|Home | About | Journals | Submit | Contact Us | Français|
The role of leisure-time physical activity in sciatica is uncertain. This study aimed to assess the association of leisure-time physical activity with lumbar radicular pain and sciatica.
Literature searches were conducted in PubMed, Embase, Web of Science, Scopus, Google Scholar and ResearchGate databases from 1964 through August 2015. A random-effects meta-analysis was performed, and heterogeneity and small-study bias were assessed.
Ten cohort (N=82,024 participants), 4 case-control (N=9350) and 4 cross-sectional (N=10,046)) studies qualified for meta-analysis. In comparison with no regular physical activity, high level of physical activity (≥4 times/week) was inversely associated with new onset of lumbar radicular pain or sciatica in a meta-analysis of prospective cohort studies (risk ratio (RR)=0.88, 95% CI 0.78-0.99, I2=0%, 7 studies, N=78,065). The association for moderate level of physical activity (1-3 times/week) was weaker (RR=0.93, CI 0.82-1.05, I2=0%, 6 studies, N=69,049), and there was no association with physical activity for at least once/week (RR=0.99, CI 0.86-1.13, 9 studies, N=73,008). On the contrary, a meta-analysis of cross-sectional studies showed a higher prevalence of lumbar radicular pain or sciatica in participants who exercised at least once/week (prevalence ratio (PR)=1.29, CI 1.09-1.53, I2=0%, 4 studies, N=10,046), or 1-3 times/week (PR=1.34, CI 1.02-1.77, I2=0%, N=7631) than among inactive participants. There was no evidence of small-study bias.
This meta-analysis suggests that moderate to high level of leisure physical activity may have a moderate protective effect against development of lumbar radicular pain. However, a large reduction in risk (>30%) seems unlikely.
Lumbar radicular pain is normally defined as pain which radiates from the low back or buttock to below the knee. It is a common symptom – reported prevalence rates ranging from 1% to 43% (Konstantinou and Dunn, 2008) – and may be accompanied by clinical findings suggestive of a herniated disc or nerve root irritation (Ropper and Zafonte, 2015) in “clinically verified sciatica”. In two Finnish cohort studies, the prevalence of lumbar radicular pain lasting longer than seven days during the past 12 months among participants who had previously been free from the symptom was 9% and 12% (Miranda et al., 2002; Shiri et al., 2010c), and that of lumbar radicular pain lasting longer than 30 days 3.0% and 3.4% (Miranda et al., 2002; Shiri et al., 2010c). The risk of lumbar radicular pain increased with age and did not differ between men and women (Miranda et al., 2002; Shiri et al., 2010c).
Risk factors for the development of lumbar radicular pain have been studied less than those for low back pain more generally. Among potentially relevant aspects of lifestyle risk, obesity (Shiri et al., 2014) and smoking (Shiri and Falah-Hassani, 2015) are associated with moderately increased risk, but the role of leisure-time physical activity is uncertain (Shiri et al., 2007). Some prospective cohort studies (Kaila-Kangas et al., 2003; Riihimäki et al., 1994) have found an increased risk in individuals who were moderately or highly active during their leisure-time, while others (Miranda et al., 2002; Shiri et al., 2013) have suggested that moderate or vigorous leisure-time physical activity protects against lumbar radicular pain, particularly in people who are overweight or obese (Shiri et al., 2013). Associations in either direction are plausible. Physical activity may impose mechanical stress on spinal structures leading to injury. On the other hand, there are indications that physical activity can be beneficial in patients who have already developed sciatica (Jensen et al., 2010; Kara et al., 2005).
Several systematic reviews have indicated an absence of association between leisure-time physical activity and low back pain overall (Bakker et al., 2009; Chen et al., 2009; Heneweer et al., 2011), but they did not explore associations more specifically with lumbar radicular pain or clinically verified sciatica. The aim of this systematic review and meta-analysis was to assess the associations of moderate and high levels of leisure-time physical activity with lumbar radicular pain and clinically verified sciatica.
