Objective. To determine the onset of low back pain (LBP) in schoolchildren and to investigate the role of mechanical and psychosocial factors as risk factors for its onset.
Methods. A prospective population-based cohort study was conducted of 1046 schoolchildren, aged 11 to 14 years at baseline, identified as being free of LBP, from 39 secondary schools in Northwest England. New onset of LBP at 1-year follow-up was measured.
Results. Children who reported high levels of psychosocial difficulties were more likely to develop LBP than their peers (relative risk: 1.6; 95% confidence interval: 1.1–2.3). An excess risk was, in particular, associated with conduct problems (2.5; 1.7–3.7). Similarly, children who reported high numbers of somatic symptoms at baseline were at greater risk of developing LBP: abdominal pain (1.8; 1.1–3.0), headaches (1.6; 0.97–2.8), and sore throats (1.5; 0.8–2.6). In contrast, we have been unable to demonstrate a strong association between daily mechanical load (schoolbag weight) and the short-term risk of new-onset LBP (highest versus lowest quintile: 1.2; 0.7–2.1).
Conclusions. In children who were initially free of LBP, adverse psychosocial factors and the presence of other preexisting somatic pain symptoms were predictive of future LBP, reflecting findings in adults. In contrast, there was little evidence of an increase in short-term risk associated with mechanical load across the range of weights commonly carried by children to school.
Low back pain (LBP) is common in youths, although medical care is seldom sought.1 Recent studies have suggested a prevalence approaching what is reported in adults, and some authors have demonstrated a cumulative incidence of 74% by 20 years of age.2 Our previous work has revealed a 1-month period prevalence of 24% in 11- to 14-year-old schoolchildren.3
Some studies4,5 but not all6 have reported an association between LBP and height, or sitting height (height minus leg length). Similarly, a number of studies have demonstrated an association between LBP and body weight and/or body mass index (BMI)4,7,8 and also with aspects of lifestyle, eg, with both high and low levels of physical activity.8 Most epidemiological studies examining childhood LBP have focused on mechanical risk factors, and several cross-sectional studies have reported an association with heavy mechanical (schoolbag) load.7,9,10
A number of studies have suggested that there may be a cross-sectional association between LBP and psychosocial factors.11,12 However, it has not previously been examined, prospectively, whether psychological distress and adverse psychosocial factors—both of which have been demonstrated as important causative factors in adults13—lead to an increased risk of LBP in children.
The aim of the current study was to investigate the role of anthropometry, physical activity, mechanical load, and adverse psychosocial factors in the onset of LBP among schoolchildren. Furthermore, we wished to determine whether the report of LBP was related to other common somatic complaints, such as headaches and abdominal pain.
A prospective population-based study that consisted of 2 cross-sectional surveys 1 year apart was conducted.
Thirty-nine schools in the Northwest of England took part in the baseline survey, which has been reported previously.3 Schools, both state funded and independent (fee paying), were randomly selected from all secondary schools in the study area. From these schools children in years 7 to 9 (age: 11–14 years) were eligible to take part in the study. When they agreed to participate, each school was asked to select between 1 and 3 classes to take part in the study. Both parental and child consent was obtained before participation in the study, and permission to approach schools in the study area was obtained from the appropriate regional directors of education. Ethical approval for the study was granted by the University of Manchester Committee of the Ethics of Research on Human Beings and also by the appropriate local research ethics committees.
In the classroom, with the investigators present, children were asked to complete a health questionnaire on LBP and its potential risk factors. Subjects were defined as having LBP when they answered positively both of the following 2 questions:
Thinking back over the past month, have you had any low back pain that has lasted for 1 day or longer?
Thinking back over the past month, have you had any pain, which lasted for one day or longer, in the preshaded area below? (See Fig 1.)
The study investigators measured height and weight, and BMI was calculated using Quetelet’s Index: [weight (kg)/height (m)2]. With the use of the questionnaire, demographic information was gathered. Data were collected regarding the extent and type of physical education, weekly sports participation, and the mode of travel to and from school.
The amount of time spent watching TV and playing computer games was recorded to compute a composite index of sedentary activity. The children were asked whether they had a part-time job and, if so, what type of job, how many hours they worked per week, and whether the job required lifting or carrying heavy objects.
