Validity of the Child Facial Coding System for the Assessment of Acute Pain in Children With Cerebral Palsy

Abstract

The purpose of the current study was to examine the concurrent and discriminant validity of the Child Facial Coding System for children with cerebral palsy. Eighty-five children (mean = 8.35 years, SD = 4.72 years) were videotaped during a passive joint stretch with their physiotherapist and during 3 time segments: baseline, passive joint stretch, and recovery. Children’s pain responses were rated from videotape using the Numerical Rating Scale and Child Facial Coding System. Results indicated that Child Facial Coding System scores during the passive joint stretch significantly correlated with Numerical Rating Scale scores (r = .72, P < .01). Child Facial Coding System scores were also significantly higher during the passive joint stretch than the baseline and recovery segments (P < .001). Facial activity was not significantly correlated with the developmental measures. These findings suggest that the Child Facial Coding System is a valid method of identifying pain in children with cerebral palsy.

Keywords

pain cerebral palsy child Child Facial Coding System facial activity procedural pain acute pain neurological impairment

Over the past 15 years there has been a growing body of research examining pain in normally developing children. Until recently, children with neurological impairment (eg, cerebral palsy, Down syndrome, traumatic brain injury) have been excluded from these studies. As a result, researchers are only beginning to understand the pain indicators for children with neurological impairment. Given that many children with neurological impairment have difficulty verbally communicating their pain experience, as well as difficulty with voluntary physical movement, this population creates pain assessment challenges for clinical researchers.

One method of resolving the barriers associated with pain assessment for children with neurological impairment is to identify facial activity that is reflective of pain experience. Facial action coding systems were first used to identify pain in infants and children who followed typical developmental milestones in cognitive and physical development. These coding systems have been an important research tool for the study of pain because they are objective, comprehensive, and based on anatomical reactions.^1,2 Several studies have documented the utility of examining facial activity associated with pain with both preterm and full-term infants,^3
–5 children with developmental delays and autism,⁶ and typically developing children.^7

–10

The few studies that have examined facial activity associated with pain in children with neurological impairment have shown conflicting results. Some studies have shown diminished facial activity in children with cerebral palsy,^11,12 while 2 studies have shown increased facial activity during venepuncture procedure in children with autism.^6,13 A number of factors can account for these mixed findings, including differences in the types of noxious events (parent assisted stretch, influenza vaccine injection, and venepuncture), small sample sizes, differences in primary diagnosis (ie, cerebral palsy and autism), age variability, and severity of cognitive and physical impairments. It is interesting to note that in the studies^11,12 that showed attenuated facial activity in children with cerebral palsy, observers’ global pain ratings were significantly higher during the injection¹² and parent assisted stretch.¹¹ Therefore, observers’ global pain judgments indicated that the children were displaying signals of pain, whereas the facial coding systems indicated that the children’s pain responses were attenuated. It is not clear what factors mediated observer global pain judgments, however, in previous research nurses reported that facial activity is an important factor in their global estimates of pain.¹⁴ More research is needed to examine the facial responses of children with cerebral palsy during clinical procedures that are part of their regular treatment program.

The aim of the current study is to examine the discriminant and concurrent validity of the Child Facial Coding System with a large sample of children who have cerebral palsy. It is predicted that the frequency and intensity of facial activity will be significantly higher during the treatment procedure then the baseline and recovery segments. It is also expected that the Child Facial Coding System scores will be significantly correlated with the Numerical Rating Scale scores for each time segment.

Method

Participants

Eighty-five children and adolescents diagnosed with cerebral palsy who were between the ages of 2 to 18 years of age participated in the study. The children consisted of 44 females and 41 males. All participants in the study had been attending regular physiotherapy sessions as part of their treatment program. This study was approved by the Interdisciplinary Committee on Ethics in Human Research at Memorial University in St. John’s, Newfoundland, Canada.

