Prevalence and incidence of low back pain among runners: a systematic review

Background Running is one of the most popular sports worldwide. Despite low back pain (LBP) represents the most common musculoskeletal disorder in population and in sports, there is currently sparse evidence about prevalence, incidence and risk factors for LBP among runners. The aims of this systematic review were to investigate among runners: prevalence and incidence of LBP and specific risk factors for the onset of LBP. Methods A systematic review has been conducted according to the guidelines of the PRISMA statement. The research was conducted in the following databases from their inception to 31st of July 2019: PubMed; CINAHL; Google Scholar; Ovid; PsycINFO; PSYNDEX; Embase; SPORTDiscus; Scientific Electronic Library Online; Cochrane Library and Web of Science. The checklists of The Joanna Briggs Institute Critical Appraisal tools were used to investigate the risk of bias of the included studies. Results Nineteen studies were included and the interrater agreement for full-text selection was good (K = 0.78; 0.61–0.80 IC 95%). Overall, low values of prevalence (0.7–20.2%) and incidence (0.3–22%) of LBP among runners were reported. Most reported risk factors were: running for more than 6 years; body mass index > 24; higher physical height; not performing traditional aerobics activity weekly; restricted range of motion of hip flexion; difference between leg-length; poor hamstrings and back flexibility. Conclusions: Prevalence and incidence of LBP among runners are low compared to the others running related injuries and to general, or specific population of athletes. View the low level of incidence and prevalence of LBP, running could be interpreted as a protective factor against the onset of LBP. Systematic review registration PROSPERO CRD42018102001.


Keypoints
Prevalence and incidence of LBP among runners seem basically low if compared with general population and other popular sports activities; Running could, cautiously, be considered a protective factor for the lumbar spine; Risk factors for the onset of LBP are generally physical impairments or training methods-related factors that could be partly modified and managed in clinical practice; Scarcity and methodological weakness of the available studies invite to conduct further research about actual prevalence and incidence as well as risk factors for LBP among runners; LBP may be better defined as Running Related Disorder instead of Running Related Injury.

