Objectives. To study the association of musculoskeletal pain with emotional and behavioral problems, especially depressive symptoms in Finnish preadolescents.
Study Design. A structured pain questionnaire was completed by 1756 third- and fifth-grade schoolchildren for identifying children with widespread pain (WSP), children with neck pain (NP), and pain-free controls for the comparative study. There were 124 children with WSP (mean age, 10.7 years), 108 children with NP (mean age, 11.1 years), and 131 controls (mean age, 10.7 years) who completed the Children's Depression Inventory (CDI) and a sleep questionnaire. A blinded clinical examination was done to detect fibromyalgia. For parental evaluation, the Child Behavior Checklist and a sociodemographic questionnaire were used. For teacher evaluation the Teacher Report Form was used.
Results. Children with WSP had significantly higher total emotional and behavioral scores than controls, according to child and parent evaluation. A significant difference in the mean total CDI scores was also found between the WSP and NP groups. Children with fibromyalgia had significantly higher CDI scores than the other children with WSP.
Conclusions. Musculoskeletal pain, especially fibromyalgia, and depressive symptoms had high comorbidity. Pain and depressive symptoms should be recognized to prevent a chronic pain problem.
Chronic widespread musculoskeletal pain and tenderness, together with several nonmusculoskeletal symptoms, are included in the criteria for fibromyalgia,1 a syndrome causing functional disability in adults as well as children and adolescents.2-4 Fibromyalgia with decreased pain threshold5,6 may represent one extreme of the widespread musculoskeletal pain disorders, although symptoms previously described as specific to fibromyalgia have also been found in other patients with widespread pain (WSP).2 Symptoms of depression and anxiety are often found in patients with fibromyalgia,7,8 with the estimated lifetime prevalence of depression ranging from 20% to 83% in clinical studies.9-12 However, all these studies had small sample sizes (from 7 to 35 patients) and the highest prevalence was found in the study with seven patients and using a nonstructured interview method.10 Subjective complaints of sleep disturbance are a prominent feature of many with fibromyalgia,13,14 and nonrapid-eye-movement sleep disturbance has been hypothesized to be related to musculoskeletal symptoms of fibromyalgia.15
The prevalence of depressive symptoms among children in the United States and in Finland has been estimated at 9.7% to 12.4%.16-18 Symptoms of depression and anxiety have been found to a greater extent in children with somatic complaints compared with healthy controls.19-24 Population studies25,26 or clinical population studies of children with WSP27 have not evaluated depression or other mood disturbances or sleep disturbances.
The aim of the study was to evaluate the association of musculoskeletal pain with emotional and behavioral problems, especially symptoms of depression. To address this question, we compared children with WSP with children with regional neck pain (NP) and with pain-free controls. Because it is known that cross-informant correlations are, in general, low to moderate,28,29 we collected data from children, parents, and teachers. Additionally, we were interested in whether children with fibromyalgia had more emotional and behavioral problems than other children with WSP.
MATERIALS AND METHODS
Pick-up Procedure of Children With Musculoskeletal Pain
The study took place in Lahti, a town in southern Finland with 94 827 inhabitants (1995). All 21 primary schools were asked to take part in the study, but 2 schools refused. The Steiner school, the hospital school, and the schools for the hearing disabled, physically disabled, and the mentally handicapped were excluded because the methods used in this study were not suitable. All pupils from the third and fifth grades completed a pain questionnaire, except those who were not at school on the day of the study.30
A structured pain questionnaire was developed and pretested for the study. The test-retest reliability of the questionnaire in detecting those who have pains at least once a week was good (κ, .9).30 The concurrent validity of the pain questionnaire was examined by comparing it with interviews of 31 third-grade and 25 fifth-grade children. The questions from the pain questionnaire were used as the basis of the interview. The children completed the pain questionnaire during a lesson and then were interviewed individually on the same day. The interviewer (M.M.) did not know the answers to the pain questionnaires. The observed agreement of pain questionnaire and interview technique was 86% [95% confidence interval (CI), 74% to 94%] and κ was .67.