We used the PRISMA statement (Moher et al., 2009) when developing our review protocol and meta-analysis. The first reviewer (RS) conducted comprehensive literature searches in PubMed, Embase, Scopus, Web of Science, Google Scholar and ResearchGate databases from 1964 through August 2015, using predefined combinations of MeSH terms (PubMed), Emtree terms (Embase) and text words (supplementary table S1). There was no restriction on language. He also hand-searched the reference lists of included articles for additional reports that might be relevant, and looked at the full text of studies on other risk factors for lumbar radicular pain or sciatica such as overweight/obesity and smoking (Shiri et al., 2007) in case they included data on leisure-time physical activity. In addition, he checked the full text of studies on associations of overweight/obesity, smoking and physical activity/inactivity with low back pain (Shiri et al., 2010a, b) for any findings that they might include on lumbar radicular pain or sciatica. Additional results were sought from the corresponding authors of two studies (Jin et al., 2015; Kääriä et al., 2011), but they were unable to provide any.
The first reviewer assessed the titles, abstracts and full texts of potentially relevant reports and determined whether they examined an association of leisure-time physical activity/inactivity with lumbar radicular pain or clinically verified sciatica. Cross-sectional studies, case-control studies (both population-based and hospital-based) and cohort studies were all potentially eligible for inclusion in the review. However, studies that were conducted among self-selected volunteers, that recruited patients with lumbar radicular pain or sciatica but lacked a control group, or that did not report quantitative data on associations with leisure-time physical activity were excluded.
The outcomes of interest were: 1) Lumbar radicular pain. Low back pain radiating down the lower limb; and 2) Clinically verified sciatica. Lumbar radicular pain that is accompanied by physical signs or radiographic evidence of herniated disc or nerve root impingement.
Following the approach of studies that were eligible for inclusion, being physically active was defined as participation in a sport or other physical activity during leisure-time, at least once a week. Moderate level of physical activity was defined as participation in such activity 1-3 times a week, or being in the middle third of the distribution of leisure-time physical activity in a study sample. High level of physical activity was defined as participation in leisure-time physical activity 4-7 times per week or being in the upper third of the distribution of such activity in a study sample.
The methodological quality of the included studies was assessed independently by two reviewers (RS and KFH) using criteria adapted from the Effective Public Health Practice Project tool (Armijo-Olivo et al., 2012). Five sources of bias: selection bias, performance bias, detection bias, confounding, and attrition bias (supplementary table S2) were assessed. Disagreements between raters were resolved through discussion.
The outcome in our main meta-analyses, was all lumbar radicular pain, whether or not it fulfilled the criteria for clinically verified sciatica. From each study, we abstracted maximally adjusted risk estimates for the exposures and outcomes of interest, together with their 95%confidence intervals. For one study (Videman et al., 1995) that did not report a confidence interval for an odds ratio (OR), the standard error (SE) was calculated using the following formula: SE = log (OR) / Z value (Higgins and Green, 2009).
A few studies (Karjalainen et al., 2013; Kääriä et al., 2011; Matsudaira et al., 2013) used high level of physical activity as a comparison group and explored the associations of inactivity vs. activity, low vs. high, and moderate vs. high level of leisure time physical activity with lumbar radicular pain or clinically verified sciatica. The sample sizes and numbers of events did not differ markedly between active and inactive subgroups. We calculated a proxy OR for high level of physical activity by inverting the OR for physical inactivity, and a proxy OR for moderate level of physical activity by dividing the OR for moderate activity by that for inactivity. We then used the SE of the estimate for physical inactivity to calculate a 95% CI for high level of activity. For moderate level of activity we used its own SE. We used natural logarithms of lower and upper limits of the confidence interval to calculate SEs and new 95% CIs.
Some studies (Mattila et al., 2008; Miranda et al., 2002; Schumann et al., 2010) reported estimates for the associations of two or more different types of sports activities with lumbar radicular pain or sciatica, in which case we combined results by a fixed-effect meta-analysis to give an overall pooled estimate for physical activity vs. physical inactivity for the study. Moreover, because some participants contributed data to more than one estimate, we corrected the variance of the pooled estimate by a method that has been suggested for combining multiple outcomes or multiple time-points within a study (Borenstein et al., 2009 ). The formula for calculating the variance (V) for combined estimate of two associations is: Vcombined =1/4 (Vestimate1 + Vestimate2 + 2r √ estimate1 √ estimate2) where r denotes correlation between two sports activities. In the absence of data in the included papers to estimate r for two sports activities, we used a value of 0.50. For these calculations we used the natural logarithms of ORs or RRs and their confidence intervals.