Background
Running is one of the most practiced sports in the adult population worldwide, due to the sustainable cost of technical materials and its great beneficial impacts on health [1][2][3][4][5][6][7][8][9][10][11]. The benefits of running include weight control and prevention of chronic health disorders, such as the cardiovascular diseases, resulting in a general reduction of mortality risk [1][2][3][4][5][6]. The health benefits associated with running are well-documented, nevertheless the attention to lifestyle, diet, fitness and competitive athletics promoted by media in the last decade, have led to a drastic increase of the levels of physical activity and interest in both competitive and recreational running, even in subjects without an appropriate knowledge on training methodology [3][4][5][6][7][8]. Although evidence suggests that running is one of the most effective ways to achieve a good state of health and fitness [9], recent studies indicate that it also involves a relatively high risk of associated injury [10,11]. Several authors have reported that 11-85% of recreational runners have at least one Running Related Injuries (RRIs) each year [10], resulting in a reduction or interruption of training in 30-90% of runners [11][12][13]. Acute RRIs are rare, almost 80% of RRIs are due to overuse, resulting from an imbalance between the resistance capacity of connective tissue and the biomechanical load of running [14,15]. The prevalence rate of RRIs among middle and long-distance runners has been reported to range between 19 and 92% [2,[16][17][18][19][20]. However, the discrepancies among studies limit the comparison of data due to the divergences in the type of runners analyzed, follow-up provided, study design, etiology and definition of RRIs [1,2,[14][15][16][17][18][19][20][21][22][23][24][25]. In 2015, Yamato et al. [20] defined the RRIs as musculoskeletal pain or physical complaint of the lower limbs or of the back/trunk due to running, causing a total restriction or interruption of running for at least seven or more days and requiring therapeutic assistance [20]. Currently a definition of RRIs is not yet fully share [20], this is reflected in the difficulty of analyzing the studies about of RRIs [18]. RRIs therefore primarily affect joints of the lower limb, pelvis and lumbar spine [18,25,26], causing painful muscles, tendons and joints, often resulting in low back pain (LBP) [14][15][16][17][18][19][20][21][22][23][24][25][26]. It is frequent in clinical practice [27][28][29][30][31][32][33], that patients contact physical therapists for consultancy on LBP which represents a common complaint of athletes [27][28][29][30][31][32][33]. In the 90% of the cases, LBP is defined as non-specific, because the pathoanatomical musculoskeletal causes are not clearly identifiable [34]. LBP is one of the most common health problems in the world, that 80% of adults experience at some point in their life [35,36]. Despite many published studies on the prevalence and incidence of LBP, there is not a clear consensus regarding its actual epidemiologic impact [37][38][39][40]. Indeed, some evidence reported a point prevalence estimate of LBP that ranged from 1 to 58% (mean 18.1%) [39,40]. One-year and lifetime's prevalence of LBP in the general worldwide population, ranged between 0.8-82.5% (mean 38,1%) and 11-84% (mean 47,1%), respectively [39,40]. Similarly, regarding a population of athletes [41], the percentage values of the prevalence of LBP remains wide, namely 1-94% in the lifetime (highest prevalence in rowing and crosscountry skiing) [41], and 18-65% for the point prevalence (lowest prevalence in basketball and highest prevalence in rowing) [41].
As seen in the general population, a big amount of athletes also experiences LBP [41][42][43][44][45][46][47][48]. Moreover, athletes of particular sport disciplines such as ski, rowing, golf, volleyball, track and fields, swimming or gymnastics are at greater risk of suffering from LBP than nonathletes population [33,41,[43][44][45][46][47][48]. The incidence rates of low back pain in athletes have been reported up to 30% depending on the specific sport they are involved in [49]. However different authors describe also a great variability in prevalence rates, that have been reported in a range from 66% [50,51] to 88.5%, respectively in young athletes and in elite athletes [52]. The incidence rate constitutes the frequency of new events of a medical disorder in the studied population considered at risk, calculated in a given period of time [53]. On the other hand, the prevalence proportion is the part (in percentage) of a defined population affected by a particular medical disorder at a given point in time, or over a specified period of time [53].
Despite several studies about the prevalence and incidence of LBP in general population and sports are retrievable [35][36][37][38][39][40][41], it seems that this topic has not been clearly investigated in the runners. Researches are mainly focused on RRIs in general but there are not Systematic Reviews (SRs) specifically addressing prevalence, incidence and risk factors for LBP in runners [11,18]. Moreover, earlier literature of LBP has been addressed to a wide range of sports or athletes [31,54] and no conclusive data were published peculiarly on the LBP among a specific population of runners. For this reason, the aims of this systematic review (SR) were to investigate among runners: 1) the prevalence and the incidence of LBP; and 2) specific risk factors for the onset of LBP.

Study design and protocol
The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) protocol was used to design the present SR [55]. This SR has been registered in PROSPERO database (number CRD42018102001).

Search strategy
An electronic literature search was conducted in the following databases from their inception to 31st of July 2019: PubMed, CINAHL (EBSCO), Google Scholar, Ovid, PsycINFO, PSYNDEX, Embase, SPORTDiscus, Scientific Electronic Library Online (SciELO), Cochrane Library and Web of Science. Research strategies were conducted and designed depending on the specific settings of each database with the supervision of an expert librarian. The research strings were developed according to the PICO model of clinical question (participants, interventions, comparison and outcomes). Free-terms or synonyms (e.g. runners; risk factors; running-related injury), and when possible MeSH (Medical Subject Headings) terms (e.g. low back pain; prevalence; incidence) were used and combined with Boolean operators (AND, OR, NOT). Additionally, a manual research has been conducted through the bibliographies of all the assessed studies to obtain an integrative cross-references full-text selection. A dedicated search strategy was prepared for each database. We have reported the full search strategy for PubMed in Additional file 1.