During a lesson in the last week of March 1995, the pain questionnaire was completed by 1756 of the third- and fifth-grade schoolchildren, representing 82.9% of all schoolchildren of these grades in Lahti. The mean age of the third-grade children was 9.8 [standard deviation (SD), .34] years, and the mean age of the fifth-grade children 11.8 (SD, .37) years. There were questions about whether they had pain or aches during the previous 3 months (“since Christmas”) in the neck, lower back, lower extremities, or other regional areas (upper extremities, chest, upper back, or buttocks). They were also asked how often they have musculoskeletal pain and aches in these areas (almost every day, more than once a week, once a week, once a month, seldom, or never). Those who had pain at least once a week were classified into different pain groups on the basis of the painful area (neck, lower back, lower extremities, and other regional pain and WSP). WSP was determined according to the criteria for fibromyalgia1 (Table1). Those who seldom or never had pain formed the controls. If the children had pain attributable to an injury, they were asked to mark the area of the injury on the pain drawing with a different color. Pain attributable to injuries was not accepted for the pain classification.30 Of those who had pain at least once a week, the following two groups were selected for the comparative study: children with WSP and children with regional NP. The pain-free controls were randomly selected from the children who had pain seldom or never.
The comparative study was carried out in May 1995. There were 132 children (7.5%) who reported WSP, 114 children (6.5%) with NP, and 506 children (28.8%) having pain seldom or never. For the children with WSP, sex- and age-matched controls were sought from the 506 children having pain seldom or never and a sample of 131 sex- and age-matched controls was randomly selected. One child with WSP was excluded because a matched control could not be found for him. Because the number of children with NP was less than the number of children with WSP, we could not find sex- and age-matched pain controls for the WSP group from the NP group as we had planned.
The age and sex distributions of the children in the comparative study are presented in Table 2. There was no significant difference between the age and sex distributions among the three groups. Of the 13 who were excluded, 8 were not at school on the day of examination. One child was excluded attributable to incorrect preliminary classification, two boys refused to take part in the clinical examination (one had a psychiatric illness), and two fathers refused to allow their sons to take part in the study. Of the 124 children with WSP, 22 (17.7%) fulfilled the criteria for fibromyalgia.
The Figure shows the procedure of the study. A physiatrist (M.M.) and a nurse went to the schools during May 1995. For the comparative study, a blinded clinical examination with tender point palpation was done of the 363 children by the physiatrist to detect fibromyalgia.1 Children were asked to complete a Children's Depression Inventory (CDI)31 and a sleep questionnaire.32 The nurse supervised the children while they completed the questionnaire. They were also asked to take home to their parents the Child Behavior Checklist (CBCL)33,34 and a sociodemographic questionnaire with a prepaid envelope. Teachers were asked to complete the Teacher's Report Form (TRF)35,36 and mail it in a prepaid envelope to the research group.
Children's Depression Inventory.
Children were asked to complete the CDI, a well-known and validated instrument for detecting depression.31 The Finnish version includes 26 of the 27 items in the English version. For ethical reasons the question about suicide was excluded.17The results were processed on the basis of two cutoff points both previously used in epidemiological studies in Finland.17,18The cutoff point of ≥13, which has also been used in a Swedish epidemiological study,37 separated out 12.4% of the 8- to 9-year-old Finnish children having a possible depressive disorder.17 For the following studies, the board of the National Epidemiological Study chose a cutoff point of 17, which was estimated to distinguish approximately 10% of the children as depressed. This cutoff point resulted in 7.6% of the 1186 8- to 9-year-old children having a possible depressive disorder on the basis of the CDI. A subpopulation of this study with their parents were interviewed. When the results of the CDI and the interview were combined, 9.7% of the children were depressed.18 A cutoff point ≥13 has since been found to be the more appropriate than 17 for general screening purposes (Kresanov K, Tuominen J, Piha J, Almqvist F. Validation of child psychiatric screening methods. Unpublished data).