Measures

Language and adaptive ability measures

The Peabody Picture Vocabulary Test–Revised¹⁵ is a nonverbal, multiple-choice test that is designed to assess the hearing vocabulary or receptive knowledge of children and adults. Studies have shown the Peabody Picture Vocabulary Test–Revised has good psychometric properties.^16,17

The Vineland Adaptive Behavior Scales Survey Form¹⁸ assesses the social abilities of individuals with and without a disability from birth through age 19. The Vineland Adaptive Behavior Scales requires that a parent familiar with the behavior of the child answer behavior-oriented questions posed by a trained examiner. Three separate domains are assessed in the current study: Communication, Daily Living Skills, and Socialization. The Vineland Adaptive Behavior Scales has been shown to be a reliable and valid measure of adaptive abilities.¹⁸

The child’s gross motor functioning was measured using the Gross Motor Function Classification System for cerebral palsy.¹⁹ The Gross Motor Function Classification System has been shown to have high interrater reliability (.93) and test–retest reliability (.79).^20,21

Observational pain measures

The Child Facial Coding System²² codes 13 discrete facial actions. Ten of the facial actions were coded as 0 (not present), 1 (present), 2 (present with intensity). Three facial actions (blink, open lip, and flaired nostril) were not coded for intensity. Child Facial Coding System has been shown to have good reliability and validity for children’s responses to acute pain.^7,9 Only the participant’s face was visible on the videotape during the Child Facial Coding System ratings to control for context cues.

The Numerical Rating Scale provides a 0 to 5 scale for estimation of pain intensity, where 0 denotes no pain and 5 denotes worst possible pain. The Numerical Rating Scale can be used both as a self-report or proxy report measure. In the current study only proxy reports were included. The Numerical Rating Scale (0-5) has been shown to demonstrate utility in measuring pain with children who have developmental delays.^11,23
–25

Procedure

Families were recruited through the main tertiary care hospital in Newfoundland. The study was introduced to caregivers and their children during a physiotherapy appointment. Those caregivers and children interested in participating in the study were met at their next physiotherapy session by the research coordinator. Children participating in the study were videotaped during a passive joint stretch with their regular physiotherapist. A passive joint stretch involves a physiotherapist moving a muscle in the child’s body (ie, hamstring or heel cord) to the end point of available range and holding this position. For the purposes of coding pain, the children were videotaped during 3 time segments: (1) 10 minutes before the pain condition (baseline), (2) during the pain condition (passive joint stretch), and (3) 10 minutes following the pain condition (recovery). A digital video camera recorded the child’s face during each time segment. An edited videotape was created for coding in which the 3 time segments for each participant were randomized, and were always contiguous. Coding was completed on 10-second segments for the baseline, passive joint stretch, and recovery for a total of 30 seconds for each participant.

A trained Child Facial Coding System coder rated the videotape segments. All Child Facial Coding System coding was completed without sound. Only the participant’s face was visible on the videotape during the Child Facial Coding System rating to control for context cues. A second trained Child Facial Coding System coder rated a random selection of 20% of the data. Using Ekman and Friesen’s²⁶ reliability formula for facial expressions (which involves dividing twice the number of facial actions agreed on by the total number of facial actions scored by each coder), interrater reliability was found to be .81.

Proxy Numerical Rating Scale ratings were completed by a trained research assistant from the same edited videotape used by Child Facial Coding System coders. All Numerical Rating Scale coding was completed without audio. Of the data, 20% were randomly selected and coded using the Numerical Rating Scale by a second rater. Reliability between Numerical Rating Scale raters was computed using an intraclass correlation coefficient and found to be .90.