Eligibility criteria
All studies were conducted on runners without age limitation. We included any type of study design aiming to investigate prevalence, incidence or risk factors for LBP as RRIs (e.g., cross-sectional, case-control, prospective and retrospective cohort studies). Moreover, single cohort designs were also considered. Runners of any kind of experience or mileage were included, whereas sprinters and track and field athletes were not considered. We defined as RRI any occurrence severe enough to avoid or even restrict the running activity for at least 24 h. We selected studies reporting at least one anatomical area included in LBP definition such as area located below the margin of the 12th rib and above inferior gluteal fold (included: pelvis/pelvis crest, sacrum and gluteus/buttock) [56] and the pathoanatomical cause of the pain cannot be determined [57]. We selected studies published in English or Italian language without limits of date of publication. Descriptive observational designs, such as case report and case series, and any study, which did not meet the inclusion criteria, were excluded.

Study selection
The selection and data collection process were done by two reviewers (FM and AC) under the supervision of a third author (MT). The whole records were screened by the management software for systematic reviews "Rayyan" (https://rayyan.qcri.org), while references were managed by the "Mendeley" software (https://www.mendeley.com). After the removal of the duplicates, titles and abstracts were screened. Then, full-texts of the identified studies were obtained for further assessment and analyzed independently according to the eligibility criteria by two reviewers (FM and AC). Where appropriate, authors were contacted in order to obtain the full-text.

Data collection
For each article, the following data was extracted: study design; author, year of publication; the number and characteristics of participants/populations; international definition and/or any diagnostic criteria for LBP; analysis of the variables and the outcome of the studies; study settings/country (e.g., marathon, half-marathon, survey, lab analysis); prevalence and incidence rates; intervention and results; follow-up or study duration; theoretical perspectives on potential risk factors on the onset of LBP: reported risk factors; outcomes and measurements to associate the risks associated with LBP (e.g., relative risk, odds ratio, etc.).

Quality assessment
The Risk of Bias (RoB) of the included studies is analyzed using the Joanna Briggs Institute Critical Appraisal tools [58] according to the specific study design (e.g., prevalence data, cross-sectional studies, case-control studies, prospective studies). In addition, as prevalence data may be sourced from several numbers of study designs, a critical appraisal checklist specifically for prevalence studies were used. Two independent researchers (FM, LS) evaluated the RoB. The score of RoB was not adopted as criteria to include/exclude studies in this review.

Data analysis
We reported the data related to the prevalence, incidence and risk factors for LBP from each study. When needed, we estimated data on prevalence, incidence and risk factors using available data of the included studies. We reported the prevalence and incidence percentage in table form.

Study selection process
Electronic database searches and the identification of additional references yielded 14,575 records, including 3952 duplicates that were removed. After screening titles and abstracts, 10,564 (including 2 full-text not available) records were excluded. Then, 59 potentially relevant studies were considered eligible for full-text assessment, resulting in 19 included studies in this SR for quality assessment, data extraction and analysis. The selection   process is described in Fig. 1 according to the PRISMA Statement [55]. Reasons for exclusions are reported in Table 1.

