Abstract
Background Children report low back pain (LBP) as young as 8 years. Preventing LBP in children may prevent or delay adult incidence.
Objectives The purpose of this study was to determine whether education and daily exercise affect LBP episodes in children compared with education alone.
Design This was a prospective, multicenter cluster randomized controlled trial.
Setting The study was conducted at 7 New Zealand primary schools.
Participants Children (n=708), aged 8 to 11 years, from 7 schools stratified by sample size (36, 114, 151, 168, 113, 45, 83) were randomized and allocated to 2 masked groups: intervention (4 schools, n=469) or control (3 schools, n=239).
Interventions Participants in the intervention group were taught 4 spinal movements for daily practice. Both groups participated in education that emphasized “back awareness.”
Measurements Low back pain history at baseline was assessed. Children reported episodes of LBP during the previous week on trial days 7, 21, 49, 105, 161, and 270. Analysis was at the individual participant level, with adjustment for school clusters.
Results There were no significant differences between groups in the odds of reporting no LBP in the previous week during the study period (odds ratio [OR]=0.72; 95% confidence interval [95% CI]=0.46, 1.14; P=.16). The intervention group reported significantly fewer episodes of LBP (OR=0.54; 95% CI=0.39, 0.74; P<.001) and significantly fewer lifetime first episodes of LBP (n=86 [34%]) compared with the control group (n=58 [47%]) (OR=0.60; 95% CI=0.39, 0.91; P=.02). The odds of an episode of LBP were greater in participants with a history of LBP (OR=4.21; 95% CI=3.07, 5.78; P<.001). Low back pain episodes decreased across the trial period for both groups (OR=0.89; 95% CI=0.84, 0.95; P<.001). Adherence to exercise was poor.
Limitations Replication in other settings is needed.
Conclusions Regular exercise and education appear to reduce LBP episodes in children aged 8 to 11 years compared with education alone.
Primary prevention of LBP is sensible1,2 because of high prevalence rates, the association between childhood and adult LBP, and the risk for LBP is greater for people with a history of LBP than for those without.1,3–7 Lifetime prevalence of LBP increases from childhood to adolescence: 6% to 33% in children under 10 years of age,8 12% to 40% by 12 years, and 39% to 71% by age 15 years.1,8 If LBP in children can be prevented, entry into a cycle of recurring episodes may be delayed and adult LBP prevalence decreased. Despite considerable research, limited evidence exists for specific or modifiable risk factors for LBP in children.1,9 Hill and Keating9 identified 9 statistically significant predictors of future LBP in children, none which have been validated in independent reports. Little guides the design of an LBP prevention strategy for children.
Reviews of interventions have concluded with arguments for more research into strategies to reduce prevalence of LBP in children.2,10 Exercise may influence spinal health in young people. In a randomized controlled trial, Jones et al11 found a significant reduction in pain and increase in physical activity after an 8-week exercise program for adolescents compared with normal activity. No reduction in LBP episodes was observed. Fanucchi et al12 compared an 8-week exercise program and education with no exercise in 12- to 13-year-olds with a history of recent LBP. They reported an absolute risk reduction for LBP prevalence at 3 months (24%) and 6 months (40%) favoring the intervention group.12
Regular exercise is recommended for prevention of disease and health promotion, but a recognized barrier is adherence.13 The effects of exercise diminish when exercise participation ceases,1,14–16 and exercise may be therapeutic only during periods of adherence.17 We do not know if children can habitualize spinal exercise (in the way they have adopted dental hygiene). However, factors that contribute to success of adult exercise programs include establishing a routine, minimizing interruption to daily life, and family involvement18; interactive exercise demonstration, education, diagrams, and written instructions19; and exercise supervision and follow-up and behavioral techniques such as positive reinforcement and goal setting.14
The primary target of this research was to encourage attention to the spine, habitualize spine exercises, and provide participants with opportunities to gain knowledge of behaviors associated with spine health. A cluster randomization trial of schools was designed to limit contamination between control and intervention groups. Participants were in the age group of 8 to 11 years, in which LBP prevalence has been reported to be low but rapidly increasing.8 The a priori hypothesis was that education about healthy spine behaviors and a short daily exercise routine would reduce episodes of LBP compared with education alone. The primary outcome was an episode of LBP defined as self-reported LBP in the previous week. A secondary outcome was lifetime first episode of LBP defined as the first episode of LBP in the study period for participants with no previous history of LBP.