Results

Descriptive statistics for the developmental and communication variables are presented in Table 1. As Table 1 shows, the mean chronological age of participating children was 8.35 years (SD = 4.72). Children scored below their chronological age on all Vineland Adaptive Behavior Scales subscales: socialization (mean = 5.79 years, SD = 4.72 years), daily living (mean = 4.45 years, SD = 4.26 years), and communication (mean = 5.43 years, SD = 44.65 years); thus, the sample as a whole had delays in adaptive behavior. The children also scored under their chronological age on the Peabody Picture Vocabulary Test–Revised with an average age of 7 years, 9 months (SD = 4 years, 1 month). Although all children had varying degrees of verbal ability, 34 children were not able to pass the Peabody Picture Vocabulary Test–Revised training because of severe sensory (hearing and vision) and cognitive impairments.

Table 1.

Descriptive Statistics for the Developmental and Communication Variables.

	Mean (years)	SD (years)	n	%
Chronological age	8.35	4.72	85	100
VABS subscales
Socialization	5.79	4.72	85	100
Daily Living	4.45	4.26	85	100
Communication	5.43	4.65	85	100
PPVT-R	8.29	4.54	85	100
GMFCS
Level I	—	—	24	28.2
Level II	—	—	12	14.1
Level III	—	—	12	14.1
Level IV	—	—	16	18.8
Level V	—	—	21	24.7

Abbreviations: GMFCS, Gross Motor Function Classification System; PPVT-R, Peabody Picture Vocabulary Test–Revised; VABS, Vineland Adaptive Behavior Scales.

Descriptive Statistics for measures of physical impairment

The median Gross Motor Function Classification System score was level 3, indicating a moderate level of physical impairment (that is, they could walk with assistive mobility devices and had limitations walking outdoors). As shown in Table 1, the physical impairments (Gross Motor Function Classification System) of children ranged from severe limitations in self-mobility to mild physical limitations that did not interfere with the children’s mobility. Cerebral palsy diagnosis included hemiplegia (8 children), quadriplegia (44 children), diplegia (30 children), athetoid (2 children), and tetraparesis (1 child), along with a diagnosis of spasticity for 79% of the sample. Of the children, 52% had a visual impairment and 9% had a hearing impairment.

Relationships between Pain Measures and Demographic Variables

Pearson product–moment correlational analysis was conducted to examine the relationship between demographic variables and the pain measures during the passive joint stretch. Significant correlations were obtained between chronological age and both Child Facial Coding System frequency scores, r(85) = –.32, P < .05, and Child Facial Coding System intensity scores, r(85) = –.31, P < .05, during the passive joint stretch segment. There was no significant correlation between chronological age and Numerical Rating Scale score during the passive joint stretch. There was a significant correlation between Gross Motor Function Classification System and the Numerical Rating Scale at passive joint stretch, r(85) = .26, P < .05; however, no significant correlation was found between Gross Motor Function Classification System and Child Facial Coding System scores. The children’s adaptive behavior (Vineland Adaptive Behavior Scales) and vocabulary (Peabody Picture Vocabulary Test–Revised) scores were also not significantly correlated with the pain measures.

Discriminant validity

Child Facial Coding System Frequency Scores Across Time

Table 2 presents the means and standard deviations for the dependent measures. A 3 (time segments) by 5 (Gross Motor Function Classification System) mixed measures analysis of covariance, with chronological age as the covariate, was conducted on the Child Facial Coding System frequency scores. Results indicated a significant main effect for time, F(2, 78) = 16.36, P < .001. The interaction between time and Gross Motor Function Classification System was not significant, F(8, 156) = 0.60, P = .78. A significant interaction was also obtained between chronological age and time, F(2, 78) = 6.17, P = .003. Bonferroni pairwise comparisons revealed that Child Facial Coding System frequency scores were significantly higher during the passive joint stretch segment than during the baseline (P < .001) and recovery segments (P < .001); there was no significant difference between the latter 2 segments.

Table 2.

Means and Standard Deviations (SDs) for the CFCS and NRS.