For assessing mechanical exposure, the children completed a schoolbag diary. Using a calibrated spring balance, the children measured the weight of their schoolbag(s) and recorded it over a 5-day period. From this, average daily mechanical load was calculated (hereafter referred to as “daily mechanical load”). Type of schoolbag and method of carriage were recorded as well as whether the child used a locker to store his or her bags while at school.
Psychosocial factors were recorded using the Strengths and Difficulties Questionnaire (SDQ), an instrument, validated in the appropriate age group, designed to assess prosocial behavior (strength) and hyperactivity, emotional problems, and behavioral dimensions such as peer problems and conduct problems (difficulties).14 For assessing the prevalence of other common somatic complaints—headache, abdominal pain, and sore throats—information was collected regarding how many days in the past month the child had had each of these symptoms. Their responses were categorized into “none,” “1 to 7 days,” and “>7 days.” Finally, the level of deprivation was assessed by the Townsend Index, using the postcode of residence of each subject.15 The study, which had a 97% participation rate, identified 1046 children (from a total of 1446) who were free of LBP.
The 1046 children who were initially free of LBP were revisited 1 year after baseline, and the 1-month period prevalence of LBP was determined by identical self-complete questions (and, by definition, all cases were new onset since baseline). As at baseline, the majority of data collection was performed in the classroom. Height and weight were measured again, and the gain over the baseline values was calculated. Also, because of a potential link between LBP and menstruation, information was gathered at follow-up on whether girls had started menstruating. Children who were absent on the day of data collection were left a questionnaire in school for completion later. When this was not convenient or when the children attended schools that declined to participate at follow-up, questionnaires were mailed (with reminders to nonresponders).
We evaluated the influence of the baseline risk factors in predicting future LBP. Our primary outcome was the new onset of LBP at follow-up. The association between exposure and outcome was calculated using a Poisson regression model, and the results are expressed as relative risks (RRs) adjusted for age and gender. Statistical analysis was conducted using Stata (version 7; Stata Corp, College Station, TX).
In total, 933 (89%) children participated at follow-up, with a median follow-up time of 12.4 months (interquartile range [IQR]: 11.9–13.6 months). A total of 777 (82.5%) children completed the questionnaire in the classroom with investigators present, 36 (3.9%) children completed the questionnaire at school at a later date, and 127 questionnaires (13.6%) were completed at home and returned by mail. A total of 168 (18.6%) children reported LBP that had lasted for 1 day or longer in the month before the follow-up survey, and the median duration of pain was 3 days (IQR: 2–7 days). The occurrence of new-onset LBP increased with age: 12.5% at 12 years versus 24.1% at 15 years (χ2trend: 10.9; P < .001; Fig 2), and the onset of pain was more frequent in girls (20.1%) than in boys (16.9%), although this was not statistically significant (difference: 3.1%; 95% confidence interval [CI]: −1.9%–8.2%). In girls, there was no association between LBP and menstruation (RR: 0.9; 95% CI: 0.5–1.6). Increasing deprivation was not associated with an increase in the likelihood of future LBP (χ2trend: 1.12; P = .29), and there was no statistically significant difference in risk between children who attended state schools or independent schools. The analysis of potential risk factors yielded similar results in boys and girls; therefore, all subsequent results are presented combined.
Baseline height, baseline weight, and change over the follow-up year did not influence the onset of LBP at follow-up (Table 1). Similarly, neither BMI nor its change over the follow-up year was associated with an increase in the risk of future LBP (data not shown).
Median daily mechanical load was 4.7 kg (IQR: 3.7–6.0 kg), equating to a relative daily mechanical load (bag weight*100/body weight) of 9.9% (IQR: 7.4–12.9%). Future LBP was not associated with daily mechanical load (see Table 2). Similarly, relative mechanical load, type of schoolbag, and the method of carrying the schoolbag were not associated with an increased risk of future pain (data not shown). It may be, however, that the measurement of schoolbag weight on only 1 occasion during the school day may be a poor proxy for daily mechanical load. For example, a student who walked to school and did not use a schoolbag locker would have a higher cumulative mechanical exposure than a student who carried the same schoolbag weight but traveled to school by car and did deposit the bag(s) in a locker. When this was examined, the risk of new-onset LBP associated with the carriage of heavy schoolbags was greater in children who walked to school than in those who traveled by other means and was also greater in those who did not use a locker. Although these results are not statistically significant, the trend in both is clear (Figs 3 and 4).