The children also completed a sleep questionnaire, which was a shorter, Finnish version of the questionnaire developed by Cook and Burd32 including six questions with five to six alternative answers per question. The questions concerned day tiredness, problems in falling asleep, naps, nightmares, and problems of waking up in the night.
Child Behavior Checklist.
The CBCL is a questionnaire for parents consisting of 118 behavior items, each scored from 0 to 2. The validity and reliability of the instrument has been well documented, eg, in the United States33,34 and in Finland.38 The instrument gives a total behavior problem score and two broad-band subscores, externalizing (E) and internalizing (I). The E-scale includes variables such as aggression, disorderly conduct, delinquent behavior, hyperactivity, and cruelty. The I-scale includes variables such as depression, anxiety, withdrawal, and somatizing. The age- and sex-specific cutoff points at the 83rd percentile were used for the limit of normal range functioning.33,34,38 Of the parents, 302 (83.2%) returned the questionnaire. The analysis of dropouts showed that children of parents who returned the questionnaire did not differ from all those included in the study according to the mean depression and sleep scores or mean total fibromyalgia tender point count. There were 24 (18.3%) from the WSP group, 19 (17.6%) from the NP group, and 18 (13.7%) from the control group whose parental evaluation was missing.
The sociodemographic questions dealt with the structure of the family, the number of children, and the education of parents. Parents were also asked about any diseases in the family and, if so, who has the disease. Demographic data gathered from the returned questionnaires are presented in Table 3.
Teacher Report Form.
The TRF is a questionnaire for teachers consisting of 118 behavior problem items, each scored from 0 to 2.35,36 The instrument gives total problem score and subscores like the CBCL. There are age- and sex-specific cutoff points for the limit of normal range functioning, and the cutoff point at the 83rd percentile was used. The validity and reliability of the instrument have been well documented.36 Teachers returned 276 questionnaires (76.0%). Of the teachers' evaluation, the data of 25 children (19.1%) of the WSP group, 30 children (27.8%) of the NP group, and 32 (24.4%) of the control group were missing. However, the mean depression and sleep score and the mean total fibromyalgia tender point count of those whose data were received from teachers did not differ significantly from the mean values of all children included in the study.
The descriptive values of variables were expressed as mean and standard deviations, median frequencies, or percentages. The most important descriptive values were expressed with confidence intervals. The differences between groups were evaluated by the χ2 test, Fisher's exact test, the nonparametric Mann-Whitney rank sum test, or the Kruskall-Wallis test. When the Kruskall-Wallis test revealed a significant difference (P < .05), the differences between the groups were localized by the Bonferroni multiple comparison method. The statistical computation was performed with the SPSS statistical program package.
The significance of differences in family structures between the groups in the comparative study dichotomously compared families with biological parents versus other family structure. No significant difference was found. The groups did not differ on the basis of parent's educational status. Although there was a minor difference in the percentages of sickness in the families of WSP children compared with the families of NP children and the control families, the difference did not reach statistical significance (Table 3).
Comparison of Total Scores Among Different Groups
Depressive Symptoms and Sleep Problems
Table 4 shows the mean and median scores of the CDI and the sleep questionnaire and the mean and median counts of tender points of the three groups. Both pain groups had higher scores on the CDI and the sleep questionnaire and had more tender points than the controls. The pain groups differed significantly from the controls in all these variables. There was also a significant difference in the mean total scores of the CDI between the two pain groups.
Of the children with WSP, 37 (29.8%) had depression scores of ≥13 on the CDI, as did 17 (15.7%) of the children with NP, and 3 (2.3%) of the controls.
On the basis of the higher cutoff point on the CDI (≥17), 22 (17.7%) of the children with WSP, 9 (8.3%) of the children with NP, and 2 (1.5%) of the control children had significant depressive symptoms. There was a statistically significant difference between the groups at both cutoff points (P < .001).