Method
Design Overview
The study was a prospective, multicenter cluster randomized controlled trial conducted in primary schools in New Zealand over 270 days. Schools were enrolled in the MySpine program and were allocated using randomization with concealment to an exercise and education group (intervention) or to education alone (control). The unit of randomization was school. Analysis of data at the participant level was adjusted for clustering by school. The method adhered to the guidelines in the Consolidated Standards of Reporting Trials (CONSORT)20 with consideration of the recommendations for conducting and reporting cluster randomized controlled trials.21
Setting and Participants
The study took place in schools in the North Shore City district of the Auckland North region of New Zealand during the 2011 primary school academic year. There were 131 primary schools in the Auckland North region, of which 47 were in North Shore City. Twenty North Shore City schools were deemed eligible, as they were either in the public or private sector and were no more than 30 minutes or 10 km from the primary researcher. This proximity was required to enable a single researcher (J.J.H.) to regularly visit all participating schools. Each school had a decile* rating that indicates the socioeconomic group of the school catchment area. All invited schools had a rating of 9 or 10. The primary researcher sent explanatory statements to school principals and met and gained consent in person from each interested school principal and school board.
Eligible participants were children, aged 8 to 11 years, in participating schools who appeared to be able to follow simple instructions and complete a child-friendly survey and whose parents approved their participation. Children who were not able to do the exercises due to spinal pathologies, neurological disorders, injuries, or physical disabilities for which movement was a contraindication or prevented standing on one leg safely and independently were excluded. The researcher sent an explanatory statement and consent form to the parents of every eligible child at each participating school via the principal. A follow-up consent form was sent 2 weeks later, which was signed by both parents and children.
Randomization and Interventions
Randomization was performed at the cluster level. The primary researcher enrolled participating schools into the study and created 2 groups, A and B, and designated one of them the intervention group. Schools (numbered 1–7) and the numbers of likely participants at each school were sent to the randomization officer (J.L.K.). The randomization officer, who was blind to all other information about the schools, ranked schools by student numbers and, using a random number sequence, randomly allocated sequential pairs of schools to condition A or B without knowledge of which was the intervention condition.
The primary researcher was an experienced physical therapist and visited participants in their classrooms (22–34 participants per class) 7 times during the 270 days of the trial, on days 1, 7, 21, 49, 105, 161, and 270. Figure 1 summarizes the intervention (MySpine program) and control conditions. All teachers were familiarized with the study and data collection procedures in a 15-minute information session prior to the day 1 school visit and were provided with an information and instruction sheet. Language and comprehension levels for the questions and instructions given to children were determined with input from teachers, parents, and children during the design phase of the study.
MySpine program. The MySpine Program may be used without written permission of the authors. Only use the MySpine name if the program is not modified, and reference should be made to the authors.
On day 1, children were given a personal identification number and completed a baseline survey in class (Fig. 2). Teachers were present during each education/exercise session and monitored all children to ensure that they completed a survey online,† or a paper version if requested, within a week of each visit by the researcher. Class teachers were provided with information about the link to the online survey. Children were shown how to log in and complete surveys in class and had a week after the visit to complete the surveys.
MySpine survey (baseline and follow-up). The MySpine Survey may be used without written permission of the authors. Only use the MySpine name if the program is not modified, and reference should be made to the authors.
At each session, the concept of back awareness and taking responsibility for “MySpine” was taught and reinforced for the intervention and control groups. Participants were taught habits thought to keep the spine healthy and principles underpinning recommended behaviors. In addition, the intervention group was taught 4 easy exercises (Fig. 1). The exercises were designed to encourage movement of the lumbar spine through flexion, extension, and lateral flexion. They could be completed quickly without supervision, were easy to remember, were enjoyable, and could be combined with existing daily routines to maximize the potential for adherence. Children were instructed to repeat each of the exercises 3 times in one session daily at a suitable time (eg, after brushing teeth, before bed, before school registration).
