 2002).

Motor Behaviors

Body movement as a pain indicator focuses on observation of arm and leg activity. Body movement is included in many pain assessment tools, but this behavioral marker changes with development from the neonatal stage through infancy and into adolescence. Increased activity, posture, and tense muscle tone are thought to indicate more pain in neonates (Hummel & van Dijk, 2006). Muscle tone and breathing pattern take an intermediate position between behavioral and physiologic indicators (van Dijk et al., 2002). Muscle tone or posture may reflect tenseness due to abdominal pain but may also reflect the behavior of a frightened child. Shallow and rapid breathing may reflect hyperventilation due to anxiety or may be caused by pain after thoracic or abdominal surgery (van Dijk et al., 2002). A major limitation of assessing body movement is that children may avoid moving because of pain (van Dijk et al., 2002); therefore, special attention needs to be paid to increased or decreased activity from a child's baseline.

Physiologic Measures

Physiologic measures of pain, which include heart rate, respiratory rate, oxygen saturation, and blood pressure, have been examined in neonates during acute pain caused by heel lance (Johnston, Stevens, Yang, & Horton, 1995) and circumcision (Howard, Howard, & Weitzman, 1994). The benefit of physiologic measures for pain assessment has been described by Sweet and McGrath (1998) but it has been debated because of lack of specificity for pain (van Dijk, de Boer, Koot, Duivenvoorden, Passchier, Bouwmeester, & Tibboel, 2001). Fluctuations in heart rate, blood pressure, and oxygen saturation may be influenced by poor circulation due to blood loss, fluid intake, body temperature, and medical interventions. Several studies have concluded that physiologic pain indicators are not ideal to assess pain and recommend that the pain assessment tools be based on behavioral pain indicators (Buchholz, Karl, Pomietto, & Lynn, 1998; Buttner & Finke, 2000; McGrath & Unruh, 1994).

In a study by van Dijk et al. (2001), the researchers examined the association between physiologic and behavioral pain measures using the COMFORT scale in children. The COMFORT scale consists of both behavioral and physiologic components. The behavioral components are described further in Table 2. Physiologic measures included mean arterial pressure, heart rate, heart rate variability, and mean arterial pressure variability. The researchers studied 204 subjects and found that as children between 0 and 3 years old increase in age, their respiratory and heart rates decrease and their overall blood pressure increases in response to postoperative pain. That study also found moderate correlations between crying and heart rate as well as between facial expressions and heart rate variability in nonverbal children. A low correlation between the physiologic measures made during the study seems to indicate that the measurements were independent of each other. Intermittent noninvasive measuring of heart rate and blood pressure could induce distress or anger in infants and children, which would increase the unreliability of such pain indicators (van Dijk, 2001).

In summary, the means for validating the pain assessment tools are provided in Table 2. Although all of the tools have behavioral components, the Crying, Requires Oxygen, Increased Vital Signs, Expression, and Sleep (CRIES) scale (Krechel & Bildner, 1995) and the COMFORT scale are the only ones that contain both physiologic and behavioral measures of pain. The correlations of the physiologic measures are lower on the studies that tested with the COMFORT scale (Ambuel, Hamlett, Marx, & Blumer, 1992; van Dijk, de Boer, Koot, Tibboel, Passchier, & Duivenvoordent, 2000). Ambuel et al., (1992) pointed out that heart rate and arterial blood pressure had the lowest inter-rater reliabilities due to the way they were measured. The researchers predict that the inter-rater reliability would increase if the physiological measures are observed for two minutes and the trend is measured (Ambuel et al., 1992). In addition, van Dijk et al. (2000) pointed out the limited validity of heart rate and mean arterial pressure as a measurements of pain and suggested the need for further research with clinical data to support these measures, because they are frequently used as indicators of pain in clinical settings. However, from a research perspective, the physiologic measures may or may not be appropriate depending on the specific pediatric population.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: