- Email: [email protected]
Measuring pain accurately in children with cognitive impairments: Refinement of a caregiver scale
Measuring pain accurately in children with cognitive impairments: Refinement of a caregiver scale Lynn M. Breau, BA, Carol Camfield, MD, FRCPC, Patrick J. McGrath, PhD, Christina Rosmus, RN, MSc, and G. Allen Finley, MD, FRCPC
Objective: To examine whether typical pain behavior, as reported by caregivers, could be used prospectively to predict future pain behavior and to derive a subset of core items from the Non-Communicating Children’s Pain Checklist. Study design: Caregivers (n = 33) of children with cognitive impairments completed the Non-Communicating Children’s Pain Checklist retrospectively and immediately after subsequent episodes of pain and distress in their homes. Odds ratios were computed for checklist items, and multiple regressions were used to predict numerical pain and distress ratings with items that had significant odds ratios. A logistic regression was used to test whether the items found to predict pain could correctly classify the presence or absence of pain in a new cohort of 63 children with similar cognitive impairments. Results: Seven of the checklist items had significant odds ratios: Cranky, Seeking Comfort, Change in Eyes, Less Active, Gesture to Part That Hurts, Tears, and Gasping. This subset of items significantly predicted numerical pain ratings by caregivers (multiple R = .70), but not distress ratings (multiple R = .31). In a second group of 63 children with cognitive impairments, this subset of items displayed 85% sensitivity and 89% specificity for pain. Conclusion: A subset of items from the Non-Communicating Children’s Pain Checklist could predict pain in children with cognitive impairments. Caregivers’ retrospective reports may be useful for clinicians making judgments about pain in these children. (J Pediatr 2001;138:721-7)
From the Departments of Psychology, Pediatrics, Psychiatry, Biomedical Engineering, and Anaesthesia, Dalhousie University and Pediatric Pain Research Lab, IWK Health Centre, Halifax, Nova Scotia, Canada.
Supported in part by a Student Fellowship granted to Lynn Breau and a Distinguished Scientist Award granted to Patrick J. McGrath, PhD, by the Medical Research Council of Canada. Partial support was also provided by The Toronto Hospital for Sick Children Foundation. Presented in part at the Annual Convention of the Canadian Pain Society (1998), Regina, Canada, and the Child Neurology Society (1998), Montreal, Canada. Submitted for publication Mar 29, 2000; revisions received July 25, 2000, and Sept 29, 2000; accepted Oct 11, 2000. Reprint requests: Lynn Breau, Pediatric Pain Research Lab, IWK Grace Health Centre, 5850 University Ave, Halifax, Nova Scotia, Canada, B3J 3G9. Copyright © 2001 by Mosby, Inc. 0022-3476/2001/$35.00 + 0 9/21/112247 doi:10.1067/mpd.2001.112247
Self-report is the best method of assessing an individual’s pain. For some groups, such as infants, adults with advanced dementia, and people with severe cognitive impairments, self-report can be problematic. VAS
Visual analog score
There has been a growing body of research into the pain of infants,1 but very little is known about the pain of individuals with severe cognitive impairments. Many people with cognitive impairments also have neurologic impairments that alter their behavior and physiology. Some individuals with cognitive impairments may also appear insensitive to pain,2,3 although this may merely reflect an inability to communicate their experience,4 observers’ inability to recognize their pain signals,5,6 or uncertainty of pain behaviors seen.7 In addition, preliminary reports suggest that physiologic, behavioral, and facial activity measures designed for those without impairments may not be sensitive to pain in those with cognitive impairments.8-10 In contrast, development of measures specifically for those with impairments has progressed.11,12 Although a preliminary validation of a measure by McGrath et al13 suggests that their tool exhibits some preliminary reliability and validity, its length makes it cumbersome. This research examines data collected as part of that validation study in an effort to determine whether a subset of items could be abstracted to reduce the size of the checklist while retaining its sensitivity and specificity. It also ex721
BREAU ET AL
THE JOURNAL OF PEDIATRICS MAY 2001
Table I. Demographic characteristics of samples 1 (n = 33) and 2 (n = 63)
Characteristic Participants’ age (y) Communicative age equivalents (mo) Language Production Language Comprehension Gestural Ability Participants’ sex Male Female Caregivers’ relation to child Parent/foster parent Health care provider Grandparent Diagnoses* Cerebral palsy Moderate to severe developmental delay/ autism (no neurologic diagnosis) Developmental malformation (micrencephaly, hydrocephalus, agenesis of the corpus callosum) Syndrome (eg, Sanfillipo, Smith-Lemi Opitz, Aicardi, Angelman) Traumatic brain injury, meningitis, perinatal complications, infantile spasms, tumor Other (neurodegenerative disorder, metabolic disorder) Seizure disorder
Sample 1 (mean ± SD)
Sample 2 (mean ± SD)
14 ± 11
11 ± 4
10 ± 3 12 ± 4 10 ± 3
12 ± 7 12 ± 3 11 ± 3
67% 33%
57% 43%
64% 36% 0%
81% 17% 2%
55% 25%
43% 22%
6%
18%
6%
11%
18%
7%
0%
13%
33%
68%
*Note: Sample 1: based on caregiver report. Sample 2: based on review of medical chart. Diagnoses are not exclusive; percentages may sum to greater than 100%.
amines whether the results can be replicated with a different sample in an effort to determine its generalizability.
METHODS Participants SAMPLE 1. Thirty-six caregivers of individuals with cognitive impairments and with no ability to communicate verbally completed the materials described as part of a larger study examining the validity and test-retest reliability of total checklist scores for two painful events.13 The study also investigated the ability of the checklist to discriminate between pain and calm and between pain and distress based on total checklist scores. No analyses 722
of individual checklist items were conducted as part of that report. All caregivers were female, spoke English, and had been the primary caregiver of the participants for at least 6 months. They were contacted through the child neurology division of a regional children’s hospital, support groups, and residential centers. The data from 3 caregivers were excluded because their description of the painful incident was not sufficient. Thus the analyses were conducted on the reports of 33 caregivers. SAMPLE 2. A second group of 63 caregivers and their children with cognitive impairments and severe verbal deficits was used to replicate the results obtained with the first group. All
were recruited through the child neurology division of a regional children’s hospital and were part of a separate cohort currently involved in a longitudinal study of pain.
Measures Caregivers completed The MacArthur Communicative Development Inventories (MCDI): Words and Gestures or Words and Sentences.14 This is a measure of a person’s ability to comprehend language and to communicate verbally and with gestures. The inventory provides age equivalents for Language Comprehension, Language Production, and Gestures. Three “versions” of the Non-Communicating Children’s Pain Checklist12 were also completed by the caregivers of Sample 1. The first asked them to indicate whether the items, phrased in the past tense, had occurred in the past during pain. The remaining two were identical except for instructions indicating whether they were to be completed during “pain” or “distress,” and the word “pain” in some items of the version to be completed during pain was also replaced with the word “distress” in the version used during distress. These were completed immediately after actual incidents. One was an incident in which it was clear the participant had pain (Pain Incident). The other was an incident during which the participant was distressed but clearly had experienced no pain (Distress Incident). Caregivers also completed a visual analog scale of pain or distress intensity for the incidents and provided written descriptions of each incident.