Risk of Bias of the included studies
Details of the RoB of the included studies are presented in Tables 3, 4, 5 and 6. Most items of all RoB assessment tools used for the quality assessment were rated as low risk. For all the studies addressing prevalence data regardless of the study design [63-65, 67, 68, 70, 81], the items rated as unclear RoB were related to the sampling methods in 3 studies [63,65,70], while in one study the items rated as high risk [68]. More in depth in one study [68], another two items were rated as high risk, one regarding the reliability of the condition measurement and one regarding the validity of identification of the condition. For cross-sectional studies, the majority of studies had low and, less commonly, unclear RoB [63][64][65][66][67][68]. However, among them, in the study of Marti et al. [68], the item related to the criteria for inclusion was rated as high risk, likewise the item about the reliability of the condition measurement in the study of Chang et al. [67]. For retrospective studies [69][70][71] there was a low RoB across all the studies, apart from comparability of groups, matching of cases and controls, adoption of the same criteria for identification of case and controls and methods to measure the exposure in 3 studies [69][70][71], which were all rated as not applicable. Finally, for prospective studies [72][73][74][75][76][77][78][79][80][81], in 6 studies [71,[73][74][75][76]79] items related to the similarity/recruitment of groups, methods of exposure were rated as not applicable. Also were judged as not applicable the items related to the time of follow-up and loss to follow-up in the study of Back et al. [72]. Moreover, in 3 studies [72,74,78] the item about strategies to address incomplete follow up was evaluated as not applicable, whereas the remaining items were commonly judged as low RoB.

Summary of findings
Results about prevalence and incidence are reported in Table 7.          cross-sectional studies [63][64][65][66][67][68], one was a retrospective study [70] and one was a prospective study [81]. The range of point prevalence ranged from a minimum of 0, 7% [66] at a maximum of 13.6% [63] and lifetime prevalence ranged from a minimum of 3.2% [67] a maximum of 20.2% [64]. Point prevalence values, 13.6 and 0.7% respectively, were reported in two studies [58,64]. Five studies reported values of 1-year prevalence [65,66,68,70,81], ranged from 14% [65,70] to 0.7% [68], and those about lifetime prevalence were two [63,64], 3.2 and 20.2% respectively. Only 1 study [66] addressed data for point and 1-year prevalence, with values of 0.7 and 13.5% respectively [66]. Also in the cross-sectional survey study of Woolf et al. [63], the point prevalence of LBP in runners was reported, and it was equal to 13.6%. The study of Marti et al. [68] reported a 1-year prevalence of LBP of 0.7%, but this value was calculated in a sample of all male runners, and it was referred only to the Grade III injuries (defined as full training involuntary interruption of running for at least 2 weeks duration). In the cross-sectional study of Teixeira et al. [65] 1-year prevalence of LBP (including pain in the lumbar spine and pain in pelvic/sacral/gluteus regions) among elite marathon runners was 14%. In the retrospective descriptive study of Ellapen et al. [70] the 1-year prevalence of lower back (including hip) among recreational halfmarathon runners was 14% (mean value; 13% for men, 15% for women). In the only prospective cohort study,  Walter et al. [81], the 1-year prevalence of LBP among 1288 runners was 4.3%. The highest lifetime prevalence rate of LBP was reported to be 20.2% in the crosssectional study of Malliaropoulos et al. [64] in a sample of 40 ultra-trail runners. Furthermore, in another cross sectional study [67], the LBP lifetime prevalence was 3.2% [67], in a sample of 893 subjects of which 80% of male runners [67].