We converted ORs to prevalence ratios (PRs) for cross sectional studies and ORs to risk ratios (RRs) for prospective cohort studies (Wang, 2013) where results were presented as ORs. Estimates of the association from different studies were combined by random-effects meta-analysis (Higgins and Green, 2009). We performed design-specific meta-analyses for cross sectional studies, case control studies, and prospective cohort studies.
The presence of heterogeneity across the studies was assessed by the I2 statistic (Higgins and Thompson, 2002), which indicates the total variation across studies that cannot reasonably be attributed to chance. An I2 statistic less than 25% corresponds to small inconsistency and a value more than 50% to large inconsistency (Ioannidis et al., 2007).
Sensitivity analyses were performed with restriction to the subset of studies that provided data on clinically verified sciatica, and restriction to studies of higher methodological quality. We used a funnel plot to explore small-study bias, Egger’s regression test to examine funnel plot asymmetry, and the trim-and-fill method to estimate the number of studies missing through small-study bias (Duval and Tweedie, 2000; Rothstein et al., 2005). Statistical significance for small-study bias was based on a P value <0.10 (Borenstein et al., 2009 ). We used Stata (metan, metafunnel and metabias commands), version 13 (Stata Corp, College Station, TX) for the meta-analyses. We used natural logarithm of the estimates and their 95% CIs in the meta-analyses.
The electronic searches identified 8121 potentially relevant publications (supplementary figure S1), and the full-text of 513 papers was examined. From 29 studies that addressed leisure time physical activity in relation to lumbar radicular pain or clinically verified sciatica, we excluded: six that were conducted among patients with sciatica but with no control group; one on self-selected volunteers; one on self-selected volunteers and using radiological lumbar disc herniation as an outcome; and one with insufficient quantitative data to estimate an odds ratio. A cross sectional study (Saraux et al., 1999) that assessed an association between playing tennis and lumbar radicular pain but ignored other types of sporting activities was also excluded. Lastly, we excluded one study (Saftic et al., 2006) that assessed only limitation of daily activity (by the SF36 questionnaire). This left 18 (10 cohort [N= 82,024 participants], 4 case control [N=9350] and 4 cross sectional [N= 10,046]) studies qualified for inclusion in meta-analyses. No relevant publications were found that had not been picked up by the original electronic searches. Nine of the 18 studies provided information on risk of clinically verified sciatica, and nine on associations with lumbar radicular pain more broadly.
Seven studies were rated as having low risk of selection bias, nine studies as having moderate risk and two as having high risk. Six studies were assessed as having low risk of performance bias, 10 as having moderate risk and two as having high risk. Thirteen studies were rated as having low risk of attrition bias and five as having moderate risk. All studies presented some risk estimates that were adjusted at least for age and sex.
A meta-analysis of four cross sectional studies showed a higher prevalence of lumbar radicular pain or sciatica in participants who exercised at least once a week compared with inactive participants (pooled PR = 1.29, 95% CI 1.09-1.53, I2 = 0%, N =10,046 participants, Figure 1). In sex-specific analysis, leisure-time physical activity for at least once a week was associated with a higher prevalence of lumbar radicular pain or sciatica in both men and women (Figure 1).
A meta-analysis of four case control studies did not show a statistically significant association between physical activity for at least once a week and clinically verified sciatica (pooled OR = 0.80, 95% CI 0.58-1.08, I2 = 84%, N =9350 participants, Figure 1). A subgroup analysis of two studies on lumbar disc herniation (pooled OR = 0.81, 95% CI 0.47-1.40, I2 = 90%, N =2438 participants) and two studies on hospitalization due to a herniated lumbar disc or sciatica (pooled OR = 0.70, 95% CI 0.30-1.64, I2 = 87%, N =6912 participants) showed similar results.