	CFCS frequency			CFCS intensity			NRS
	Mean	SD	Range	Mean	SD	Range	Mean	SD	Range
Baseline	2.94	0.25	0-11.50	2.02	0.22	0-9.60	0.18	0.07	0-1.00
PJS	5.10	0.37	0-14.80	4.02	0.35	0-14.40	1.33	0.13	0-4.00
Recovery	2.73	0.23	0-11.40	1.89	0.21	0-10.20	0.19	0.06	0-1.00

Abbreviations: CFCS, Child Facial Coding System; NRS, Numerical Rating Scale; PJS, passive joint stretch.

To determine whether Child Facial Coding System frequencies for each of the 13 action units were significantly different across time segments, separate 1-way analyses of variance with repeated measures were conducted (see Table 3). As Table 3 shows, only 2 of the 13 Child Facial Coding System action units were nonsignificant across time segments: blink and lip corner puller. Bonferroni pairwise comparisons indicated that Child Facial Coding System frequencies were significantly higher during the passive joint stretch than during the baseline and recovery segments for brow lower, eye squeeze, nose wrinkler, nasolabial furrow, cheekraiser, open lips, upper lip raise, and vertical mouth.

Table 3.

Analysis of Variance Statistics for CFCS Action Unit: A Comparison Among Baseline, PJS, and Recovery Phases.

Facial action unit	df	F	P
Brow lower	2, 83	11.71	.000
Squint	2, 83	4.10	.02
Eye squeeze	2, 83	9.22	.000
Blink	2, 83	1.75	.18
Flared nostril	2, 83	4.22	.02
Nose wrinkler	2, 83	8.48	.000
Nosolabial furrow	2, 83	12.27	.000
Cheekraiser	2, 83	9.26	.000
Open lips	2, 83	11.73	.000
Upper lip raise	2, 83	16.07	.000
Lip corner puller	2, 83	0.21	.81
Vertical mouth	2, 83	11.87	.000
Horizontal mouth	2, 83	3.26	.04

Abbreviations: CFCS, Child Facial Coding System; PJS, passive joint stretch.

Child Facial Coding System intensity scores across time

A 3 (time) by 5 (Gross Motor Function Classification System) mixed measures analysis of covariance with chronological age as the covariate was conducted on the Child Facial Coding System intensity scores. Results indicated a significant main effect for time, F(2, 78) = 16.53, P < .001, and a significant interaction between time and chronological age, F(2, 78) = 6.13, P = .003. The interaction between time and Gross Motor Function Classification System was not significant, F(8, 156) = 0.56, P = .81. Bonferroni pairwise comparisons revealed that Child Facial Coding System intensity scores were significantly higher during the passive joint stretch segment than during the baseline (P < .001) and recovery (P < .001) segments; as anticipated, there were no significant differences in Child Facial Coding System intensity scores between the baseline and recovery segments.¹

Concurrent validity

Table 4 shows the Pearson correlations between Numerical Rating Scale pain ratings and both Child Facial Coding System frequency scores and Child Facial Coding System intensity scores. The Child Facial Coding System frequency and intensity scores were significantly correlated with the Numerical Rating Scale scores for each of the corresponding time segments (all P < .01). Most notably, Numerical Rating Scale scores during the passive joint stretch were significantly positively correlated with both Child Facial Coding System frequency scores and Child Facial Coding System intensity scores during the passive joint stretch (r = .72 and .71).

Table 4.

Correlations Between NRS and CFCS Scores.

	1	2	3	4	5	6	7	8	9
1. CFCS frequency baseline	1.00
2. CFCS frequency PJS	.15	1.00
3. CFCS frequency recovery	.59**	.11	1.00
4. CFCS intensity baseline	.99**	.15	.58**	1.00
5. CFCS intensity PJS	.13	.99**	.08	.14	1.00
6. CFCS intensity recovery	.60**	.13	.99**	.59**	.16	1.00
7. NRS baseline	.57**	.09	.28*	.50**	.09	.24*	1.00
8. NRS PJS	.03	.72**	.06	–.01	.71**	.04	.11	1.00
9. NRS recovery	.29*	.01	.36**	.24*	.06	.35**	.40**	.10	1.00

Abbreviations: CFCS, Child Facial Coding System; NRS, Numerical Rating Scale; PJS, passive joint stretch.

