Some pain behaviors appear to be automatic, reflexive manifestations of pain, whereas others present as voluntarily controlled. This project examined whether this distinction would characterize pain cues used in observational pain measures for children aged 4–12. To develop a comprehensive list of cues, a systematic literature search of studies describing development of children’s observational pain assessment tools was conducted using MEDLINE, PsycINFO, and Web of Science. Twenty-one articles satisfied the criteria. A total of 66 nonredundant pain behavior items were identified. To determine whether items would be perceived as automatic or controlled, 277 research participants rated each on multiple scales associated with the distinction. Factor analyses yielded three major factors: the “Automatic” factor included items related to facial expression, paralinguistics, and consolability; the “Controlled” factor included items related to intentional movements, verbalizations, and social actions; and the “Ambiguous” factor included items related to voluntary facial expressions. Pain behaviors in observational pain scales for children can be characterized as automatic, controlled, and ambiguous, supporting a dual-processing, neuroregulatory model of pain expression. These dimensions would be expected to influence judgments of the nature and severity of pain being experienced and the extent to which the child is attempting to control the social environment.
Success in communicating an experience of pain to others can be a vital social transaction when there are threats to personal safety [
Judgments of pain in others represent complex decisions influenced by both characteristics of observers and available information [
The purposes of the current study were (1) to compile a comprehensive list of all of the pain cues in all identified observational pain scales for children aged 4 to 12 years and (2) to establish whether the cues systematically varied in terms of whether they were perceived as automatic or controlled. The long-term objective is to develop a preliminary scale composed of items that reflect automatic, controlled, and ambiguous pain expression.
The systematic review was conducted according to PRISMA guidelines [
Relevant articles were chosen according to specific inclusion criteria. First, articles were screened by title to exclude articles that did not meet the required age range (4 to 12 years). Titles containing the terms “infant,” “toddler,” “adolescent,” “adult,” and “elderly” were deleted, although titles that included the word “child” or a synonymous term were kept for further review. Articles with ambiguous titles were retained for further screening. Second, articles not published in peer-reviewed journals were excluded to maintain the empirical integrity of the included articles, eliminating commentaries, dissertations/theses, and book chapters. Review articles were retained as potential sources of relevant scales through the “snowball method” that is effective to identify obscure articles that might otherwise be overlooked [
A total of 21 articles that met the criteria were found in the literature search [
The 21 articles yielded 368 cues, but these often were unclear and some represented combination items, as noted below. Compiling a list of independent items referring to pain behaviors was accomplished using the following criteria. (i) Combined/composite behaviors were separated into individual items. For example, the Noncommunicating Children’s Pain Checklist-Revised [
In order to eliminate redundant items, five raters independently reviewed the list of 381 items. The raters were instructed to nominate items for exclusion that were identical, closely related, or synonymous (e.g., “grimace” and “grimacing”), vague or nondescriptive (e.g., “altered expression”), and referring only to an emotion (e.g., “angry”), incorporated features of other items (e.g., “reluctant to smile” as an element of the item “inconsolable”), and only quantified the behavior in terms of time (e.g., “grimace for less than 50% of the time” and “long grimace > 50% of time”). Cues with 100% agreement for inclusion among the raters were retained. Cues with 20–80% agreement concerning retention were discussed until 100% consensus concerning inclusion or exclusion was reached.
To determine the degree of automaticity of the observational pain cues, an online survey was undertaken. Participants were instructed to picture a child between the ages of 4 and 12 years who was experiencing pain and exhibiting a wide range of painful actions. The participant was instructed to rate each pain behavior cue according to their position on seven 10-point Likert subscales using anchors developed by Moors and De Houwer [
The survey was posted on the online hosting service Mechanical Turk (
The study protocol was approved by the Behavioral Research Ethics Board at the University of British Columbia. Written informed consent was obtained from participants before beginning the study.
Application of the criteria described above yielded a list of 66 cues. These items appear in Tables
Final list of pain behaviors extracted from pediatric observational pain scales that loaded on factor 1, which we labeled as the “Controlled” factor.
Pain behavior | Factor 1 | Factor 2 | Factor 3 | Factor 4 |
---|---|---|---|---|
Taking medication when he/she normally refuses | | | | |
Seeking physical comfort or closeness | | | | |
Pulling head away | | | | |
Lying very still | | | | |
Screaming | | | | |
Shifting torso | | | | |
Making aggressive physical contact | | | | |
Resisting being moved | | | | |
Cursing | | | | |
Raising voice | | | | |
Asking for help | | | | |
Complaining of pain | | | | |
Self-harming | | | | |
Angry verbalizations | | | | |
Whining | | | | |
Quiet | | | | |
Kicking | | | | |
Resistant | | | | |
Final list of pain behaviors extracted from pediatric observational pain scales that loaded on factor 2, which we labeled as the “Automatic” factor.