Procedure This research was approved by the Research Ethics Boards of the IWK Grace Health Centre and of Dalhousie University. Informed consent was obtained from all caregivers. SAMPLE 1. Each caregiver was mailed the measures described, and they were telephoned after receiving them to en-
BREAU ET AL
THE JOURNAL OF PEDIATRICS VOLUME 138, NUMBER 5 sure they understood each questionnaire. All materials were returned by mail. Caregivers completed the retrospective checklist when they received the package. They then waited for a painful and a distressful incident to occur to complete the two additional checklists. SAMPLE 2. The second group of caregivers completed a 1-week diary that included the Non-Communicating Children’s Pain Checklist as part of their participation in a larger 2-year study. The information collected from day 1 of the diary was used in these analyses.
Statistical Analyses All data were analyzed by using SPSS 8.0 software.15 Descriptive statistics were generated for the characteristics of the participants (means, SDs, percentages). Relations between pain intensity and demographic characteristics of Sample 1 were made with Spearman correlations for ranked data, whereas correlations with categorical characteristics were conducted by using Eta coefficients. Comparisons of children with or without pain in Sample 2, based on noncategorical demographic characteristics, were conducted by using Mann-Whitney U tests for nonparametric data. To examine the predictive utility of each item based on caregiver retrospective report, odds ratios were computed, and comparisons of the occurrence of items were conducted by using matched-sample t tests. Multiple linear regression and logistic regression were used to examine the sensitivity and specificity of items with significant odds ratios. Significance was set at α = .05 for all tests, and Bonferroni corrections were used when multiple tests were conducted.
RESULTS Demographic Information SAMPLE 1. The demographic characteristics of the participants in Sample 1 are shown in Table I. All had severe cogni-
Table II. Pain experienced by sample 2 (n = 63) as a function of level of limb function
Level of limb function Upper limbs Full use Limitations No use Lower limbs Full use Limitations No use
tive delays and a wide range of physical impairments. VAS pain ratings for the episode observed were not related to the children’s age, Language Comprehension, Language Production, gestural ability, or physical impairments. Pain intensity was also not related to sex or pain type (acute, long-term). SAMPLE 2. The demographic characteristics of Sample 2 are also shown in Table I. All had severe cognitive impairments. According to their caregivers, 40% of the children had full upper limb function and 25% had full lower limb function, whereas 14% had no upper limb use and 41% had no lower limb use. In addition, 63% had some visual impairment and 17% had some hearing impairment. Children with limited or no use of their upper limbs (Mann-Whitney U = 226, P < .01) or lower limbs (Mann-Whitney U = 211, P < .01) were more likely to experience pain on the day the diary was completed than children with full upper or lower limb function (Table II).
Descriptions of the Painful and Distressful Incidents Caregivers’ descriptions of the painful and distressful incidents were reviewed to verify that evidence of pain or distress was present for the time they completed the checklists. Descriptions of pain included a bee sting, burns, respiratory tract infections, post-surgical and procedural
No pain on day 1 of diary (n = 45)
Pain on day 1 of diary (n = 18)
51% 40% 9%
11% 61% 28%
36% 32% 31%
0% 33% 67%
pain, and injuries. Descriptions of distress included the child being denied a pleasurable activity and events the child considered aversive (eg, bath, hair brushed/washed, fearful reactions to dogs/loud noises). VAS pain ratings for the pain episode ranged from 0 to 10 (mean = 6.6, SD = 2.4). VAS ratings of distress ranged from 3 to 10 (mean = 6.6, SD = 2.0). Eight of the 30 items of the Noncommunicating Pain Checklist were omitted from these analyses for the following reasons: (1) a response rate of <30% for the pain incident (Increase in sleep, Floppy, Shivering, Breath holding), (2) report during long-term pain only (Eat less, Decrease in sleep), (3) unsuitability for quantitative analyses (Specific sound for pain/distress, Specific movement for pain/distress).
Odds Ratios for the Checklist Items To determine whether caregivers’ retrospective reports of the items from the checklist could predict their reoccurrence during the subsequent pain incident, odds ratios were computed for each item. Odds ratios for which the CI does not include 1.0 are considered significant.16 The greatest odds of occurring were for Seeking Comfort, Gesturing, and Tears (Table III). In all, caregivers’ retrospective reports were a good indication that 7 items would re-occur, although the large CIs suggest there was large variability 723
BREAU ET AL
THE JOURNAL OF PEDIATRICS MAY 2001
Table III. Odds ratios for items of the Non-Communicating Children’s Pain Checklist occurring during a pain incident based on caregivers’ retrospective report (n = 33).
Pain incident Checklist items
Odds ratio
Vocal behavior Moaning, whining, whimpering (fairly soft) Crying (moderately loud) Screaming/yelling (very loud) Social Not co-operating, cranky, irritable, unhappy* Less interaction with others; withdrawn Seeks comfort or physical closeness* Difficult to distract, not able to satisfy or pacify Facial expression Furrowed brow Change in eyes* including squinching of eyes, eyes open wide, eyes frown Turn down of mouth, not smiling Lips pucker up, tight, pout or quiver Clenches or grinds teeth, chews; thrusts tongue out Activity Not moving, less active, quiet* Jumping around, agitated, fidgety Body and limbs Stiff, spastic, tense rigid Gestures to or touches part of body that hurts* Protects, favors, or guards part of body that hurts Flinches or moves body part away; sensitive to touch Physical signs Change in color; pallor Sweating; perspiring Tears* Sharp intake of breath; gasping*
95% CI
2.6 7.0 3.3
(.42–15.92) (.74–65.94) (.63–16.79)
12.0 2.8 42.0 1.4
(1.23–117.41) (.58–13.31) (4.12–428.66) (.33–6.17)
0.5
(.09–2.52)
8.9 3.8 2.0 5.3
(1.34–58.80) (.64–22.04) (.41–9.71) (.99–27.90)
9.2 2.8
(1.49–56.30) (.58–13.31)
1.8 38.0 3.7 3.7
(.35–8.80) (3.31–436.94) (.70–19.59) (.77–17.43)
2.3 6.0 23.8 6.5
(.23–23.91) (.54–67.28) (2.34–242.91) (1.20–35.57)
*Items with significant odds ratios.
among the children. This indicates that the 7 items are more consistently shown across episodes of pain than their non-significant counterparts.