*Correlation significant at P < .05 (2-tailed). **Correlation significant at P < .01 (2-tailed).

Discussion

The main purpose of the current study was to examine the concurrent and discriminant validity of the Child Facial Coding System during a passive joint stretch with children who have cerebral palsy. As expected, the Child Facial Coding System frequency and intensity scores significantly correlated with Numerical Rating Scale scores indicating concurrent validity. Significantly higher Child Facial Coding System intensity and frequency scores demonstrated discriminant validity during the passive joint stretch segment when compared to the baseline and recovery segments (when chronological age was controlled). These findings are inconsistent with previous research on facial activity associated with pain in children with cerebral palsy. For example, Hadden and von Baeyer¹¹ found that the Child Facial Coding System frequency scores of 19 children with cerebral palsy were not significantly higher during a parent assisted stretch segment when compared to baseline and recovery segments. Oberlander and colleagues¹² also did not find significantly higher Child Facial Coding System scores for 9 adolescents with cerebral palsy during a needle injection segment. It is possible that the reason previous research did not identify a significant difference in Child Facial Coding System scores during the intervention is due to the small samples sizes, limiting power. The current study is the first to examine facial activity (Child Facial Coding System) associated with pain in a large sample of children with cerebral palsy. Given the large variability in cognitive and physical disabilities inherent in the diagnostic categories of children with cerebral palsy, and those with other neurological impairments, it is essential that large samples of children be investigated. A second explanation could be that the passive joint stretch (assisted by a physiotherapist) is a more noxious stimulus than needle injection or parent assisted stretching. Although physiotherapy assessment procedures are not invasive—that is, they do not involve tissue-damaging procedures—they are nevertheless painful and distressing.²⁷ Parents can feel the urge to “go easy” on their child during assisted stretching exercises, while procedures administered by physiotherapists can be more focused on clinical benefits with the recognition of a certain degree of pain that is inherent in the process. Future research should assess the validity of Child Facial Coding System in children with neurological impairment across a variety of painful stimuli and durations.

Although the current results are inconsistent with prior research with children with cerebral palsy, there is some evidence to suggest that Child Facial Coding System has been effective in identifying pain in other paediatric populations. For example, Rattaz and colleagues⁶ found that children with autism displayed significant Child Facial Coding System facial activity during venepuncture. The current results are also consistent with past research examining the discriminant validity of Child Facial Coding System with typical children.^8,10

It is noteworthy that in the current study, 5 of the 6 Child Facial Coding System action units (brow lower, squint, flared nostril, nose wrinkle, and vertical mouth stretch) were associated with the typical “pain face.”^7,8 These results suggest that children with cerebral palsy can display similar facial activity to that of typically developing children. These findings are interesting given that researchers tend to compare children with neurological impairment in terms of their differences with typically developing children, while ignoring many of the similarities that can exist between groups.

Given our study findings, facial coding systems such as Child Facial Coding System seem to provide an objective and valid method of measuring pain in children with cerebral palsy. Facial activity can be particularly salient for children who are unable to provide self-report due to the severity of their cognitive and language impairments. Given that current facial coding systems require slow-motion replay and extensive training, these systems are probably most appropriate for research settings. However, future research should examine whether the facial actions (brow lower, squint, flared nostril, nose wrinkle, lip corner pull, and vertical mouth stretch) specific to pain in typical children could be reliably coded in real time by caregivers and health care providers for children with cerebral palsy in clinical settings. Researchers have shown some success in the use of real-time coding of facial activity with infants.^4,5