Pain behavior | Factor 1 | Factor 2 | Factor 3 | Factor 4 |
---|---|---|---|---|
Socially withdrawn | | | | |
Wincing | | | | |
Deepening of nasolabial furrow | | | | |
Whimpering | | | | |
Flailing arms and legs | | | | |
Grimacing | | | | |
Hard to console or comfort | | | | |
Watery eyes | | | | |
Tensing up | | | | |
Restless | | | | |
Squeezing eyes | | | | |
Grinding teeth | | | | |
Widening eyes | | | | |
Hyper alert | | | | |
Irregular, indrawn breathing | | | | |
Crying | | | | |
Tears running down face | | | | |
Squinting | | | | |
Moaning | | | | |
Shivering torso | | | | |
Looking pale | | | | |
Gasping | | | | |
Has disturbed sleep | | | | |
Final list of pain behaviors extracted from pediatric observational pain scales that loaded on factor 3 and/or factor 4, loaded on two or more factors, or did not load on any factors. This factor was labeled as the “Ambiguous” factor.
Pain behavior | Factor 1 | Factor 2 | Factor 3 | Factor 4 |
---|---|---|---|---|
Reduction of activity | | | | |
Puckering | | | | |
Rear up body of the trunk/sit up | | | | |
Jerking | | | | |
Furrowing brow | | | | |
Opening mouth | | | | |
Eating less than usual | | | | |
Flexion or tightening of toes | | | | |
Localizing pain | | | | |
Arching back | | | | |
Thrusting tongue out | | | | |
Rubbing | | | | |
Grunting | | | | |
Chewing | | | | |
Looking more flushed than normal | | | | |
Irritable | | | | |
Corners of mouth downward | | | | |
Eyes almost closed | | | | |
Being difficult to distract | | | | |
Playing less than usual | | | | |
Drawn up with arms and legs to the body | | | | |
Flexion or tightening of fingers | | | | |
Clenching jaw | | | | |
Protecting/favoring/guarding part of body that hurts | | | | |
Squirming | | | | |
A total of 296 participants completed the online survey. The data were screened for duplicate IP addresses and for nonsensical responses (e.g., the same response on all of the Likert scale for all questions). This resulted in the exclusion of the data of 19 participants, leaving 277 complete data sets. To prepare the data, three of the automaticity subscales were reverse-scored and the average of all seven scales was calculated in line with methods utilized in previous research [
The majority of participants were male (65%). Participant mean age was 24 years and ranged from 18 to 65 years. All were fluent in English. The majority of participants cited English as their first language (
To determine the factorial structure of these ratings, an Exploratory Factor Analysis was conducted using Unweighted Least Squares (ULS) extraction with a Direct Obliman [
Pain behavior items categories according to the factor analysis.
A considerable diversity of actions has been identified as signifying pain in children, including behaviors that could be characterized as verbal (e.g., “asking for help,” “complaining of pain,” and “cursing”), facial activity (e.g., “wincing,” “furrowed brow,” and “widening eyes”), nonverbal vocalizations (e.g., “whimpering,” “crying,” and “moaning”), limb action (e.g., “flailing arms and legs,” “rubbing,” and “protecting/favoring/guarding part of body that hurts”), body action (e.g., “tensing up” and “restless”), physiological manifestations (e.g., “looking pale,” “irregular breathing,” and “shivering torso”), and social behaviors (e.g., “withdrawn,” “hard to console,” and “angry verbalizations”). While there were commonalities across scales in the general categories employed, there was also considerable diversity in how actions in the different domains were described.
Descriptions of methods used by authors of the original papers to select items for inclusion in their scales referred primarily to expert judgment, either provided by the authors of the scales themselves or surveys of others who spend time with children in pain, for example, nurses or parents. Exceptions to personal nomination of items or the use of items already described in the literature were studies that used direct observation of children (e.g., [
There is room for improvement in descriptions of items. Chang et al. [
We note that on some scales certain cues function on their own, whereas on other scales these same cues are subcategories of other cues. For example, the Dalhousie Everyday Pain Scale [
Weights assigned to the different pain behavior items may be problematic. Scales that add up the number of endorsed items to provide an index of pain severity de facto assign equivalent weights to items. It would seem that items that are only sensitive but not specific to pain (e.g., manifestations of aversive, but nonnoxious states) should receive less weight than both specific and sensitive items. Scales that include self-report items may artificially lower scores in children with cognitive or communication impairment if weight is attached to these items.