Incidence of Checklist Items During Pain The previous analyses indicated that the 7-item subset was displayed consistently over time but did not examine whether these 7 items were displayed by many of the children, making them suitable candidates for a standardized instrument, or whether some items did not achieve significance because they 724
were only displayed during pain by a small number of the children. To investigate this, the number of items with significant odds ratios (Odds-High) and the number with non-significant odds ratios (Odds-Low) displayed by the children during pain were compared (Table IV). The results (t(32) = –.09, P = .93) indicated that there were equal numbers of Odds-High behaviors (mean = .44, SD = .30) and OddsLow behaviors shown (mean = .44, SD = .19). This indicates that low predictive ability was not due to low incidence of Odds-Low items and confirms
that high predictive ability was due to the consistency of the Odds-High items (as measured by the predictive utility of caregiver retrospective report), not their greater incidence across children during the pain episode.
Sensitivity and Specificity to Pain To confirm the sensitivity of the set of 7 Odds-High items to pain, a multiple regression was conducted. Because it was not expected that any one item would predict pain ratings independently, the set of 7 items was entered in one step, with caregivers’ VAS pain ratings as the dependent variable. This produced an R value of .70, which accounted for a significant amount of the variance in caregivers’ VAS ratings (F (7,22) = 3.1, P < .05). Five items were positively related to pain ratings: Cranky, Seeking Comfort, Change in Eyes, Less Active, and Tears. Two items, Gestures and Gasp, were negatively related to pain intensity. This provides further evidence that the 7item set is sensitive to pain. To test the discriminative ability of these items, a second multiple regression was performed with caregivers’ ratings of distress intensity during the distressful incident as the dependent variable. Again, all 7 items were entered in one step. This resulted in multiple R value of .31, which was non-significant (F(7,23) = .34, P > .05). Thus occurrence of the set of 7 items could not predict the intensity of distress, indicating that the 7-item set is specific to pain.
Replication As a final test of the ability of this subset to detect pain, analyses were conducted to determine whether these 7 items could classify a new group of 63 children with cognitive impairments as having pain or not having pain on a given day. Eighteen of the participants had pain on the day examined, whereas 45 had no pain that day. The mean pain rating for those who had pain, on a scale of 0 to 100, was 65 (SD = 37).
BREAU ET AL
THE JOURNAL OF PEDIATRICS VOLUME 138, NUMBER 5
Table IV. Occurrence of items from the Non-Communicating Children’s Pain Checklist during an observed pain incident as a function of retrospective report (n = 33)
Percentage of caregivers who observed item during pain incident Checklist items Vocal behavior Moaning, whining, whimpering (fairly soft) Crying (moderately loud) Screaming/yelling (very loud) Social Not cooperating, cranky, irritable, unhappy Less interaction with others; withdrawn Seeks comfort or physical closeness Difficult to distract, not able to satisfy or pacify Facial expression Furrowed brow Change in eyes, including squinching of eyes, eyes open wide, eyes frown Turn down of mouth, not smiling Lips pucker up, tight, pout or quiver Clenches or grinds teeth, chews; thrusts tongue out Activity Not moving, less active, quiet Jumping around, agitated, fidgety Body and limbs Stiff, spastic, tense rigid Gestures to or touches part of body that hurts Protects, favors, or guards part of body that hurts Flinches or moves body part away; sensitive to touch Physical signs Change in color; pallor Sweating; perspiring Tears Sharp intake of breath; gasping
Four children were missing scores for at least one item. Logistic regressions were conducted both with and without replacing those data, with no difference in significance. Thus the results obtained without replacing the missing values are presented below (16 children with pain and 43 without pain). All 7 items were entered in one step. The χ2 test indicated that 53% of variation in pain/no pain scores was accounted for (χ2 = 25.3, P < .001) and that the 7-item subset could correctly classify 11 (69%) of the 16 children who had pain
Item reported retrospectively
Item not reported retrospectively
Of total sample
86% 92% 67%
14% 8% 33%
68% 41% 30%
94% 60% 92% 62%
6% 40% 8% 38%
55% 30% 42% 43%
53% 84%
47% 16%
63% 70%
79% 50% 64%
21% 50% 36%
70% 36% 41%
85% 60%
15% 40%
33% 33%
64% 67% 67% 73%
36% 33% 33% 27%
44% 31% 31% 52%
83% 75% 94% 70%
17% 25% 6% 30%
21% 14% 60% 34%
and 41 (95%) of the 43 who did not have pain. Two children were predicted to have pain and did not. Five children were predicted not to have pain and did. One method of expressing the relation between sensitivity and specificity of a diagnostic test is to calculate the “predictive utility” of a positive or negative result.17 The current results suggest that 85% of those predicted to have pain did, in fact, have pain, indicating high positive predictive value or sensitivity, and 89% of those predicted not to have pain did not have pain, indicating high nega-
tive predictive value or specificity. The percentage of children displaying each of the 7 behaviors is shown in Table V. These results suggest that the 7-item subset derived with one sample could detect pain in a separate cohort of children with similar impairments and provide evidence that the items are generalizable.