The current study has several limitations. First, the degree of the children’s cognitive deficit was not fully examined in the current study. Although a more detailed cognitive assessment would advance our knowledge of pain responses in children with neurological impairment, barriers to accurately assessing cognitive impairment in children with severe physical, sensory, and language impairments are well recognized.²⁸ In the current study, 34 children were unable to complete the cognitive measure (Peabody Picture Vocabulary Test–Revised) as a result of severe sensory (hearing and vision) and cognitive impairments. Although supplementary analyses revealed that performance on the Peabody Picture Vocabulary Test–Revised did not significantly influence the original findings, further research should utilize a wider range of cognitive measures to determine whether cognitive challenges influence acute pain responses for children with neurological impairment.

A second limitation of the current study was that global pain intensity and Child Facial Coding System ratings were completed from videotape recordings that were randomized to control for context cues. While these restrictions increased experimental control and improved raters adherence to facial cues for mediating their pain judgments, the generalizability of findings to the clinical pain decision-making process is not known.

Facial activity has long been understood as important for predicting the psychological states and behavior of human beings.²⁹ Facial responses to pain have been shown to be a significant factor in pain judgments from infancy through adulthood.^{1,6,8,11
–13,30,31} To our knowledge, the current study is the first to find significantly higher facial activity for children with cerebral palsy during a noxious stimulus. The strength of the current study was that facial activity was coded for a large number of children with cerebral palsy from a clinical procedure that was part of their regular treatment program. The findings indicated that Child Facial Coding System is a valid method of determining pain in children with cerebral palsy.

Footnotes

Acknowledgments

The authors would like to thank all the caregivers, children, and the physiotherapists who participated in this study.

Note

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research has been supported by grants to the first author as primary investigator from the Canadian Institutes of Health Research (CIHR) Regional Partnership Program, the Newfoundland & Labrador Centre for Applied Health Research, Canada-Newfoundland Comprehensive Economic Development Agreement, the Province of Newfoundland: Department of Health and Community Services, and the Janeway Children’s Hospital Foundation.

References

Prkachin

. Assessing pain by facial expression: facial expression as nexus. Pain Res Manag. 2009;14(1):53–58.

Hadjistavropoulos

LaChapelle

MacLeod

Snider

Craig

. Measuring movement-exacerbated pain in cognitively impaired frail elders. Clin J Pain. 2000;16(1):54–63.

Peters

Koot

Grunau

. Neonatal Facial Coding System for assessing postoperative pain in infants: item reduction is valid and feasible. Clin J Pain. 2003;19(6):353–363.

Grunau

Oberlander

Holsti

Whitfield

. Bedside application of the Neonatal Facial Coding System in pain assessment of premature neonates. Pain. 1998;76(3):277–286.

Guinsburg

de Araujo Peres

Branco de Almeida

. Differences in pain expression between male and female newborn infants. Pain. 2000;85(1-2):127–133.

Rattaz

Dubois

Michelon

Viellard

Poinso

Baghdadli

. How do children with autism spectrum disorders express pain? A comparison with developmentally delayed and typically developing children. Pain. 2013;154(10):2007–2013.

Breau

Camfield

McGrath

Rosmus

Finley

. Measuring pain accurately in children with cognitive impairments: refinement of a caregiver scale. J Pediatr. 2001;138(5):721–727.

Gilbert

Lilley

Craig

. Postoperative pain expression in preschool children: validation of the child facial coding system. Clin J Pain. 1999;15(3):192–200.

Chambers

Johnston

. Developmental differences in children’s use of rating scales. J Pediatr Psychol. 2002;27(1):27–36.

10.

Cassidy

Reid

McGrath

Smith

Brown

Finley

. A randomized double-blind, placebo-controlled trial of the EMLA patch for the reduction of pain associated with intramuscular injection in four to six-year-old children. Acta Paediatr. 2001;90(11):1329–1336.

11.