Specific adherence-enhancing strategies were incorporated into the school visits. These strategies included initial supervision of survey completion, exercise instruction, individual and group feedback as required at each session, regular follow-up visits to reinforce exercise behavior, education and behavioral techniques such as positive reinforcement and goal setting, and exercise reminders in the form of an A5 laminated card with diagrams depicting each exercise with a brief written instruction. Participants were advised to put the exercise cards in a place where they could see them every day.
Outcomes and Follow-up
Children were asked in the baseline (day 1) survey (Fig. 2) if they had ever experienced LBP in the area shown on the body chart. They also were asked if they had had “LBP today.” If they answered “yes” to either of these questions, they were classified as having a history of LBP at baseline.
The participants completed the follow-up survey (Fig. 2) within a week of the researcher visiting schools on study days 7, 21, 49, 105, 161, and 270. The survey asked: “Have you had back pain in the area shaded in the picture in the last week.” This was defined as pain in an area between T12 and S1 designated on a body chart provided with each survey. If a participant answered “yes,” this was recorded as an episode of LBP.
The primary outcome was an episode of LBP during the study period. Participants who reported no episodes of LBP in the previous week in any survey were classified as having no back pain during the study period. The secondary outcome was a lifetime first episode of LBP. Participants with no previous history of LBP prior to the study were monitored for a lifetime first episode of LBP and the follow-up assessment at which this episode was reported. Children were asked to report the duration of the LBP, choosing 1, 2, 3, 4, 5, 6, 7, or more than 7 days. Low back pain duration was later dichotomized into shorter episodes (1–2 days) and longer episodes (≥3 days). Participants also were asked if their LBP affected activity and required time off from school, ceasing participation in school sports, or visiting a health care professional.
At each assessment on days 7, 21, 49, 105, 161, and 270, participants were asked how many times in the previous week they had completed the exercises (Fig. 2). Participants were scored for adherence between 0 and 4 depending on how often they completed exercises: not at all (0), about once a week (1), about twice a week (2), every second day (3), or every day (4). We also calculated a total adherence score (0–24) for each participant across the 6 follow-up assessments. Additionally, participants were questioned about reasons for nonadherence, with the options: sickness/injury, forgot, cannot remember how to do the movements, find the movements uncomfortable, too busy, or other. Comprehensive analysis of adherence data will be reported elsewhere (unpublished data).
Sample Size Calculations
To estimate the sample size required, we combined data from an LBP prevalence review8 and recent exercise trials11,12 to estimate the number of LBP episodes we were likely to observe over the trial period. We used estimates of LBP incidence in children aged 8 to 11 years of 15% for the control group and 9% for the intervention group. We used Donner and colleagues' model22 to adjust estimates for effects due to cluster randomization. With a cluster size of 7 (3 in the control group and 4 in the intervention group), a conservative intracluster correlation coefficient (ICC) of .2, 80% power, a nondirectional test, and an alpha of .05, our estimated total sample size was 714.
Data Analysis
Analyses were performed at the individual participant level and adjusted for the effect of clusters. Intervention and control groups were compared at baseline for sex, age, and prior episodes of LBP before or on the day of the study using regression analysis adjusted for school clusters.
The primary outcome was episodes of LBP during the study period. The first analysis compared groups on reports of no LBP episodes during the trial using logistic regression.
All reports of LBP episodes over the study period were then compared for intervention and control groups using mixed-effects logistic modeling (clustered by school and student ID). The covariates “data collection point” (2–7) and “whether the participant had experienced a previous LBP” were included in the model.
The primary comparative analyses were conducted using an intention-to-treat analysis. Missing data were accommodated by logistic regression analysis.23
Groups were compared using ordered logistic regression on reports of lifetime first episode of LBP during the study period and on time to lifetime first episode of LBP. Participants with LBP at baseline were excluded from this analysis. Groups were compared for duration of LBP episodes using cluster adjusted logistic regression analysis. Low back pain severity (reports of missed sport, missed school, or visited health practitioner) was compared using mixed-effects logistic modeling (clustered by school and student ID).