DISCUSSION These findings highlight the fact that although many behaviors and signs 725
BREAU ET AL
THE JOURNAL OF PEDIATRICS MAY 2001
Table V. Percentage of participants displaying checklist items in sample 2 (n = 63)
Percentage displaying each item on day 1 of diary Checklist items Not cooperating, cranky, irritable, unhappy Seeks comfort or physical closeness Change in eyes, including: squinching of eyes, eyes open wide, eyes frown Not moving, less active, quiet Gestures to or touches part of body that hurts Tears Sharp intake of breath; gasping
may occur during any given episode of pain, only some are sufficiently reliable over time to predict pain and to be used for its identification. It is also significant that the 7 items, derived from one sample, could distinguish between the presence and absence of pain in a larger cohort. Replication with a new sample is essential to showing that a solution can be generalized to others in a similar population. One unique aspect of this study was that the items found to predict pain were based on caregivers’ retrospective reports, rather than on observations of previous discrete episodes. This means that their reports could be used as a guide by other observers. Although not directly tested, it is possible that use of the checklist assisted caregivers in providing more accurate information regarding their children’s previous pain behavior. Another important finding was that the subset predicted pain intensity ratings, but not distress intensity ratings, confirming their specificity to pain. Most intriguing, however, was the fact that an increase in 5 items (Cranky, Seeking Comfort, Change in Eyes, Less Active, and Tears) but a decrease in 2 (Gestures and Gasping) predicted greater pain intensity. One explanation for the decrease in Gestures and Gasping with more intense pain is that they both necessitate moving, which may 726
With pain (n = 18)
Without pain (n = 45)
65% 59% 65%
38% 37% 27%
35% 18% 62% 56%
20% 0% 18% 7%
exacerbate pain. The pain experienced by these children was particularly intense, with an average pain rating of 6.6 (± 2) of 10, a level usually considered “moderate.”18 Therefore it is possible that the difference in Gestures and Gasping found here was due to the high intensity of pain felt by this group. Another possible explanation for the decrease in Gestures lies in the complexity of executing that task. The children in this study had the gestural ability of infants 8 to 16 months of age and cognitive impairments that could make organizing and orchestrating such a behavior difficult. Many also had physical limitations that would add to the difficulty. Intense pain may have compounded this difficulty, because stressful situations can interfere with cognitive functioning.19 The reduction in Gasping is more difficult to explain, and contradictory findings have been reported. Several studies have reported that breathing pattern20 and rate21 change with pain in infants. However, there is currently no consensus on which changes are most likely to signify pain. Craig et al22 reported a reduction in breathing rate during pain and raised the possibility of Breath holding. Breath holding is one of the 30 items in the original NonCommunicating Children’s Pain Checklist. A review of the observations for this study indicated, however, that
only 2 children were reported to have held their breath during pain. It could be that caregivers were unable to recognize this behavior and that a reduction in short, sharp breaths and gasping actually reflects the fact that the participants were holding their breath. One limitation of the studies was the heterogeneity of the groups in terms of diagnosis. Another limitation of this study was that the observations on which odds ratios of an item occurring were computed were made by the same caregivers who completed the retrospective report. However, the fact that the subset did so well in classifying the second group of children experiencing pain provides evidence that the items can be generalized to other children and their caregivers. In addition, the CIs of the odds ratios generated for individual items were large, suggesting that the predictive ability of single items may vary somewhat from individual to individual. It is notable that the 7 items were from a variety of subscales including the Social, Facial, Activity, Body and Limb, and Physical Signs Subscales. This suggests that the usefulness of caregivers’ retrospective reports is not limited to one class or type of behavior. It is also interesting that no items from the Vocal Behavior Subscale were predictive of pain. This could reflect the fact that these children do not display much vocal behavior during pain, that their vocal behavior is less consistent than other pain behaviors, or that caregivers are less observant of or less reliant on their vocal behavior during pain. Regardless, the fact that no vocal behaviors were significant supports the premise that pain tools designed for individuals without cognitive impairments, which generally weight verbal behavior strongly,23 are not appropriate for this special group and underscores the need for development of pain assessment tools specifically designed for these children. These findings bode well for the development of pain assessment tools for
BREAU ET AL
THE JOURNAL OF PEDIATRICS VOLUME 138, NUMBER 5 children who cannot communicate verbally because of cognitive impairments. They also negate the common belief that the pain behavior of those with cognitive impairments may be too idiosyncratic to be measured by standardized instruments. As with development of any diagnostic instrument, these results should be regarded as only partial evidence of validity. Further research with different observers and in variety of settings is required before formal use of this tool can be recommended. The authors would like to recognize the contribution of the parents and caregivers who participated. Without their assistance, this research would not have been possible.
6.
7.
8.
9.
10.
REFERENCES 1. Anand KJS, Stevens B, McGrath PJ, editors. Pain in neonates. 2nd revised and enlarged ed. Amsterdam: Elsevier; 2000. 2. Biersdorff KK. Pain insensitivity and indifference: alternative explanations for some medical catastrophes. Ment Retard 1991;29:359-62. 3. Biersdorff KK. Incidence of significantly altered pain experience among individuals with developmental disabilities. Am J Ment Retard 1994;98: 619-31. 4. Oberlander TF, O’Donnell ME, Montgomery CJ. Pain in children with significant neurological impairment. Dev Behav Pediatr 1999;20:235-43. 5. Roy A, Simon GB. Intestinal obstruction as a cause of death in the mentally
11.
12.
13.
14.
handicapped. J Ment Defic Res 1987; 31:193-7. Jancar J, Speller CJ. Fatal intestinal obstruction in the mentally handicapped. J Intellect Disabil Res 1994;34:413-22. Parmelee PA, Smith B, Katz IR. Pain complaints and cognitive status among elderly institution residents. J Am Geriatr Soc 1993;41:517-22. Porter FL, Malhotra KM, Wolf CM, Morris JC, Miller JP, Smith MC. Dementia and response to pain in the elderly. Pain 1996;68:413-21. Oberlander TF, Gilbert CA, Chambers CT, O’Donnell ME, Craig KD. Biobehavioral responses to acute pain in adolescents with a significant neurologic impairment. Clin J Pain 1999;15:201-9. Koh JL, Fanurik D. Postoperative pain assessment in children with cognitive impairment. Presented at the 16th Annual Meeting of the American Pain Society; 1997 Oct 23-27; New Orleans, La. Giusiano B, Jimeno MT, Collignon P, Chau Y. Utilization of a neural network in the elaboration of an evaluation scale for pain in cerebral palsy. Methods Inf Sci 1995;34:498-502. McGrath PJ, Rosmus C, Campbell MA, Camfield C, Hennigar A. Behaviors caregivers use to determine pain in nonverbal, cognitively impaired children. Dev Med Child Neurol 1998;40:340-3. Breau LM, McGrath PJ, Camfield C, Rosmus C, Finley GA. Preliminary validation of an observational pain checklist for cognitively impaired, nonverbal persons. Dev Med Child Neurol 2000;42:609-12. Fenson L, Dale PS, Reznick JS, Thal D, Bates E, Hartung JP, et al.
15. 16.
17.
18.
19.
20.
21.
22.
23.
MacArthur Communicative Development Inventories: user’s guide and technical manual. San Diego: Singular Publishing Group, Inc; 1993. SPSS version 8 [computer software]. Chicago: SPSS Inc; 1998. Howell DC. Statistical methods for psychology. 3rd ed. Belmont (CA): Duxbury Press; 1992. Browner WS, Newman TB, Cummings SR. Designing a new study: III. Diagnostic tests. In: Hulley SB, Cummings SR, editors. Designing clinical research: an epidemiological approach. Baltimore: Williams & Wilkins; 1988. p. 87-97. Collins SL, Moore RA, McQuay HJ. The visual analogue pain intensity scale: what is moderate pain in millimetres? Pain 1997;72:95-7. Champion GD, Goodenough B, von Baeyer CL, Thomas W. Measurement of pain by self-report. In: Finley GA, McGrath PJ, editors. Measurement of pain in infants and children. Seattle: IASP Press; 1998. p. 123-60. McIntosh N, van Veen L, Brameyer H. The pain of heel prick and its measurement in preterm infants. Pain 1993;52:71-4. Rawlings DJ, Miller PA, Engel RR. The effect of circumcision on transcutaneous PO2 in term infants. Am J Dis Child 1980;134:676-8. Craig KD, Whitfield MF, Grunau RVE, Linton J, Hadjistavropoulos HD. Pain in the preterm neonate: behavioral and physiological indices. Pain 1993;52:287-99. McGrath PJ. Behavioral measures of pain. In: Finley GA, McGrath PJ, editors. Measurement of pain in infants and children. Seattle: IASP Press; 1998. p. 83-102.