A meta-analysis of nine prospective cohort studies showed no association between physical activity for at least once a week and lumbar radicular pain or sciatica (Pooled RR = 0.99, 95% CI 0.86-1.13, I2 = 43%, N = 73,008 participants, Figure 2). Three of the afore-mentioned nine studies were on clinically verified sciatica, and the pooled adjusted RR from those three studies was 1.01 (95% CI 0.73-1.41, I2 = 23%, N = 61,032). In sex-specific analysis, leisure-time physical activity for at least once a week was not associated with the new onset of lumbar radicular pain or sciatica in either men or women (Figure 2).
A meta-analysis of two cross sectional studies (Karjalainen et al., 2013; Leino-Arjas et al., 2008) showed a higher prevalence of lumbar radicular pain or sciatica in moderately active participants compared with inactive (Pooled PR = 1.34, 95% CI 1.02-1.77, I2 = 0%, N = 7631 participants), while a meta-analysis of six prospective cohort studies showed an inverse, but non-significant, association between moderate levels of physical activity and lumbar radicular pain or sciatica (Pooled RR = 0.93, 95% CI 0.82-1.05, I2 = 0%, N = 69,049 participants, Figure 3). In a subgroup analysis of six cohort studies, the association was similar for lumbar radicular pain (pooled RR = 0.91, 95% CI 0.76-1.08, I2 = 20%, N = 7017) and clinically verified sciatica (pooled RR = 0.92, 95% CI 0.64-1.31, I2 = 0%, N = 62,032).
A meta-analysis of two cross sectional studies (Karjalainen et al., 2013; Leino-Arjas et al., 2008) showed a higher prevalence of lumbar radicular pain or sciatica, but non-significant, in highly active participants than in inactive participants (Pooled PR = 1.27, 95% CI 0.92-1.74, I2 = 0%, N = 7631 participants). On the other hand, a meta-analysis of seven prospective cohort studies showed a statistically significant inverse association between high level of physical activity and lumbar radicular pain or sciatica (Pooled RR = 0.88, 95% CI 0.78-0.99, I2 = 0%, N = 78,065 participants). In a subgroup analysis, high level of leisure time physical activity was associated with lumbar radicular pain (Pooled RR = 0.84, 95% CI 0.73-0.96, I2 = 0%, 3 studies, N = 7017 participants, Figure 4), but not with clinically verified sciatica (Pooled RR = 1.06, 95% CI 0.81-1.38, I2 = 0%, 4 studies, N = 71,048 participants, Figure 4). In sex-specific analysis (Figure 4), high level of physical activity was associated with lower risk of lumbar radicular pain in both men and women. The association was, however, significant in women only. None of the three studies on lumbar radicular pain was rated as having high risk of selection bias.
There was no heterogeneity among cross sectional studies and cohort studies on moderate or high level of physical activity. The level of heterogeneity was moderate among cohort studies on physical activity for at least once a week vs. no activity. Heterogeneity was high among case control studies on sciatica. There were only four case control studies, which limited exploration of the sources of heterogeneity.
A funnel plot of nine cohort studies on the association between leisure-time physical activity for at least once a week and lumbar radicular pain or sciatica was symmetrical (P for Egger test = 0.54, Supplementary Figure S2). The trim-and-fill method imputed one missing study attributed to small-study bias. A funnel plot of seven prospective cohort studies on the association between high level of leisure-time physical activity and lumbar radicular pain or sciatica was also symmetrical (P for Egger test = 0.24). The trim-and-fill method imputed one missing study due to small-study bias.
This meta-analysis suggests that there is a modest inverse association between leisure-time physical activity and onset of lumbar radicular pain, which might reflect a protective effect. A contrasting positive association with prevalence of lumbar radicular pain in cross sectional studies may have occurred because some patients with sciatica were advised to exercise, or may be a spurious finding.
There are reasons to expect that moderate to high level of leisure-time physical activity might protect against the development of lumbar radicular pain or sciatica. Regular moderate-intensity physical activity during leisure-time improves muscle strength, spinal posture and blood circulation in the skeletal muscles, and reduces visceral adipose tissue and systemic low-grade inflammation (Bruunsgaard, 2005; Rauramaa et al., 2004). Moreover, regular moderate intensity leisure-time physical activity has been shown to prevent the recurrence of sciatica and associated disability (Jensen et al., 2010; Kara et al., 2005). Also, in patients with lumbar radicular pain, low level of leisure-time physical activity predicts greater subsequent intensity of pain and disability (Jensen et al., 2010), and lack of regular leisure-time physical activity is associated with reoperation for recurrent herniated lumbar disc (Kara et al., 2005).