Survey data regarding adherence were explored for association with LBP episode during the study period, lifetime first episode of LBP during the study period, and previous history of LBP using logistic regression with adjustment for school clusters. STATA version 12 (StataCorp LP, College Station, Texas) was used for statistical analysis.
Results
Recruitment and Randomization of Schools
Participants were recruited in April 2011. Seven schools consented to participate; 13 schools declined to participate. Participating schools were randomized to control (n=3) and intervention (n=4). There were 764 children in the 7 participating schools who were eligible to participate; 710 children consented to data being used in analysis (Fig. 3). Participant numbers at each school were: control group (n=45, 82, and 113) and intervention group (n=36, 114, 152, and 168). Two participants (1 control group, 1 intervention group) left their participating school and were lost to follow-up. No participants changed groups during the study.
Flow of participants and clusters (schools) throughout the study.
Baseline Data
Participant demographics and characteristics are presented in Table 1. Baseline characteristics were similar between groups. There were no significant differences between groups in age (P=.61), sex (P=.39), or prior episodes of LBP before or on the day of the study (P=.94).
Baseline Participant and School Characteristics: Baseline Characteristics at Individual and Cluster Levels
Response Rates
The response rate remained ≥70% (range=70%–92%) for both the control and intervention groups at all data collection points (data collection points 2–7, respectively: 83%, 76%, 81%, 88%, 74%, and 77%). There was no significant difference between groups in response attrition.
LBP Episodes
During the trial, 35% (n=164) of the intervention group and 28% (n=66) of the control group reported no LBP episodes. Logistic regression (clustered by school) indicated no significant effect for group (OR=0.72; 95% CI=0.46, 1.14; P=.16).
Comparison of Reports of LBP Episodes
Reports of episodes of LBP for all participants (regardless of previous history) at each of the data collection points were compared for intervention and control group participants (Tab. 2). Mixed-effects logistic modeling (clustered by school and student ID) was applied to predict a report of an episode of LBP when group, data collection points, and whether the participant had experienced previous LBP were included in the model. The odds of reporting an episode of LBP during the study were lower in the intervention group than in the control group (OR=0.54; 95% CI=0.39, 0.74; P<.001). The odds of reporting an episode of LBP (regardless of group) was greater in participants with a history of LBP at baseline than in participants with no history of LBP prior to participation in the study (OR=4.21; 95% CI=3.07, 5.78; P<.001). The covariate “data collection day” was significant (OR=0.89; 95% CI=0.84, 0.95; P<.001), with the number of episodes of reported LBP decreasing for both groups across time.
Low Back Pain (LBP) Episodes and Lifetime First Episodes of LBP Reported During the Studya
Lifetime First Episode of LBP
Participant reports of lifetime first episode of LBP (ie, participants with no previous episodes of LBP) at each time point are reported in Table 2. Logistic regression (clustered by school) indicated significantly fewer reports of lifetime first episode LBP for the intervention group over the study period (OR=0.60; 95% CI=0.39, 0.91; P=.02). The time to lifetime first episode LBP was significantly different between groups (OR=1.58; 95% CI=1.04, 2.39; P=.031).
Measures of Duration and Severity
Low back pain lasted 1 to 2 days in 67% of total reported episodes (72% of control group cases and 62% of intervention group cases). Remaining episodes lasted ≥3 days. There was no significant difference between groups for duration of reported LBP episodes (OR=1.27; 95% CI=0.88, 1.86; P=.20).
There were no differences between groups in the reports of missed sport (control group: 28 [9%], intervention group: 36 [9%]; OR=1.03; 95% CI=0.49, 2.18; P=.941), days off from school (control group: 15 [5%], intervention 21 [5%]; OR=0.87; 95% CI=0.40, 1.86; P=.710), or visits to a health care practitioner (control group: 22 [9%], intervention group: 33 [9%]; OR=0.80; 95% CI=0.36, 1.80; P=.593).