Children who, at baseline, participated in sports on a frequent basis were at an increased risk of developing LBP at follow-up (Table 2). However, neither the amount of physical education at school nor the extent of sedentary activities had any effect on the future onset of pain (Table 2). Having a part-time job at baseline significantly increased the risk of LBP at follow-up, after adjusting for age and gender (RR: 1.5; 95% CI: 1.1–2.1). However, no differences in risk were observed between job type, number of hours worked per week, or whether the job required the lifting of heavy items.
Calculated by summing the 4 “difficulties” dimensions of the SDQ, a high “total difficulties” score—indicating high levels of adverse psychosocial exposure—was associated with an increased risk of developing LBP at follow-up (RR: 1.6; 95% CI: 1.1–2.3; Table 3). Of the 4 individual psychosocial constructs that make up the total difficulties score, the presence of conduct problems (eg, anger, disobedience, violence) was a strong, significant predictor of future back pain (RR: 2.5; 95% CI: 1.7–3.7). Also, high levels of hyperactivity (restlessness, distraction, lack of concentration) were associated with an increased risk but to a lesser extent (RR: 1.4; 95% CI: 0.98–2.1). In contrast to the “difficulties” dimensions of the SDQ, there was no significant relationship between level of prosocial behavior (a strength) and future LBP.
Other Common Somatic Symptoms
A total of 7.5% of children (95% CI: 5.8–9.2) reported headaches on >7 days during the month before the baseline survey, 6.9% (5.2%–8.5%) reported sore throats, and 8.5% (6.7%–10.3%) reported abdominal pain. The presence of these other somatic symptoms at baseline was associated with an increased risk of LBP at follow-up (Table 4).
We report the results of the first prospective population-based study to examine the role of mechanical and psychosocial factors on the onset of LBP in youths. Adverse psychosocial factors, particularly the existence of conduct problems, are predictive of future pain. The reporting of back symptoms is also related to other childhood somatic symptoms. In contrast, we have been unable to demonstrate a strong association between daily mechanical load and the short-term risk of new-onset LBP.
A high follow-up response rate (89%) was achieved. The nature of the data collection process—in the classroom with the researchers present—minimized nonparticipation. Attempts were made to rerecruit children who were absent from class for any reason on the day of study, although the majority of nonresponders came from the few schools that were unable to participate at follow-up rather than from individual absenteeism. Rather than particular issues about the study, the reasons for schools’ not participating at follow-up were always related to a lack of time. Children who were absent on the day of follow-up but who completed a questionnaire at a later date did not differ from those present in terms of the rate of new-onset LBP or in terms of the prevalence of the major risk factors for its development (data not shown). It is unlikely, therefore, that bias was introduced in this manner.
A potential source of error concerned the accuracy with which the children recorded their schoolbag weight. The investigators weighed the schoolbags of a subset of individuals (approximately 70% of the sample) on the first day of the data collection week. In separate analysis using this data (not shown), a heavier bag weight on day 1 was not associated with the onset of LBP. From this, we conclude that any inaccuracies in bag weight measurement by the children did not contribute to our failing to find an association between daily mechanical load and LBP. Another possible weakness of the study is that baseline daily mechanical load may not have accurately reflected exposure during the course of the follow-up year. Mechanical exposure, however, was also recorded at follow-up, and the rank correlations of exposures at baseline and follow-up showed good agreement (Spearman’s ρ: 0.56; P < .001).
In a population of 12- to 15-year-old schoolchildren free of LBP 12 months previously, we demonstrated a 1-month period prevalence of new-onset LBP of 18.6%. These results are similar to levels of occurrence reported by other authors: Balagué et al16 reported a 1-week period prevalence of 16% in a 7- to 17-year-old Swiss population; Olsen et al,17 in the United States, reported a 1-year period prevalence of 22.0% in an 11- to 17-year-old population; and Nissinen et al5 demonstrated a 1-year incidence of 17.6% in Finnish children aged 12 to 13 years. As in other studies,10,16,18 we report an increase in LBP with age. With a higher prevalence in girls, particularly with the increase in occurrence with age, it has been proposed that pain might be related to menstruation. However, this seemed not to be the case: after age was adjusted for, girls who had started menstruating were at no greater risk of future LBP than their peers (RR: 0.9; 95% CI: 0.5–1.6).
Negrini et al19 reported that Italian schoolchildren are regularly exposed to, on average, a daily mechanical load equivalent to 22% of their body weight. Our results show that children in the United Kingdom carry much lighter loads. Cross-sectionally, some authors7,9,10 but not all20 have shown an association between heavy schoolbag loads and LBP. We have demonstrated prospectively that these factors alone are not contributory to LBP onset, although children with a high cumulative exposure may be at increased risk.