727
Objective: To examine whether typical pain behavior, as reported by caregivers, could be used prospectively to predict future pain behavior and to derive a subset of core items from the Non-Communicating Children’s Pain Checklist. Study design: Caregivers (n = 33) of children with cognitive impairments completed the Non-Communicating Children’s Pain Checklist retrospectively and immediately after subsequent episodes of pain and distress in their homes. Odds ratios were computed for checklist items, and multiple regressions were used to predict numerical pain and distress ratings with items that had significant odds ratios. A logistic regression was used to test whether the items found to predict pain could correctly classify the presence or absence of pain in a new cohort of 63 children with similar cognitive impairments. Results: Seven of the checklist items had significant odds ratios: Cranky, Seeking Comfort, Change in Eyes, Less Active, Gesture to Part That Hurts, Tears, and Gasping. This subset of items significantly predicted numerical pain ratings by caregivers (multiple R = .70), but not distress ratings (multiple R = .31). In a second group of 63 children with cognitive impairments, this subset of items displayed 85% sensitivity and 89% specificity for pain. Conclusion: A subset of items from the Non-Communicating Children’s Pain Checklist could predict pain in children with cognitive impairments. Caregivers’ retrospective reports may be useful for clinicians making judgments about pain in these children. (J Pediatr 2001;138:721-7)
From the Departments of Psychology, Pediatrics, Psychiatry, Biomedical Engineering, and Anaesthesia, Dalhousie University and Pediatric Pain Research Lab, IWK Health Centre, Halifax, Nova Scotia, Canada.
Supported in part by a Student Fellowship granted to Lynn Breau and a Distinguished Scientist Award granted to Patrick J. McGrath, PhD, by the Medical Research Council of Canada. Partial support was also provided by The Toronto Hospital for Sick Children Foundation. Presented in part at the Annual Convention of the Canadian Pain Society (1998), Regina, Canada, and the Child Neurology Society (1998), Montreal, Canada. Submitted for publication Mar 29, 2000; revisions received July 25, 2000, and Sept 29, 2000; accepted Oct 11, 2000. Reprint requests: Lynn Breau, Pediatric Pain Research Lab, IWK Grace Health Centre, 5850 University Ave, Halifax, Nova Scotia, Canada, B3J 3G9. Copyright © 2001 by Mosby, Inc. 0022-3476/2001/$35.00 + 0 9/21/112247 doi:10.1067/mpd.2001.112247
Self-report is the best method of assessing an individual’s pain. For some groups, such as infants, adults with advanced dementia, and people with severe cognitive impairments, self-report can be problematic. VAS
Visual analog score
There has been a growing body of research into the pain of infants,1 but very little is known about the pain of individuals with severe cognitive impairments. Many people with cognitive impairments also have neurologic impairments that alter their behavior and physiology. Some individuals with cognitive impairments may also appear insensitive to pain,2,3 although this may merely reflect an inability to communicate their experience,4 observers’ inability to recognize their pain signals,5,6 or uncertainty of pain behaviors seen.7 In addition, preliminary reports suggest that physiologic, behavioral, and facial activity measures designed for those without impairments may not be sensitive to pain in those with cognitive impairments.8-10 In contrast, development of measures specifically for those with impairments has progressed.11,12 Although a preliminary validation of a measure by McGrath et al13 suggests that their tool exhibits some preliminary reliability and validity, its length makes it cumbersome. This research examines data collected as part of that validation study in an effort to determine whether a subset of items could be abstracted to reduce the size of the checklist while retaining its sensitivity and specificity. It also ex721
BREAU ET AL
THE JOURNAL OF PEDIATRICS MAY 2001
Table I. Demographic characteristics of samples 1 (n = 33) and 2 (n = 63)
Characteristic Participants’ age (y) Communicative age equivalents (mo) Language Production Language Comprehension Gestural Ability Participants’ sex Male Female Caregivers’ relation to child Parent/foster parent Health care provider Grandparent Diagnoses* Cerebral palsy Moderate to severe developmental delay/ autism (no neurologic diagnosis) Developmental malformation (micrencephaly, hydrocephalus, agenesis of the corpus callosum) Syndrome (eg, Sanfillipo, Smith-Lemi Opitz, Aicardi, Angelman) Traumatic brain injury, meningitis, perinatal complications, infantile spasms, tumor Other (neurodegenerative disorder, metabolic disorder) Seizure disorder
Sample 1 (mean ± SD)
Sample 2 (mean ± SD)
14 ± 11
11 ± 4
10 ± 3 12 ± 4 10 ± 3
12 ± 7 12 ± 3 11 ± 3
67% 33%
57% 43%
64% 36% 0%
81% 17% 2%
55% 25%
43% 22%
6%
18%
6%
11%
18%
7%
0%
13%
33%
68%
*Note: Sample 1: based on caregiver report. Sample 2: based on review of medical chart. Diagnoses are not exclusive; percentages may sum to greater than 100%.
amines whether the results can be replicated with a different sample in an effort to determine its generalizability.
METHODS Participants SAMPLE 1. Thirty-six caregivers of individuals with cognitive impairments and with no ability to communicate verbally completed the materials described as part of a larger study examining the validity and test-retest reliability of total checklist scores for two painful events.13 The study also investigated the ability of the checklist to discriminate between pain and calm and between pain and distress based on total checklist scores. No analyses 722
of individual checklist items were conducted as part of that report. All caregivers were female, spoke English, and had been the primary caregiver of the participants for at least 6 months. They were contacted through the child neurology division of a regional children’s hospital, support groups, and residential centers. The data from 3 caregivers were excluded because their description of the painful incident was not sufficient. Thus the analyses were conducted on the reports of 33 caregivers. SAMPLE 2. A second group of 63 caregivers and their children with cognitive impairments and severe verbal deficits was used to replicate the results obtained with the first group. All
were recruited through the child neurology division of a regional children’s hospital and were part of a separate cohort currently involved in a longitudinal study of pain.
Measures Caregivers completed The MacArthur Communicative Development Inventories (MCDI): Words and Gestures or Words and Sentences.14 This is a measure of a person’s ability to comprehend language and to communicate verbally and with gestures. The inventory provides age equivalents for Language Comprehension, Language Production, and Gestures. Three “versions” of the Non-Communicating Children’s Pain Checklist12 were also completed by the caregivers of Sample 1. The first asked them to indicate whether the items, phrased in the past tense, had occurred in the past during pain. The remaining two were identical except for instructions indicating whether they were to be completed during “pain” or “distress,” and the word “pain” in some items of the version to be completed during pain was also replaced with the word “distress” in the version used during distress. These were completed immediately after actual incidents. One was an incident in which it was clear the participant had pain (Pain Incident). The other was an incident during which the participant was distressed but clearly had experienced no pain (Distress Incident). Caregivers also completed a visual analog scale of pain or distress intensity for the incidents and provided written descriptions of each incident.