Hadden

von Baeyer

. Global and specific behavioral measures of pain in children with cerebral palsy. Clin J Pain. 2005;21(2):140–146.

12.

Oberlander

Gilbert

Chambers

O’Donnell

Craig

. Biobehavioral responses to acute pain in adolescents with a significant neurologic impairment. Clin J Pain. 1999;15(3):201–209.

13.

Nader

Oberlander

Chambers

Craig

. Expression of pain in children with autism. Clin J Pain. 2004;20(2):88–97.

14.

Goodenough

Addicoat

Champion

. Pain in 4- to 6-year-old children receiving intramuscular injections: a comparison of the Faces Pain Scale with other self-report and behavioral measures. Clin J Pain. 1997;13(1):60–73.

15.

Dunn

. Peabody Picture Vocabulary Test–III. Circle Pines, MN: American Guidance Service; 1997.

16.

Altepeter

Handal

. A factor analytic investigation of the use of the PPVT-R as a measure of general achievement. J Clin Psychol. 1985;41(4):540–543.

17.

Breen

. Comparison of the Wechsler Intelligence Scale for Children–Revised and the Peabody Picture Vocabulary Test–Revised for a referred population. Psychol Rep. 1981;49(3):717–718.

18.

Sparrow

Balla

Cicchetti

. Vineland Adaptive Behavior Scales: Interview Edition Survey Form. Circle Pines, MN: American Guidance Service; 1984.

19.

Palisano

Rosenbaum

Walter

Russell

Wood

Galuppi

. Development and reliability of a system to classify gross motor function in children with cerebral palsy. Dev Med Child Neurol. 1997;39(4):214–223.

20.

Boyce

Gowland

Rosenbaum

. Gross motor performance measure for children with cerebral palsy: study design and preliminary findings. Can J Pub Health. 1992;83(suppl 2):S34–S40.

21.

Wood

Rosenbaum

. The gross motor function classification system for cerebral palsy: a study of reliability and stability over time. Dev Med Child Neurol. 2000;42(5):292–296.

22.

Chambers

Cassidy

Gilbert

Craig

. Child Facial Coding System: Revised Manual. Halifax, NS, Canada: IWK Grace Health Centre, Dalhousie University and University of British Columbia; 1996.

23.

Fanurik

Koh

Harrison

Conrad

Tomerlin

. Pain assessment in children with cognitive impairment. An exploration of self-report skills. Clin Nurs Res. 1998;7(2):103–119; discussion 120-124.

24.

Fanurik

Koh

Schmitz

Harrison

Conrad

. Children with cognitive impairment: parent report of pain and coping. J Dev Behav Pediatr. 1999;20(4):228–234.

25.

Fanurik

Koh

Schmitz

Harrison

Roberson

Killebrew

. Pain assessment and treatment in children with cognitive impairment: a survey of nurses’ and physicians’ beliefs. Clin J Pain. 1999;15(4):304–312.

26.

Ekman

Friesen

. Investigator’s Guide to the Facial Action Coding System. Palo Alto, CA: Consulting Psychologists Press; 1978.

27.

von Baeyer

Tupper

. Procedural pain management for children receiving physiotherapy. Physiother Can. 2010;62(4):327–337.

28.

Sattler

. The Assessment of Children: Cognitive Foundations. 5th ed. California: Sattler; 2008.

29.

Ekman

Rosenberg

. What the face reveals: basic and applied studies of spontaneous expression using the Facial Action Coding System (FACS). Oxford, UK: Oxford University Press; 1997.

30.

Craig

Hadjistavropoulos

Grunau

Whitfield

. A comparison of two measures of facial activity during pain in the newborn child. J Pediatr Psychol. 1994;19(3):305–318.

31.

Defrin

Lotan

Pick

. The evaluation of acute pain in individuals with cognitive impairment: a differential effect of the level of impairment. Pain. 2006;124(3):312–320.