Incidence of LBP
Twelve [69,[71][72][73][74][75][76][77][78][79][80][81] of the 19 included studies addressed incidence of LBP among runners. Ten were prospective  * higher credits as a sum of sex and age of the runner, difficulty level of previous racespositive height difference, the vertical climb index, and the distance in kmand performance.1 Previous LBP; a Subgroup of runners without insert; 2 Current LBP; †value calculated by authors using data from the full-text; ‡ value reported directly from the full-text studies [72][73][74][75][76][77][78][79][80][81], and two were a retrospective study [69,71]. Overall, the incidence of LBP among runners ranged from a minimum of 0.35% (in 6 weeks) [76] and maximum value of 22% (in 1-year) [72]. The highest incidence rate of LBP was reported as equal to 22% (7 male; 3 female) in the prospective study of Bach et al. [72] in a sample of 45 runners. The minimal rate of incidence, below 1%, was found in the studies by Kluitenberg et al. [76] and Rasmussen et al. [69] with values of 0.35 and 0.5%, respectively. Furthermore, overall low incidence values, beneath 5%, were found in other six studies [69,73,75,77,78,80,81]. Among them, a value of 1.6% (in 13 weeks) was found in the prospective study of Tauton et al. [73] for the distribution of injuries in the lower back. A similar value (1.8%) was found in Walter et al. [81]. In a more recent prospective cohort study of Dallinga et al. [80] an incidence rate of 1.9% (in 12 weeks) was found in a sample of recreational runners, during the training period for a running event. More in depth, the analysis of Van Der Worp et al. [77] showed a rate of 2.7% (in 12 weeks) in a sample of adult women runners. Moreover, the prospective cohort study of Von Rosen et al. [78] reported the incidence of injuries in the lower back of 2.8% of all injuries recorded between young female runners (mean age 17 years). Lastly, in the study of Buist et al. [73], a value of 4.8% (in 8 weeks) was found among a sample of novice runners, in runners with previous experience who have started running again and runners engaged in regular running [73]..
In the remaining two prospective cohort studies [74,79], the incidence rate of LBP was found to be slightly higher. Indeed, Lysholm et al. [74] reported a 1-years incidence equal to 5% among a small sample of 39 runners and in the recent study of Messier et al. [79] the incidence (in 2 years) of LBP among runners was 6%, considering the anatomical sites of back and pelvis. In the end, the retrospective analysis of Clement et al. [69] among 1650 runners revealed similar findings: the 2 years-incidence of injuries localized in the lower back was 3.7% (3.3% for men and 4.3% for women).

Risk factors for LBP
The risk factors for the onset of LBP are reported in Table 8. Four studies [63,64,69,70] addressed specific risk factors for LBP in runners. Two of them were retrospective studies [69,70] and two were cross sectional studies [63,64]. The retrospective analysis of Clement et al. [69] indicated as possible risk factors for the development of non-specific back pain in runners leg-length discrepancy, poor hamstrings flexibility and poor back flexibility [69]. However, the authors did not specify the strength of the associations with LBP and the values of statistical significance. In another retrospective study [70] on recreational runners' tightness of hip flexors and hip flexion angle measured both with the Thomas test (a clinical and physical test used to measure the flexibility of the hip flexors, which includes the iliopsoas muscle group, the rectus femoris, pectineus, gracillis as well as the tensor fascia latae and the sartorius [82] and goniometer were defined as potential intrinsic factors predisposing to lower back/hip injuries. Indeed, the hip flexion angles of female runners who have suffered lower back/ hip musculoskeletal injuries were significantly greater, than those of their not-injured counterparts (p < 0.01) [70]. Risk factors are listed in Table 8.
Moreover, the cross-sectional study of Malliaropoulos et al. [64] highlighted that having more than 6 years of experience of running could represent a predicting factor for getting injured in the lower back (p = 0.012) [64].
Lastly, Woolf et al. [63], in a cross sectional study conducted on a wide sample of runners, showed that runners who have previously suffered LBP, have reported greater shoe wear on either the inside or outside. Conversely, an equal shoe wear was less likely to relate a previous history of LBP (p = 0.034) [63].
In the same study [63], a previous history of LBP was reported by runners who did not use orthotics (such as insert, insole, heel, foot-bed, etc.) (p = 0.011), by who had a body mass index higher than 24 (p < 0.01) and by who did not perform weekly traditional aerobics activity (p < 0.05).
Moreover, again in the study of Woolf et al. [63], runners who did not regularly play contact sports (e.g., football, soccer, basketball, wrestling, boxing, rugby) were more likely (p < 0.04) to have suffered LBP than those who do [63]. Current LBP was reported by high stature (p < =0.02) runners and by who perform a long time flexibility exercises routine before the training (p < = 0.05) [63].

Discussion
The aim of this SR was to investigate the prevalence and incidence of LBP and to identify risk factors for the onset of LBP among runners. To the best of the authors' knowledge, this is the first SR addressing these outcomes in this specific population.