Procedure This research was approved by the Research Ethics Boards of the IWK Grace Health Centre and of Dalhousie University. Informed consent was obtained from all caregivers. SAMPLE 1. Each caregiver was mailed the measures described, and they were telephoned after receiving them to en-
BREAU ET AL
THE JOURNAL OF PEDIATRICS VOLUME 138, NUMBER 5 sure they understood each questionnaire. All materials were returned by mail. Caregivers completed the retrospective checklist when they received the package. They then waited for a painful and a distressful incident to occur to complete the two additional checklists. SAMPLE 2. The second group of caregivers completed a 1-week diary that included the Non-Communicating Children’s Pain Checklist as part of their participation in a larger 2-year study. The information collected from day 1 of the diary was used in these analyses.
Statistical Analyses All data were analyzed by using SPSS 8.0 software.15 Descriptive statistics were generated for the characteristics of the participants (means, SDs, percentages). Relations between pain intensity and demographic characteristics of Sample 1 were made with Spearman correlations for ranked data, whereas correlations with categorical characteristics were conducted by using Eta coefficients. Comparisons of children with or without pain in Sample 2, based on noncategorical demographic characteristics, were conducted by using Mann-Whitney U tests for nonparametric data. To examine the predictive utility of each item based on caregiver retrospective report, odds ratios were computed, and comparisons of the occurrence of items were conducted by using matched-sample t tests. Multiple linear regression and logistic regression were used to examine the sensitivity and specificity of items with significant odds ratios. Significance was set at α = .05 for all tests, and Bonferroni corrections were used when multiple tests were conducted.
RESULTS Demographic Information SAMPLE 1. The demographic characteristics of the participants in Sample 1 are shown in Table I. All had severe cogni-
Table II. Pain experienced by sample 2 (n = 63) as a function of level of limb function
Level of limb function Upper limbs Full use Limitations No use Lower limbs Full use Limitations No use
tive delays and a wide range of physical impairments. VAS pain ratings for the episode observed were not related to the children’s age, Language Comprehension, Language Production, gestural ability, or physical impairments. Pain intensity was also not related to sex or pain type (acute, long-term). SAMPLE 2. The demographic characteristics of Sample 2 are also shown in Table I. All had severe cognitive impairments. According to their caregivers, 40% of the children had full upper limb function and 25% had full lower limb function, whereas 14% had no upper limb use and 41% had no lower limb use. In addition, 63% had some visual impairment and 17% had some hearing impairment. Children with limited or no use of their upper limbs (Mann-Whitney U = 226, P < .01) or lower limbs (Mann-Whitney U = 211, P < .01) were more likely to experience pain on the day the diary was completed than children with full upper or lower limb function (Table II).
Descriptions of the Painful and Distressful Incidents Caregivers’ descriptions of the painful and distressful incidents were reviewed to verify that evidence of pain or distress was present for the time they completed the checklists. Descriptions of pain included a bee sting, burns, respiratory tract infections, post-surgical and procedural
No pain on day 1 of diary (n = 45)
Pain on day 1 of diary (n = 18)
51% 40% 9%
11% 61% 28%
36% 32% 31%
0% 33% 67%
pain, and injuries. Descriptions of distress included the child being denied a pleasurable activity and events the child considered aversive (eg, bath, hair brushed/washed, fearful reactions to dogs/loud noises). VAS pain ratings for the pain episode ranged from 0 to 10 (mean = 6.6, SD = 2.4). VAS ratings of distress ranged from 3 to 10 (mean = 6.6, SD = 2.0). Eight of the 30 items of the Noncommunicating Pain Checklist were omitted from these analyses for the following reasons: (1) a response rate of <30% for the pain incident (Increase in sleep, Floppy, Shivering, Breath holding), (2) report during long-term pain only (Eat less, Decrease in sleep), (3) unsuitability for quantitative analyses (Specific sound for pain/distress, Specific movement for pain/distress).
Odds Ratios for the Checklist Items To determine whether caregivers’ retrospective reports of the items from the checklist could predict their reoccurrence during the subsequent pain incident, odds ratios were computed for each item. Odds ratios for which the CI does not include 1.0 are considered significant.16 The greatest odds of occurring were for Seeking Comfort, Gesturing, and Tears (Table III). In all, caregivers’ retrospective reports were a good indication that 7 items would re-occur, although the large CIs suggest there was large variability 723
BREAU ET AL
THE JOURNAL OF PEDIATRICS MAY 2001
Table III. Odds ratios for items of the Non-Communicating Children’s Pain Checklist occurring during a pain incident based on caregivers’ retrospective report (n = 33).
Pain incident Checklist items
Odds ratio
Vocal behavior Moaning, whining, whimpering (fairly soft) Crying (moderately loud) Screaming/yelling (very loud) Social Not co-operating, cranky, irritable, unhappy* Less interaction with others; withdrawn Seeks comfort or physical closeness* Difficult to distract, not able to satisfy or pacify Facial expression Furrowed brow Change in eyes* including squinching of eyes, eyes open wide, eyes frown Turn down of mouth, not smiling Lips pucker up, tight, pout or quiver Clenches or grinds teeth, chews; thrusts tongue out Activity Not moving, less active, quiet* Jumping around, agitated, fidgety Body and limbs Stiff, spastic, tense rigid Gestures to or touches part of body that hurts* Protects, favors, or guards part of body that hurts Flinches or moves body part away; sensitive to touch Physical signs Change in color; pallor Sweating; perspiring Tears* Sharp intake of breath; gasping*
95% CI
2.6 7.0 3.3
(.42–15.92) (.74–65.94) (.63–16.79)
12.0 2.8 42.0 1.4
(1.23–117.41) (.58–13.31) (4.12–428.66) (.33–6.17)
0.5
(.09–2.52)
8.9 3.8 2.0 5.3
(1.34–58.80) (.64–22.04) (.41–9.71) (.99–27.90)
9.2 2.8
(1.49–56.30) (.58–13.31)
1.8 38.0 3.7 3.7
(.35–8.80) (3.31–436.94) (.70–19.59) (.77–17.43)
2.3 6.0 23.8 6.5
(.23–23.91) (.54–67.28) (2.34–242.91) (1.20–35.57)
*Items with significant odds ratios.
among the children. This indicates that the 7 items are more consistently shown across episodes of pain than their non-significant counterparts.