The mean depression and sleep scores of the 22 children with fibromyalgia, a subgroup of the children with WSP, were 14.7 and 7.4, respectively. The median of the CDI was 12.0 and of the sleep score 6. The mean count of tender points in children with fibromyalgia was 13.4. Children with fibromyalgia had higher scores on the CDI than the other children with WSP, and the difference between the groups reached statistical significance (P = .001). The difference between mean tender point counts of children with fibromyalgia and other children with WSP was also statistically significant (P < .001).
Of the children with fibromyalgia, 10 (45.5%) had a depressive symptom score on the CDI, whereas 27 (26.5%) of the other children with WSP had CDI scores ≥13. These 10 children with fibromyalgia represented 27.0% of the children with WSP who scored greater than the cutoff point of 13. Using the CDI ≥ 13 criterion for depressive symptoms, no significant difference between children with fibromyalgia and other children with WSP was identified (P = .12). However, when the cutoff point of 17 was used, 8 children (36.4%) with fibromyalgia met the criterion, whereas 14 (13.7%) of the other children with WSP met this criterion. At this cutoff point the difference between groups was significant (P= .026).
Table 5 shows the mean raw scores of the CBCL for the three groups and the statistical significance of the differences among the groups. Children with WSP and children with NP differed from the controls in CBCL total scores, E-scores and I-scores, and in subscores of withdrawn, somatic complaints, anxious/depressive, social, attention, and aggressive behavior problems. Differences between children with WSP and children with NP were not statistically significant.
We determined the effect of sex and school grade on the CBCL scores. Differences between the sexes on the third and fifth grades were not statistically significant (P = .44 andP = .96, respectively). Likewise differences between children in the third and fifth grades were not statistically significant (P = .35).
Sixteen (16%) of the children with WSP, 14 (15.7%) of the children with NP, and 10 (8.8%) of the controls achieved scores more than the cutoff point. The difference in abnormal functioning between the groups did not reach statistical significance. The effects of age and sex on the difference in abnormal functioning were not significant.
There were 15 children with fibromyalgia whose parents had completed the CBCL. On the basis of median CBCL scores, children with fibromyalgia did not differ from other children with WSP.
Of the children with fibromyalgia, 4 (26.7%) achieved scores more than the cutoff point compared with 12 (14.1%) of the other children with WSP. The difference was not statistically significant.
Table 6 shows the mean raw TRF scores of the three groups and the statistical significance of the differences among the groups. Children with WSP differed from controls in TRF total scores, E-scores, and subscores of somatic complaints, attention, and aggressive behavior problems.
Twenty (20.2%) of the children with WSP, 13 (16.7%) of the children with NP, and 8 (8.1%) of the controls achieved scores more than the cutoff point. With the χ2 test, the differences among the groups reached statistical significance (P = .049). Twenty-three (20.4%) of the third-grade children and 18 (10.8%) of the fifth-grade children scored more than the cutoff point. There was not a significant difference between the two groups (P = .04).
There were 12 children with fibromyalgia whose teachers returned the questionnaire. The median of the TRF total score was 16 in children with fibromyalgia and 12 in other children with WSP. However, the difference did not reach statistical significance (P = .63).
Of the children with fibromyalgia, 4 (33.3%) were rated more than the cutoff point by teachers, whereas 14 (16.1%) of the other children with WSP achieved scores more than the TRF cutoff point.
Our study shows that children with WSP had more emotional and behavioral problems than controls, according to the children themselves and parents. Furthermore, children with musculoskeletal pain had depressive symptoms and sleep problems to a greater extent than controls. In regard to depressive symptoms, children with WSP achieved higher mean scores than children with NP or controls on the CDI. There were more children from the WSP group who exceeded the cutoff level of depression compared with controls. Moreover, the role of depressive symptoms emerged in children with fibromyalgia. Although children with WSP and NP rated higher scores on the CDI, the mean scores still remained at normal level. However, children with fibromyalgia had a mean depression score more than the lower cutoff point and more than one-third of children with fibromyalgia reached the higher cutoff point. Thus, musculoskeletal pain, especially fibromyalgia, and symptoms of depression had high comorbidity.