Prevalence and incidence of LBP
Despite running is one of the most practiced sports worldwide, and the prevalence rate of RRI is well documented in scientific literature [1][2][3][4][5][6][7][8][9][10][11], prevalence and incidence of LBP among runners are still unclear. The relatively low number of studies that we were able to include in the present review confirms the scarcity of literature on this topic.
Overall, the findings of this SR revealed that LBP prevalence and incidence among runners, compared to the general population [35][36][37][38][39][40], were low. In detail, according to the results, within the most represented sports population among the analyzed studies (i.e. 20-50 years of age), running does not appear to be related to higher rates of incidence and prevalence of LBP, if compared to the general population presenting same age [40,83,84].
Moreover, it should be noted that the results of two studies reporting high prevalence (20.2% lifetime) [64], and high incidence (22% 1-year) [72], is probably depending from the very small [64,72] and the specific sample of 40 ultra-trail runners (that face with races taking place on mountain, desert, or forest and it includes uphill, downhill and is similar in duration to an ultramarathon, that is beyond the distance of a regular marathon of 42.195 km) [64].
Regarding the mileage, it is worth pointing out that LBP prevalence in runners seems to be somehow independent of the running distance. In Besomi et al. [66] the largest sample (4380) within studies included in our SR, prevalence was assessed on a race of three difference distances (10, 21 and 42 km). The rate of prevalence in the 42 km-runners was similar (7.5%) to the rate among the 21 km-runners, (7.5%).
Although prevalence and incidence of LBP appear low if compared to the general population, this conclusion should be taken cautiously. Indeed, out of the scarcity of the available studies, there are many points in the included studies that weaken the generalizability of this statement.
Further studies are needed in order to extend the results of this systematic review to the older adult population, because LBP incidence and prevalence values increase with age [83,84].
Transversely to all the included studies, participants were heterogeneous for individual characteristics (age, sex), training level and previous injuries. Therefore, it is reasonable that various samples of populations (ex. young elite athletes or middle-aged recreational runners) may led to different prevalence or incidence rates. Furthermore, as reported in some prospective studies, not all the participants were exposed to the same running or training methods. In the cross-sectional survey study of Woolf et al. [63] for example, the rate of LBP pointprevalence, was calculated not only within experienced runners, but also between novice runners. Instead, in the study of Marti et al. [64] the 1-year prevalence of LBP, 0.7%, was estimated in a wide sample, 4358 runners, but constituted of only male runners; In the cross-sectional study of Teixeira et al. [65] which is the only one to report the International Association for the Study of Pain (IASP) definition of pain [86], the prevalence of LBP was calculated among elite marathon runners who compete at international and/or national level and perform high volume of training, up to 160 km/week. In the crosssectional study of Chang et al. [67] in a sample of 893 runners (mostly composed of male) although the lifetime prevalence rate was low, 3.2%, runners were not specifically asked if they had the symptom at the time of completing the questionnaire. Concerning the incidence, in the prospective study of Bach et al. [72] the highest rate of LBP (22%; 7 males, 3 females) was found within a small sample of 45 runners. In the two prospective cohort studies [77,78], the rate of incidence was assessed in samples made up exclusively of female runners and four studies evaluated incidence rates of LBP in only novice runners [75][76][77]79]. Clement et al. [69] was the only study that used the term "Non-specific lower back pain", as reported by the most recent literature [57,88] and only seven among the included studies [63,64,66,69,72,74,78], to define an injury affecting the lumbar spine, adopted specific terms such as low/lower back pain, LBP, Non-specific lower back pain.