Incidence of Checklist Items During Pain The previous analyses indicated that the 7-item subset was displayed consistently over time but did not examine whether these 7 items were displayed by many of the children, making them suitable candidates for a standardized instrument, or whether some items did not achieve significance because they 724
were only displayed during pain by a small number of the children. To investigate this, the number of items with significant odds ratios (Odds-High) and the number with non-significant odds ratios (Odds-Low) displayed by the children during pain were compared (Table IV). The results (t(32) = –.09, P = .93) indicated that there were equal numbers of Odds-High behaviors (mean = .44, SD = .30) and OddsLow behaviors shown (mean = .44, SD = .19). This indicates that low predictive ability was not due to low incidence of Odds-Low items and confirms
that high predictive ability was due to the consistency of the Odds-High items (as measured by the predictive utility of caregiver retrospective report), not their greater incidence across children during the pain episode.
Sensitivity and Specificity to Pain To confirm the sensitivity of the set of 7 Odds-High items to pain, a multiple regression was conducted. Because it was not expected that any one item would predict pain ratings independently, the set of 7 items was entered in one step, with caregivers’ VAS pain ratings as the dependent variable. This produced an R value of .70, which accounted for a significant amount of the variance in caregivers’ VAS ratings (F (7,22) = 3.1, P < .05). Five items were positively related to pain ratings: Cranky, Seeking Comfort, Change in Eyes, Less Active, and Tears. Two items, Gestures and Gasp, were negatively related to pain intensity. This provides further evidence that the 7item set is sensitive to pain. To test the discriminative ability of these items, a second multiple regression was performed with caregivers’ ratings of distress intensity during the distressful incident as the dependent variable. Again, all 7 items were entered in one step. This resulted in multiple R value of .31, which was non-significant (F(7,23) = .34, P > .05). Thus occurrence of the set of 7 items could not predict the intensity of distress, indicating that the 7-item set is specific to pain.
Replication As a final test of the ability of this subset to detect pain, analyses were conducted to determine whether these 7 items could classify a new group of 63 children with cognitive impairments as having pain or not having pain on a given day. Eighteen of the participants had pain on the day examined, whereas 45 had no pain that day. The mean pain rating for those who had pain, on a scale of 0 to 100, was 65 (SD = 37).
BREAU ET AL
THE JOURNAL OF PEDIATRICS VOLUME 138, NUMBER 5
Table IV. Occurrence of items from the Non-Communicating Children’s Pain Checklist during an observed pain incident as a function of retrospective report (n = 33)
Percentage of caregivers who observed item during pain incident Checklist items Vocal behavior Moaning, whining, whimpering (fairly soft) Crying (moderately loud) Screaming/yelling (very loud) Social Not cooperating, cranky, irritable, unhappy Less interaction with others; withdrawn Seeks comfort or physical closeness Difficult to distract, not able to satisfy or pacify Facial expression Furrowed brow Change in eyes, including squinching of eyes, eyes open wide, eyes frown Turn down of mouth, not smiling Lips pucker up, tight, pout or quiver Clenches or grinds teeth, chews; thrusts tongue out Activity Not moving, less active, quiet Jumping around, agitated, fidgety Body and limbs Stiff, spastic, tense rigid Gestures to or touches part of body that hurts Protects, favors, or guards part of body that hurts Flinches or moves body part away; sensitive to touch Physical signs Change in color; pallor Sweating; perspiring Tears Sharp intake of breath; gasping
Four children were missing scores for at least one item. Logistic regressions were conducted both with and without replacing those data, with no difference in significance. Thus the results obtained without replacing the missing values are presented below (16 children with pain and 43 without pain). All 7 items were entered in one step. The χ2 test indicated that 53% of variation in pain/no pain scores was accounted for (χ2 = 25.3, P < .001) and that the 7-item subset could correctly classify 11 (69%) of the 16 children who had pain
Item reported retrospectively
Item not reported retrospectively
Of total sample
86% 92% 67%
14% 8% 33%
68% 41% 30%
94% 60% 92% 62%
6% 40% 8% 38%
55% 30% 42% 43%
53% 84%
47% 16%
63% 70%
79% 50% 64%
21% 50% 36%
70% 36% 41%
85% 60%
15% 40%
33% 33%
64% 67% 67% 73%
36% 33% 33% 27%
44% 31% 31% 52%
83% 75% 94% 70%
17% 25% 6% 30%
21% 14% 60% 34%
and 41 (95%) of the 43 who did not have pain. Two children were predicted to have pain and did not. Five children were predicted not to have pain and did. One method of expressing the relation between sensitivity and specificity of a diagnostic test is to calculate the “predictive utility” of a positive or negative result.17 The current results suggest that 85% of those predicted to have pain did, in fact, have pain, indicating high positive predictive value or sensitivity, and 89% of those predicted not to have pain did not have pain, indicating high nega-
tive predictive value or specificity. The percentage of children displaying each of the 7 behaviors is shown in Table V. These results suggest that the 7-item subset derived with one sample could detect pain in a separate cohort of children with similar impairments and provide evidence that the items are generalizable.