Comparing adult fibromyalgia patients with other patients with chronic pain, there was not a significant difference in depression or anxiety symptoms on the SCL-90R scale. Patients with fibromyalgia had a greater extent of somatization.39 In that study the chronic pain group included regional chronic pain patients but the duration of pain was at least 6 months compared with 3 months in our study. In our study, not only children with fibromyalgia but all the children with WSP had significantly higher depressive symptom scores than children with NP, and the difference persisted when the questions about somatic symptoms were excluded.
According to parents and teachers, both pain groups had more aggressive behavior problems than controls. Depression in childhood typically includes irritability.16 One explanation is that aggressive behavior is a manifestation of depressive mood in children. Sleep disturbance, found to a greater extent in children with pain, may also be one of the depressive symptoms. It has been reported that a child with depressive symptoms is more aware of sleep disturbance than parents realize.16
As far as we know, this is the first population-based epidemiological study of the associations between widespread musculoskeletal pain and psychiatric symptoms in children. In a clinical study of psychosomatic musculoskeletal pain, clinical depression (CDI ≥ 17) was lower than in our study among children with WSP (11% vs 17.7%). If we compare the mean rate of depressive symptoms of both of our pain groups to that in the clinical study, the rate of depressive symptoms is similar (12% vs 11%).40 The results of the present study raise clinically important questions, eg, how to recognize depression behind multiple somatic complaints in children, to what extent do milder depressive symptoms become more serious, and what is the effect of age on the process? This cross-sectional study is unable to give answers to these questions and, as some authors have pointed out, longitudinal studies of somatization and pain problems in children are needed.27,41 Importantly, the present study design does not imply causalities between musculoskeletal pain and depressive symptoms.
One problem in previous epidemiological studies of pediatric pain has been that classification of simple occurrence or nonoccurrence of pain has been used.42 In our study we classified pain according to extent and frequency. The shortcoming of our study was that severity of pain was not determined in the pickup procedure. Another shortcoming was that information about behavioral symptoms was based on questionnaires, a method which does not give a clinical diagnosis. However, it was not possible to carry out diagnostic child interviews in this population-based study. The assessment of the validity of the pain questionnaire was limited because we did not test the three-month recall of pain among children with pain diaries or collect reports from parents. The fairly high prevalence rates of musculoskeletal pain in our study may raise a question whether Finnish children are prone to pain. However, Finnish children have reported local musculoskeletal pain symptoms to the same amount or less than children in other European countries.43-46 Probably, children are reporting fairly low intensities of pain symptoms.
Both fibromyalgia and depression may be pathophysiological disturbances related to serotonin metabolism.47,48 However, only some adult fibromyalgia patients have reacted as depression patients in the dexamethasone suppression test.49,50 The methods of this study are not accurate enough to make a diagnosis of depression, but the results indicate that depression should be kept in mind when children complain of musculoskeletal symptoms. The methods of this study do not exclude somatic diseases, and follow-up of the children will show any need for more accurate examinations.
To prevent a chronic pain state from developing, both depression16 and musculoskeletal pain should be recognized and treated as early as possible. To prevent the chronic pain syndrome in adulthood, it is important to know whether chronic musculoskeletal pain symptoms are already developed in childhood. In a 30-month, prospective follow-up study of fibromyalgia in children, fibromyalgia symptoms had a tendency to disappear spontaneously.26 This is contrary to the results of longitudinal studies in adults,51,52 which show that only 9% to 25% of patients improve during follow-up, whereas 75% of patients report an unchanged or worsened situation.