Adherence to Exercise
There was no significant effect of frequency of exercise (coded from 0 [no exercise] to 4 [exercises daily]) in the odds (adjusted for clusters) of an episode of LBP over the study period (OR=1.05; 95% CI=0.61, 1.81; P=.86) or lifetime first episode LBP (OR=0.61; 95% CI=0.24, 1.59; P=.31). No relationship was found at any data collection point between a preintervention history of LBP and adherence (OR=0.97; 95% CI=0.77, 1.23; P=.77).
Discussion
Participation in a simple exercise program had no effect on the proportion of children reporting no LBP over the study period. However, children in the exercise group did report fewer episodes of LBP and fewer lifetime first episodes of LBP than those in the control group. Low back pain lasted 1 to 2 days in 67% of reported episodes. Few participants experienced LBP that affected sport participation (9%) or school absence (5%) or resulted in a visit to a health care practitioner (9%). A previous history of LBP was associated with greater odds of reporting an episode of LBP.
At inception, 47% of the participants reported a previous episode of LBP. This figure appears higher than some of the reports in a recent systematic review of prevalence of LBP in children,8 with estimates of lifetime prevalence LBP or lifetime episode LBP of 12%8,24 for 10- to 15-year-olds and of 9%8,25 for 13- to 17-year-olds. However, prevalence rates similar to those of our study also were found in many studies: Balagué et al26 reported 51% lifetime prevalence in 12- to 17-year-olds,8 Newcomer et al27 reported 51% in 10- to 18-year olds, Sardelic and Ryan28 reported 53% in 12- to 17-year-olds, and McMeeken et al29 reported 50% in 9- to 18-year-olds. For these studies, we could find no relationship between method quality scores or LBP definition and the reported prevalence rates.
We observed a decrease in episodes of LBP across the study period (intervention group: from 23% at day 7 to 13% at day 270; control group: from 33% at day 7 to 24% at day 270). This decrease by both groups across the study also was reported by Fanucchi et al,12 who found an absolute reduction in risk of LBP prevalence at 3 months (24%; 95% CI=4%, 41%) and at 6 months (40%; 95% CI=18%, 57%) in the intervention group compared with the control group. In that study, both groups showed a reduced prevalence of LBP across time (intervention group: from 67% at 3 months to 42% at 6 months; control group: from 91% at 3 months to 81% at 6 months).
Participants in the intervention group were less likely to report a lifetime first episode of LBP than the controls. Differences between groups in reported episodes of LBP appeared at the first data collection period on day 7 of the study, and small differences between groups were evident at each of the subsequent data collection points in favor of the intervention (exercise) group. If the intervention was responsible for this difference between groups, the initial effect occurred very early (within 7 days).
The odds of reporting an episode of LBP (regardless of group) were much greater in a participant with a previous episode of LBP at baseline compared with those who had never had an LBP. This finding aligns with data for adults.1,3,5,6 Hestbaek et al30 monitored 10,000 twins prospectively for 8 years from adolescence to adulthood and concluded that LBP in adolescence predicts adult LBP and that a longer duration of LBP at baseline predicts a longer duration at follow up. The odds of adult LBP were greater for those adults who reported persistent LBP in adolescence compared with those who did not (OR=4.30; 95% CI=3.5, 5.3). This finding concurs closely with our findings that the odds of reporting an episode LBP was greater in participants with a history of LBP at baseline than in participants with no history of LBP prior to participation in the study (OR=4.21; 95% CI=3.07, 5.78; P<.001).
Few participants reported missing school or sports or visiting a health care provider for their LBP episode. This finding concurs with population surveys indicating that many people do not report LBP, take time off from work or school, or visit a physician.31 Reported LBP episodes were commonly less than 2 days' duration (total sample: 67%; control group: 72%; intervention group: 62%). The remaining cases were ≥3 days' duration. We do not know whether pain was intermittent or constant over the specified durations. Defining appropriate duration metrics is contentious and challenging32 and may explain variations in published reports.8
It has been suggested that it may be inappropriate to apply adult definitions of LBP when assessing children33 and that children may conceptualize pain in ways that differ from adults.34 We used a clear diagram of the target pain area and specific time frames for LBP duration to create an unambiguous and reproducible platform for validation studies. Regular monitoring of participants was designed to minimize recall errors, as LBP prevalence studies have shown less variation in short recall periods compared with longer recall periods.8,35 We found no guidelines for appropriate recall periods for children.