DISCUSSION These findings highlight the fact that although many behaviors and signs 725
BREAU ET AL
THE JOURNAL OF PEDIATRICS MAY 2001
Table V. Percentage of participants displaying checklist items in sample 2 (n = 63)
Percentage displaying each item on day 1 of diary Checklist items Not cooperating, cranky, irritable, unhappy Seeks comfort or physical closeness Change in eyes, including: squinching of eyes, eyes open wide, eyes frown Not moving, less active, quiet Gestures to or touches part of body that hurts Tears Sharp intake of breath; gasping
may occur during any given episode of pain, only some are sufficiently reliable over time to predict pain and to be used for its identification. It is also significant that the 7 items, derived from one sample, could distinguish between the presence and absence of pain in a larger cohort. Replication with a new sample is essential to showing that a solution can be generalized to others in a similar population. One unique aspect of this study was that the items found to predict pain were based on caregivers’ retrospective reports, rather than on observations of previous discrete episodes. This means that their reports could be used as a guide by other observers. Although not directly tested, it is possible that use of the checklist assisted caregivers in providing more accurate information regarding their children’s previous pain behavior. Another important finding was that the subset predicted pain intensity ratings, but not distress intensity ratings, confirming their specificity to pain. Most intriguing, however, was the fact that an increase in 5 items (Cranky, Seeking Comfort, Change in Eyes, Less Active, and Tears) but a decrease in 2 (Gestures and Gasping) predicted greater pain intensity. One explanation for the decrease in Gestures and Gasping with more intense pain is that they both necessitate moving, which may 726
With pain (n = 18)
Without pain (n = 45)
65% 59% 65%
38% 37% 27%
35% 18% 62% 56%
20% 0% 18% 7%
exacerbate pain. The pain experienced by these children was particularly intense, with an average pain rating of 6.6 (± 2) of 10, a level usually considered “moderate.”18 Therefore it is possible that the difference in Gestures and Gasping found here was due to the high intensity of pain felt by this group. Another possible explanation for the decrease in Gestures lies in the complexity of executing that task. The children in this study had the gestural ability of infants 8 to 16 months of age and cognitive impairments that could make organizing and orchestrating such a behavior difficult. Many also had physical limitations that would add to the difficulty. Intense pain may have compounded this difficulty, because stressful situations can interfere with cognitive functioning.19 The reduction in Gasping is more difficult to explain, and contradictory findings have been reported. Several studies have reported that breathing pattern20 and rate21 change with pain in infants. However, there is currently no consensus on which changes are most likely to signify pain. Craig et al22 reported a reduction in breathing rate during pain and raised the possibility of Breath holding. Breath holding is one of the 30 items in the original NonCommunicating Children’s Pain Checklist. A review of the observations for this study indicated, however, that
only 2 children were reported to have held their breath during pain. It could be that caregivers were unable to recognize this behavior and that a reduction in short, sharp breaths and gasping actually reflects the fact that the participants were holding their breath. One limitation of the studies was the heterogeneity of the groups in terms of diagnosis. Another limitation of this study was that the observations on which odds ratios of an item occurring were computed were made by the same caregivers who completed the retrospective report. However, the fact that the subset did so well in classifying the second group of children experiencing pain provides evidence that the items can be generalized to other children and their caregivers. In addition, the CIs of the odds ratios generated for individual items were large, suggesting that the predictive ability of single items may vary somewhat from individual to individual. It is notable that the 7 items were from a variety of subscales including the Social, Facial, Activity, Body and Limb, and Physical Signs Subscales. This suggests that the usefulness of caregivers’ retrospective reports is not limited to one class or type of behavior. It is also interesting that no items from the Vocal Behavior Subscale were predictive of pain. This could reflect the fact that these children do not display much vocal behavior during pain, that their vocal behavior is less consistent than other pain behaviors, or that caregivers are less observant of or less reliant on their vocal behavior during pain. Regardless, the fact that no vocal behaviors were significant supports the premise that pain tools designed for individuals without cognitive impairments, which generally weight verbal behavior strongly,23 are not appropriate for this special group and underscores the need for development of pain assessment tools specifically designed for these children. These findings bode well for the development of pain assessment tools for
BREAU ET AL
THE JOURNAL OF PEDIATRICS VOLUME 138, NUMBER 5 children who cannot communicate verbally because of cognitive impairments. They also negate the common belief that the pain behavior of those with cognitive impairments may be too idiosyncratic to be measured by standardized instruments. As with development of any diagnostic instrument, these results should be regarded as only partial evidence of validity. Further research with different observers and in variety of settings is required before formal use of this tool can be recommended. The authors would like to recognize the contribution of the parents and caregivers who participated. Without their assistance, this research would not have been possible.
6.
7.
8.
9.
10.
REFERENCES 1. Anand KJS, Stevens B, McGrath PJ, editors. Pain in neonates. 2nd revised and enlarged ed. Amsterdam: Elsevier; 2000. 2. Biersdorff KK. Pain insensitivity and indifference: alternative explanations for some medical catastrophes. Ment Retard 1991;29:359-62. 3. Biersdorff KK. Incidence of significantly altered pain experience among individuals with developmental disabilities. Am J Ment Retard 1994;98: 619-31. 4. Oberlander TF, O’Donnell ME, Montgomery CJ. Pain in children with significant neurological impairment. Dev Behav Pediatr 1999;20:235-43. 5. Roy A, Simon GB. Intestinal obstruction as a cause of death in the mentally
11.
12.
13.
14.
handicapped. J Ment Defic Res 1987; 31:193-7. Jancar J, Speller CJ. Fatal intestinal obstruction in the mentally handicapped. J Intellect Disabil Res 1994;34:413-22. Parmelee PA, Smith B, Katz IR. Pain complaints and cognitive status among elderly institution residents. J Am Geriatr Soc 1993;41:517-22. Porter FL, Malhotra KM, Wolf CM, Morris JC, Miller JP, Smith MC. Dementia and response to pain in the elderly. Pain 1996;68:413-21. Oberlander TF, Gilbert CA, Chambers CT, O’Donnell ME, Craig KD. Biobehavioral responses to acute pain in adolescents with a significant neurologic impairment. Clin J Pain 1999;15:201-9. Koh JL, Fanurik D. Postoperative pain assessment in children with cognitive impairment. Presented at the 16th Annual Meeting of the American Pain Society; 1997 Oct 23-27; New Orleans, La. Giusiano B, Jimeno MT, Collignon P, Chau Y. Utilization of a neural network in the elaboration of an evaluation scale for pain in cerebral palsy. Methods Inf Sci 1995;34:498-502. McGrath PJ, Rosmus C, Campbell MA, Camfield C, Hennigar A. Behaviors caregivers use to determine pain in nonverbal, cognitively impaired children. Dev Med Child Neurol 1998;40:340-3. Breau LM, McGrath PJ, Camfield C, Rosmus C, Finley GA. Preliminary validation of an observational pain checklist for cognitively impaired, nonverbal persons. Dev Med Child Neurol 2000;42:609-12. Fenson L, Dale PS, Reznick JS, Thal D, Bates E, Hartung JP, et al.
15. 16.
17.
18.
19.
20.
21.
22.
23.
MacArthur Communicative Development Inventories: user’s guide and technical manual. San Diego: Singular Publishing Group, Inc; 1993. SPSS version 8 [computer software]. Chicago: SPSS Inc; 1998. Howell DC. Statistical methods for psychology. 3rd ed. Belmont (CA): Duxbury Press; 1992. Browner WS, Newman TB, Cummings SR. Designing a new study: III. Diagnostic tests. In: Hulley SB, Cummings SR, editors. Designing clinical research: an epidemiological approach. Baltimore: Williams & Wilkins; 1988. p. 87-97. Collins SL, Moore RA, McQuay HJ. The visual analogue pain intensity scale: what is moderate pain in millimetres? Pain 1997;72:95-7. Champion GD, Goodenough B, von Baeyer CL, Thomas W. Measurement of pain by self-report. In: Finley GA, McGrath PJ, editors. Measurement of pain in infants and children. Seattle: IASP Press; 1998. p. 123-60. McIntosh N, van Veen L, Brameyer H. The pain of heel prick and its measurement in preterm infants. Pain 1993;52:71-4. Rawlings DJ, Miller PA, Engel RR. The effect of circumcision on transcutaneous PO2 in term infants. Am J Dis Child 1980;134:676-8. Craig KD, Whitfield MF, Grunau RVE, Linton J, Hadjistavropoulos HD. Pain in the preterm neonate: behavioral and physiological indices. Pain 1993;52:287-99. McGrath PJ. Behavioral measures of pain. In: Finley GA, McGrath PJ, editors. Measurement of pain in infants and children. Seattle: IASP Press; 1998. p. 83-102.
727