Risk factors
Only four studies addressed specific risk factors for the onset of LBP among runners [63,64,69,70] and great caution is required for translating their results to general practice being those studies two retrospective studies [69,70] and the two cross-sectional studies [63,64], which do not represent the most reliable study design to assess risk factors.
According to the Comprehensive Model for Injury Causation [60] and the Conceptual Model for the Determinants of RRIs [89], intrinsic and extrinsic factors are responsible for the increase of running injury risk. Intrinsic factors are hardly or not modifiable; they include sex [60,89], age [60,89], BMI [60,89], history of previous injury [60,89], physical fitness and psychological factor have been found to predispose runners to injury [60,89]. Otherwise, extrinsic factors are modifiable, and comprise training volume or other characteristics, as sport equipment and training environment, which rise the runner's susceptibility to injury [60,89]. Intrinsic risk factors proposed for the onset of LBP among runners included: BMI ≥24 [63]; higher physical height [63]; tightness of hip flexors (measured by Thomas Test) [70] and hip flexion angles (only in female and measured by goniometer) [70]; but, as referred by the authors, there is no strong literature to explain this two last finding [70]. Moreover, the identification by Clement et al. [67] of physical impairments, like reduced hamstring or back flexibility and leg length discrepancy, was not supported by statistical evaluation. Notably, if the runners are compared to non-runners, these seem to present a significantly lower degree of hip flexion with the knee extended, indicating a tightness of hamstrings (p < 0.001). Nonetheless, no correlation was found between muscular tightness in runners and the incidence of LBP [72]. Due to the scarcity of available studies and the clinical impression that muscles tightness could be a risk factor for RRIs and LBP, this topic should be investigated in large samples using prospective design.
The main extrinsic risk factors for the onset of LBP among runners were: high competitive level [64]; more than 6 years of experience in running [64]; some patterns of shoes' wear [63] and do not performing weekly aerobics activity [63].
Also in this case the findings extracted from the two selected studies [63,64] cannot be directly translated to the daily practice, but could only serve as possible additional elements to support the clinician in the interpretation of the athlete's condition. Indeed, the exposure to a single risk factor is often insufficient to produce an overuse injury: the RRI is the result of a number of superposing factors (like training increase, muscular impairments, unsuitable equipment, etc.) [66].

Consistency
There is a need of a standard and internationally acceptable definitions for LBP and a clearer definition and terminology of RRI. RRI is defined as an overuse injury due to an unbalance between the resistance capacity of connective tissue and the biomechanical solicitations of running [14,15]. Therefore, here the meaning of "injury" differs from usual meaning which is related to an acute trauma and, in a clinical perspective, very rare among runners [3,14,15].
In our view, a more suitable word may be "Disorder" (Running Related Disorders -RRDs) that better describes multifactorial conditions which include, beside structural aspect, also psychosocial elements often present in nonspecific painful disorders like LBP [90][91][92][93].
Our SR confirmed, also for running, the findings of a recent SR [33] which concluded that the evidence about prevalence of LBP in athletes of some popular sports are scarce and derived from studies not of good methodological quality. This SR showed a quite high LBP prevalence among athletes, but this finding was relative to a wider sample of sports including volleyball, track and fields, swimming, golf, ski, gymnastics and rowing [33,[43][44][45][46][47][48], not specifically including running.

Clinical implications
Despite a significant correlation between spinal shrinkage, running speed and distance covered exists, it is not correlated to the onset or presence of LBP [94]. Moreover, some studies suggested that running could have an anabolic role towards the intervertebral disc [95][96][97][98], among them Belavý et al. [97] reported that longdistance runners and joggers showed better hydration and glycosaminoglycan levels than the non-athletic individuals.
These findings, together with the low level of incidence and prevalence of LBP among runners, cautiously invite thinking running as a protective factor from LBP and to consider of prescribing running as a preventive exercise for LBP.
Although the data available on risk factors are weak and not conclusive, nevertheless most of proposed running related risk factors were modifiable by specific intervention and adapted training and they should be taken into consideration by physical therapists and trainers.