A number of studies have reported an increase in the likelihood of LBP associated with higher levels of sedentary activity.2,10 These studies were cross-sectional and therefore unable to address the temporal relationship between pain and inactivity. We demonstrate, prospectively, that previous sedentary activity cannot be considered a short-term risk factor for future LBP. In accordance with previous studies,8,16 we have shown an increased risk in those who undertake a high level of physical exercise, although we have been unable to establish a dose-risk relationship to support these findings.
Several authors have reported a cross-sectional association between psychosocial factors, or exposure to adverse psychosocial events, and the presence of LBP in children,11 and this has given rise to debate as to the temporal nature of this relationship. In adults who were initially free of LBP, it has been shown that work-related psychosocial factors and psychological distress both are associated with the development of LBP over a 12-month period.13 This is the first study to examine longitudinally the role of adverse psychosocial factors in the onset of pain symptoms in youths. We have demonstrated that high levels of the negative psychosocial dimensions, particularly conduct problems and hyperactivity, lead to an increased likelihood of new LBP 12 months subsequently.
Studies have suggested that LBP may be just 1 of a number of associated somatic symptoms, each with a similar cause.21 In children, Mikkelsson et al22 demonstrated that children with widespread pain syndrome not only had more emotional and behavioral problems than pain-free individuals but also experienced depressive symptoms and sleep problems to a greater degree than their peers. We have demonstrated that the reporting of other somatic symptoms in children increases the likelihood of future LBP.
Medical consultations for childhood LBP are relatively few. However, previous work has demonstrated that children who report LBP frequently report a degree of disability: 94% reported limitation in at least 1 aspect of the modified Hanover Low Back Pain Disability Questionnaire.3,23 This suggests that, although frequently not serious enough to warrant medical consultation, LBP in children ought not be considered a trivial condition.
Although it is commonly perceived that LBP occurs predominantly in adults, there is increasing evidence that it is also common in childhood.2,3,7 We propose that the origins of the adult back pain “career” may begin at least as early as adolescence. Some evidence exists for the long-term effects of psychosocial factors in relation to LBP: Power et al24 reported that poor emotional adjustment between the ages of 7 and 16 years was significantly associated with LBP at 33 years of age. Our study has gone further to demonstrate that these factors are also associated with an increased risk in the short term. In contrast, across the weight range carried by schoolchildren, this study has provided little evidence to suggest that daily mechanical load is a short-term risk factor for LBP, although it cannot be ruled out as a potential risk factor in the long term.
Funding was provided by the Colt Foundation; Dr Watson was sponsored by an MRC PhD studentship.
Drs Macfarlane and Silman designed the study, and Dr Macfarlane provided overall supervision. Dr Symmons contributed to monitoring study conduct. Dr Watson coordinated the baseline survey. Mr Jones coordinated the follow-up survey and was responsible for data processing and analysis, drafting the paper, and coordinating subsequent revisions, to which all authors contributed.
We thank the staff and pupils of the 39 schools that took part in the study and Stewart Taylor and Ann Papageorgiou for invaluable assistance with data collection.
- Received June 22, 2002.
- Accepted October 18, 2002.
- Reprint requests to (G.T.J.) ARC Epidemiology Unit, School of Epidemiology and Health Sciences, University of Manchester, Stopford Building, Manchester, UK, M13 9PT. E-mail:
- ↵McCormick A. Morbidity Statistics From General Practice: Fourth National Study 1991–92/a Study Carried Out by the Royal College of General Practitioners, the Office of Population Censuses and Surveys, and the Department of Health. London, United Kingdom: HMSO; 1995
- ↵Negrini S, Carabalona R, Pinochi G, Malengo R, Sibilla P. Backpack and back pain in schoolchildren: is there a direct relationship? J Bone Joint Surg Am.1998;80B :247
- ↵Townsend P, Phillimore P, Beattie A. Health and Deprivation: Inequalities and the North. London, United Kingdom: Croom Helm; 1988
- ↵Mikkelsson M, Sourander A, Piha J, Salminen JJ. Psychiatric symptoms in preadolescents with musculoskeletal pain and fibromyalgia. Pediatrics.1997;100 :220– 227
- Copyright © 2003 by the American Academy of Pediatrics