Lower lifetime first episode LBP, longer time to onset of first episode, and lower numbers of reports of episode LBP in the intervention group were surprising. Despite adherence to exercise declining across the study, a reduction in reported LBP episodes occurred in both groups as the study progressed. The benefits of the program cannot be confidently attributed to exercise participation. It is unlikely (although not impossible) that the 4 exercises in this study were sufficient to have a physiological effect. It is possible that monitoring participants and talking about the spine, drawing attention to the vulnerability of the low back and introducing the concept of back care, movement, and spinal awareness, confers some therapeutic effect or provides a strategy that children used to control the influence of factors that increase the risk of a LBP event. This approach has been observed in other fields of study.36 Vanwormer et al36 evaluated studies investigating the impact of regular self-weighing for both weight loss and weight maintenance. In the absence of any other strategies, frequent self-weighing was associated with greater weight loss or weight gain prevention. Perhaps vigilance creates opportunities for control. By participating in the MySpine program, children may have been empowered to identify and adjust behaviors to reduce the risk of LBP.
Limitations
A number of limitations were encountered, in part, due to the constraints of studies involving children in an educational environment involving many centers. The response rate was below 85%, which was fair but below ideal. The children were not blind to the fact that they were exercising, although they were unaware of the control conditions, and the researcher who analyzed the data was not blinded to groups.
We have little data on the validity of children as witnesses to pain history, but replication of baseline data by independent researchers would not be difficult. If children are not valid witnesses, an alternative explanation is needed for the observation that at baseline both groups reported similar proportions of previous LBP and similar proportions at each assessment time point. More research into the validity of children as witnesses to episodes of LBP is warranted.
We advised children to report any LBP they could recall. Children selected from specified duration options, the smallest of which was 1 day. In the future, we would consider providing response options that include smaller duration periods (eg, <1 hour). These response options would improve our understanding of the child's experience. If the survey methods used to collect data were not adequately reliable, this may account for occasions where no relationship or effect was observed. Although poor reliability might explain a failure to detect an effect, it would not explain systematic differences between groups. The method was at least reliable enough to detect those differences between groups that were observed.
We based our sample size calculation on an estimated correlation between clusters of .2. Subsequent calculations of the ICC directly from the data identified that it was effectively zero. However, the effect of the intervention was much smaller than anticipated, and our power (assessed post hoc) to observed differences between groups in episodes of LBP was 0.73.
Participants were from a narrow social status (decile 9/10), which affects the generalizabilty of the results. We do not know if groups were different at baseline with regard to history of LBP in the previous week. Only “LBP today” and “any previous history of LBP” were reported at baseline.
In conclusion, the MySpine program provides a program of spine exercises and education that may reduce the episodes of LBP in a group of children aged 8 to 11 years. It is unlikely that the beneficial effects experienced by the participants can be attributed to the specific exercises. Replication studies are needed to validate these findings. Regular exercise and education appear to reduce LBP events in children aged 8 to 11 years compared with education alone.
Footnotes
Both authors provided concept/idea/research design, writing, data analysis, and project management. Ms Hill provided data collection. The authors thank the teachers and their schools for participating in this study.
This study was approved by the Northern Y Ethics committee (Project Number NTY/10/11/093, February 16, 2011) and subsequently by the Monash University Human Research Ethics Committee (Project Number 2011000216, March 2, 2011).
Clinical Trial Registration Number: ACTRN12611000551998.
↵* Decile ratings are determined by the Ministry of Education. Schools are ranked by socioeconomic status, divided into 10 equal-sized units rated 1 through 10. A rating of 1 indicates a poor area; a rating of 10 indicates a high level of affluence.
↵† Data obtained using Qualtrics software (Qualtrics, Provo, Utah), http://www.qualtrics.com.
- Received June 19, 2014.
- Accepted November 25, 2014.
- © 2015 American Physical Therapy Association