Implications for research
More high-quality studies that analyse the prevalence and incidence of LBP in runners are needed before drawing strong and definitive conclusions. The actual prevalence and incidence of LBP in runners should be investigated by large cohort studies, adopting better definition of the clinical symptoms, rather than just pain distribution in anatomical districts. Moreover, a consensus on the definition of RRIs that consider the inclusion of psychosocial aspect and widens the usual pathoanatomic approach is advisable due the characteristics of conditions like LBP.
Risk factors should be assessed by methodologically sound prospective studies on more homogeneous populations (in terms of demographic characteristics, training level of participants, gender, age, etc.). As reported by our results, running seems to represent a sport without an increased rate of LBP: data of LBP prevalence and incidence among runners are lower than those found in other sports [33,41]. However, caution is required when assuming that run could be a good practice in order to prevent LBP. Our findings mostly derive from novice runners or recreational runners, so it would be pretentious to apply in other sports and in elite/professional contexts.

Perspective
Running is one of the most practiced sports [1][2][3][4][5][6][7][8][9][10][11] and although evidence suggests that is one of the most effective ways to achieve a good state of health [9], recent studies indicate that it also involves a relatively high risk of injuries [10,11]. Currently a definition of RRIs is not yet fully share, this is reflected in the difficulty of analyzing the studies about of RRIs [18]. RRIs primarily affect joints of the lower limb and lumbar spine [18,25,26,99], causing painful muscles, tendons and joints, also resulting in LBP [14][15][16][17][18][19][20][21][22][23][24][25][26], but despite several studies about the prevalence and incidence of LBP in sports are retrievable [35][36][37][38][39][40], it seems that this topic has not been clearly investigated in the runners. Therefore, the aetiology, the prevalence and the incidence of LBP, likewise the RRI, have been reviewed. Specifically, it is important to consider how often the effectiveness of a given RRI prevention intervention is dependent on an easy modification of etiologic factors, and on and their consistency with a biologically plausible causal mechanism [24]. Therefore, the investigation of how different factors affect the lumbar spine, in terms of structure-specific load and/or loadability, and the dose-response relationship between running participation and injury risk [24]. These considerations allow researchers to move beyond traditional risk factor identification. Just so, research findings could be reliable, not only in terms of the observed cause-effect association but also translatable in clinical practice [24].
Moreover, although the encouraging results, they are limited to the population most represented among the studies analysed (20-50 years of age), from this perspective, running activity could be used as a strategy to maintain a healthy lifestyle in the adult population, as indicated by the World Health Organization guidelines (WHO) [100].

Limitations
This SR has several limits. Studies written in languages other than English or Italian were excluded and, due to the heterogeneity of the included studies were not possible to perform a meta-analysis.
We have not included sprinter runners in the research strategies, as keyword. Moreover, we have adopted a strict topographical definition of LBP [56], while in other epidemiological studies the authors referred as a LBP a generic "back pain", which could involve even the thoracic region [34,40,83,84,101]. Moreover, a homogeneous definition of LBP was not adopted in all studies, populations investigated were different and prevalence, incidence or risk factors for the onset of LBP are investigated by questionnaires that are exposed to recall bias.
Furthermore there is a high risk of selection bias in the studies, in that persons with LBP may not be able to run, increasing the rate of prevalence of LBP in general population.
Lastly, being unavailable a specific and validated assessment tool for retrospective studies, for the assessment of the methodological quality of the included studies was adopted the tool designed for case-control studies.
Finally, it would be also necessary to tighten up the definitions of both incidence and prevalence rates [53], which are sometimes confused or inverted, and therefore create difficulties in the interpretation of data.

Conclusion
Despite the small number of included studies, the heterogeneity of the samples investigated and of running modalities did not allow to gain conclusive results, the prevalence and incidence of LBP among runners appear to be low if compared to the general population and to other RRIs. Most of the physical and training-related risk factors for the onset of LBP, even based on weak evidence, are potentially modifiable by a careful intervention of the clinician and should be considered when LBP prevention is sought.