The results of our study provide evidence that children with WSP form a heterogeneous group, children with fibromyalgia representing those with more tenderness and overpresenting depressive symptoms. In addition, children with NP showed a tendency for an increased number of psychological symptoms, and, in the future, our follow-up will show whether children with NP develop multiple musculoskeletal and/or behavioral symptoms. As Wolfe14 has suggested, nociceptive input leading to sustained hyperalgesia might be causally related to fibromyalgia. High levels of psychological distress in adult fibromyalgia patients and their families have been found,14and high levels of life stress found in patients with fibromyalgia may affect psychological responses.53 The present study suggests that there is an association between musculoskeletal pain and depressive symptoms in children. However, what is most important is that both pain and depression in children are recognized and treated to prevent the limitations of a chronic pain problem.
This research was supported by the Signe and Ane Gyllenberg Foundation.
We are also grateful to Mr Hannu Kautiainen, BA, for statistical assistance.
- Received June 6, 1996.
- Accepted December 26, 1996.
Reprint requests to (M.M.) Rehabilitation Center, Rheumatism Foundation Hospital, 18120 Heinola, Finland.
- WSP =
- widespread pain •
- NP =
- neck pain •
- CI =
- confidence interval •
- SD =
- standard deviation •
- CDI =
- Children's Depression Inventory •
- CBCL =
- Child Behavior Checklist •
- TRF =
- Teacher Report Form •
- E =
- externalizing •
- I =
- Wolfe F,
- Ross K,
- Anderson J,
- Russel IJ,
- Hebert L
- Moldofsky H,
- Scarisbrick P,
- England R,
- Smythe H
- ↵Almqvist F, Tuompo-Johansson E, Panelius E, Aronen E, Kairemo A-C. Depressive symptoms according to the Children's Depression Inventory developed by Kovacs. Reports of Psychiatrica Fennica. Report No. 91. 1991:47–52
- ↵Puura K, Tamminen T, Bredenberg P, Escartin T, et al. Psychiatric disturbances in children within the catchment area of Tampere University Hospital. Foundation for Psychiatric Research. Publication Series 110 (in Finnish). Helsinki, Finland: Psychiatria Fennica Oy; 1995
- Walker LS,
- Greene JW
- Zuckerman B,
- Stevenson J,
- Bailey V
- ↵Sourander A. Child Psychiatric Short-Term Inpatient Treatment. A Prospective Study on Content and Outcome. Annales Universitatis Turkuensis, Ser. D, Medica-Odontologica; 1995:105–106; Thesis
- ↵Kovacs M. Rating scales to assess depression in school-aged children. Acta Paedopsychiatr. 1980/81;46:305–315
- ↵Achenbach TM, Edelbrock C. Manual for the Child Behavior Checklist and Revised Child Behavior Profile. Burlington, VT: University of Vermont, Department of Psychiatry; 1983
- ↵Achenbach TM. Manual for the Child Behavior Checklist and 1991 Profile. Burlington, VT: University of Vermont, Department of Psychiatry; 1991
- ↵Achenbach TM, Edelbrock CS. Manual for the Teacher's Report Form and Teacher Version of the Child Behavior Profile. Burlington, VT: University of Vermont, Department of Psychiatry; 1986
- ↵Achenbach TM. Manual for the Teacher's Report Form and 1991 Profile. Burlington, VT: University of Vermont, Department of Psychiatry; 1991
- Almqvist F,
- Bredenberg P,
- Suominen I,
- Leijala H
- Sherry DD,
- McGuire T,
- Mellins E,
- Salmonson K,
- Wallace CA,
- Nepom B
- ↵King A, Wold B, Tudor-Smith C, Harel Y. The Health of Youth—A Cross-National Survey. Canada: WHO Regional Publications; European Series No. 69. 1996:69
- Salminen JJ,
- Pentti J,
- Terho P
- ↵Meltzer HY, Lowy MT. The serotonin hypothesis of depression. In: Meltzer HY, ed. Psychopharmacology: The Third Generation of Progress. New York, NY: Raven Press; 1987:513–526
- Copyright © 1997 American Academy of Pediatrics