The scales provide a mixture of automatic and controlled items, with other items ambiguous with respect to these categories. Research participants consistently identified variations in these qualities in the behavioral cues. Automatic items are perceived to be more reflexive, unconscious, nondeliberate, and stimulus driven, whereas controlled items are construed as reflecting conscious experience, deliberation, and effort to achieve some specific purpose [
Most facial expressions were perceived to be automatic, which is consistent with the facial expression literature [
The controlled behaviors were mostly comprised of verbalizations and instrumental motor activities. In this case, the motor behavior was distinct from the automatic reflexive behaviors, as it was characterized by extended duration and coordinated action [
Twenty-five cues were rated as ambiguous, indicating that the source of action is difficult to discern. In some cases, it was unclear whether the cue referred to a social behavior or motor activity. For example, “reduction of activity” was rated as ambiguous, possibly due to the lack of context. As a result, the cue could be seen as either a reflexive or controlled behavior. In any given scenario, context is important in drawing conclusions. In a clinical setting this might be more readily resolved.
Less commonly observed cues would appear to have greater corresponding ambiguity ratings. Certain cues, such as “eating less than usual” and “playing less than usual,” are not commonly found in pain scales and are not normally used as direct indicators of pain. Similarly, most facial cues were identified as automatic, although some, for example, “puckering” and “chewing” were found to be ambiguous. These cues tend not to be reported in the descriptive empirical literature on children’s pain facial expressions and are also not prevalent in pain scales. Some cues were inherently unclear, such as “rear up body of the trunk/sit up.” Cultural backgrounds also need to be considered when explaining variations in interpretation of items, given research indicating that attitudes, beliefs, and psychological states of ethnic groups can affect the observational judgment of pain [
The substantial number and diversity of behaviors nominated as expressive of children’s pain should not be surprising given the salience and potency of painful experiences: mobilization of biobehavioral resources in all their complexity would be anticipated. What perhaps is more important in scale development is selective refinement in identifying those cues that are the most sensitive and specific to pain. Items referring to pain specific behaviors should be favored. The behaviors listed provide numerous illustrations of actions likely to be sensitive, but not specific to pain, such as “crying.” These play an important role as they attract attention to the child and precipitate a search for more explicit information concerning the nature of the child’s distress, for example, pain versus fear, anger, or irritability.
There is also merit in pursuing an understanding of the underlying structure of pain expression, with the functional value of the actions of particular importance. The functional distinction between protective and communicative actions has already been described. Nociceptive reflexes, guarded behaviors, or protective postures, for example, are useful sources of information about pain, but their communicative value would appear to be secondary to their functional value in avoiding or diminishing painful experience. This paper confirms that the dual neuroregulatory processes associated with automatic and controlled actions provide implicit structure for observer reactions to children’s behavioral cues of pain. Automatic/reflexive actions are more likely to carry specific information concerning pain in fulfilling protective and communicative functions without the use of conscious deliberation, whereas controlled reactions represent efforts to cope with painful events, either through protective behavior or social communication requiring executive functions or higher levels of neuroregulatory systems [
The present study is not without limitations. The country of residence for the majority of the sample recruited on Mechanical Turk was India (71.9%). Although previous studies using Mechanical Turk estimated that 70–80% of registered users live in the United States and provide a more representative sample of Americans than traditional undergraduate student samples [
Furthermore, participants were asked to picture a child between the ages of 4 and 12 displaying the list of pain behavior cues. It is possible that participants may have imagined the expression/display of pain differently, which may have influenced the subsequent ratings of automaticity for the behavior. Future research should consider using visual stimuli, such as pictures or videos of children displaying the various pain cues to determine if the pain cues would have the same underlying factor structure.
The present findings may help inform the development of a new observation pain scale to improve pain assessment in children. Although it is possible that items with higher factor loadings may be good candidates for inclusion on a new scale, it would be important to evaluate the validity and reliability of each pain cue to determine whether each item can discriminate between a child experiencing pain from a child not experiencing pain or another negative affective state. Although we suspect that some items will be more likely to have higher sensitivity and specificity than other items (e.g., cues that refer to specific behaviors, describing facial expressions), further clinical research is required to confidently make this distinction.
A portion of the research in this manuscript was presented as a poster at the Canadian Pain Society Conference in May 2013. An abstract of the poster was published in
The authors declare that there is no conflict of interests regarding the publication of this paper.
This research was funded by a Social Sciences and Humanities Research Council of Canada grant. Samantha Fashler is supported by a Canadian Institutes of Health Research (CIHR) Vanier Canada Graduate Scholarship.