However, the role of leisure-time physical activity in lumbar radicular pain and sciatica has not been widely investigated. After comprehensive literature searches, only 18 studies qualified for our meta-analysis, most of which were based on only small samples of participants. Furthermore, few had been designed specifically to explore the role of leisure-time physical activity in lumbar radicular pain or clinically verified sciatica. The meta-analysis therefore had low statistical power to detect a statistically significant association between leisure-time physical activity and lumbar radicular pain in sex-specific meta-analyses. Nevertheless, the confidence intervals around meta-estimates of risk for both sexes combined were reasonably tight, and it seems unlikely that a large protective effect of leisure-time physical activity (relative risk < 0.70) has been missed simply by chance.
Other limitations of the evidence base relate to the populations studied, the assessment and classification of exposures, heterogeneity of clinical outcomes investigated, and incomplete control for potential confounders. Two-thirds of the included studies were conducted in Finland, which may limit the generalizability of the findings. Most examined associations of lumbar radicular pain or sciatica only with specific types of physical activity during leisure-time such as walking or jogging. They did not include all types of physical activity during leisure-time and did not use a physical activity index or metabolic equivalent of task (MET) method. In addition, they classified leisure-time physical activity in different ways. Any resultant non-differential misclassification of exposure would be expected to bias associations towards the null. There were also differences in the comparison groups used to derive risk estimates. Some studies used low level of leisure-time physical activity as the comparison group, some studies used low/moderate level, and some studies used high level of physical activity.
As regards the specification of clinical outcomes, some studies focused on self-reported lumbar radicular pain, which is less specific than clinically verified sciatica. Others looked at hospitalization or surgery for sciatica, which is more specific, but occurs less frequently, limiting statistical power. For example, one cohort study (Mattila et al., 2008), which recruited over 57 thousand participants and took surgery for lumbar disc herniation as an outcome, contributed rather little to the pooled estimate of risk because only 0.4% of the participants underwent lumbar discectomy.
Many studies applied only limited control for potential confounding factors, failing to take into account body mass index and exposure to occupational physical activities. Thus, the inverse association between leisure-time physical activity and lumbar radicular pain or sciatica may have occurred because people who are obese or exposed to physically demanding work are less likely than others to exercise during their leisure-time (Seiluri et al., 2011).
Only one reviewer screened the abstracts and selected studies for the meta-analysis. This may be considered as a limitation of the review process. However, many databases have been searched several times. The applied exclusion criteria for this review were clear cut (having no control group, self-selected volunteers, or not reporting quantitative data). Moreover, the quality assessment was carried out by two reviewers.
The significant positive association of leisure-time physical activity with prevalent lumbar radicular pain in cross-sectional studies is a further reason for caution in interpretation. It may reflect reverse causation (i.e. patients with sciatica being advised to take exercise), but it could also be an indication that the reduced risk of onset in cohort studies is unrepresentatively low. Again, it weighs against a larger protective effect having been missed.
This systematic review and meta-analysis suggests that moderate to high level of physical activity during leisure-time may have a moderate protective effect against the development of lumbar radicular pain. However, a large reduction in risk (greater than 30%) seems unlikely.
This study was supported by the Finnish Ministry of Education and Culture (grant no: 253715).
Conflict of interest
The authors declare that they have no conflicts of interest.
PubMed, Embase, Web of Science, Scopus, Google Scholar and ResearchGate.
What does this review adds
Leisure-time physical activity reduces the risk of developing lumbar radicular pain.
ContributorsRS developed the review protocol, conducted the literature searches and extracted data. RS and KFH rated the methodological quality of included studies. RS performed the meta-analyses and drafted the manuscript. All authors interpreted the results and contributed to the writing of the manuscript. All have approved the final manuscript to be submitted for publication.