We screened 1981 titles and abstracts for eligibility (Fig. 1). Of these, 118 full-texts were screened. We critically appraised eight eligible cohort studies [24,25,26,27,28,29,30,31] and all were deemed of acceptable quality (i.e., low risk of bias). No additional studies were found with hand-searching of reference lists of eligible studies.
Risk of bias
The accepted studies had some methodological limitations (Table 1). For example, in most studies, it was unclear (and marked as “can’t say” according to the SIGN criteria) if the method of exposure assessment was reliable (6/8) [24,25,26,27, 30, 31] and there was no evidence that the method of outcome assessment was valid and/or reliable (6/8) [24, 25, 27, 29,30,31]. Additionally, it was unclear (“can’t say”) if the assessment of outcome was made blind to exposure status or if there was recognition that knowledge of exposure status could have influenced the assessment of outcome in all studies where these criteria were applicable (3/3) [27, 30, 31]. Potential confounders were not clearly identified in two studies [29, 31] that aimed to assess causal factors; therefore, we synthesised risk factors as risk markers rather than determinants.
A summary of study characteristics is presented in Table 2. The majority of eligible studies were conducted in the U.S. (5/8) [24, 25, 27, 29, 30], and one each were conducted in Sweden , Finland , and Israel . Two studies assessed Marines [24, 28], two assessed Army personnel [27, 29], and the remaining studies assessed the military as a whole [25, 26, 30, 31].
All studies were cohort studies (5/8 single-group cohorts [24, 26, 28, 29, 31]), which assessed risk factors for incident LBP in the active military population. Three studies were prospective cohort studies [26, 28, 29], while five were historical cohort studies conducted using pre-existing administrative and/or clinical data [24, 25, 27, 30, 31]. All studies examined non-causal associations between candidate risk factors and incident LBP, as there were either no clear a priori variables defined as potentially important for predicting incident LBP by the studies or the necessary confounding variables were not identified a priori and 1controlled for; therefore, only risk markers were identified. No studies identified included prediction or causal modelling; therefore, risk predictors and risk determinants could not be identified. Half of the studies examined risk factors in more than one category (e.g., physical, sociodemographic, and/or occupational) [24,25,26, 29]. Four studies examined physical risk factors (e.g., physical fitness, body characteristics) [26,27,28,29], three studies examined sociodemographic risk factors (e.g., age, sex, education) [24,25,26], and six studies examined occupational risk factors (e.g., occupational tasks, military service) [24,25,26, 29,30,31]. No studies assessed psychological risk factors for LBP.
Overview of risk factors
In the eight studies included in our review, 37 risk factors (all risk markers) were examined: 13 physical factors, 16 sociodemographic factors, and 8 occupational factors. Among prospective cohort studies, all used self-reported questionnaires to identify the risk factors [26, 28, 29]. The historical cohort studies used administrative data to identify the risk factors [24, 25, 27, 30, 31]. There were no consistent confounding variables that were adjusted for by all studies; however, age (5/8) [24,25,26,27, 30] and sex (4/8) [24, 25, 27, 28] were most commonly adjusted for. The outcomes and key findings for each risk factor studied is presented in Table 3.
Consistent associations between physical risk factors and LBP
A history of LBP demonstrated a consistent association with LBP during active duty military service [28, 29]. Monnier et al. concluded that back pain within six months prior was a risk factor for both LBP (HR 2.47, 95% CI 1.41–4.31) and LBP limiting work ability (HR 3.58, 95% CI 1.44–8.90)  and Roy and Lopez concluded that a history of LBP prior to military service was associated with LBP in the Brigade Support Battalion (OR 5.03, 95% CI 1.61–15.72), the Brigade Special Troops Battalion (OR 8.91, 95% CI 1.71–46.46), and the Infantry Battalion (OR 2.20, 95% CI 1.2–4.04) compared to those without a history of LBP . Similarly, previous injury (e.g., lower extremity injury or sports injury) consistently demonstrated an association with LBP [26, 27]. Taanila et al.  concluded that having a sports injury during the prior month (HR 1.7, 95% CI 1.0–2.8) was a risk factor for LBP, while Seay et al.  concluded that lower extremity injury was a risk factor for LBP (HR 1.70, 95% CI 1.66–1.74) irrespective of sex (males—HR 1.76, 95% CI 1.72–1.80; females—HR 1.43, 95% CI 1.36–1.50). The amount of time spent on physical training also had an association with LBP, with one study demonstrating that those participating in fewer physical training sessions per week had a greater risk of LBP limiting work ability than those participating in more physical training sessions per week (HR 2.96, 95% CI 1.19–7.39) , while another study demonstrated that participation in more strength training was associated with a lower risk for LBP (OR 0.88, 95% CI 0.78–0.99) .
Non-associations between physical risk factors and LBP
Based on a single study by Taanila et al., the following factors were not found to be associated with LBP: body mass index, waist circumference, self-assessed health, chronic disease, regular medications, orthopedic surgery, chronic impairment due to prior MSK injury, and self-assessed physical fitness .
Inconsistent associations between physical risk factors and LBP
There was conflicting evidence on whether poor performance on various physical fitness tests were associated with LBP. For example, Monnier et al. demonstrated that performing less pull ups was associated with incident LBP (HR 1.87, 95% CI 1.17–3.01) , but Taanila et al. found no association . Similarly, Taanila et al. found an association between poor performance on certain combinations of physical fitness tests (e.g., poor results in the combination of push-up and Cooper test (12-min running test) (HR 2.1, 95% CI 1.1–4.2); poor results in the combination of back lift and Cooper test (HR 2.4, 95% CI 1.1–5.4); poor results in the combination of sit-up and push-up test (HR 2.2, 95% CI 1.1–4.5); and poor results in the combination of push-up and back lift test (HR 2.8, 95% CI 1.4–5.9)) but not individual physical fitness tests (e.g., push-up or Cooper test alone) . There was also conflicting evidence on the association between height and LBP, with one study reporting an association between shorter height and LBP (HR 1.98, 95% CI 1.19–3.29) and LBP limiting work ability (HR 4.48, 95% CI 2.01–9.97) , while another study found no association .
Consistent associations between sociodemographic risk factors and LBP
Being female was the only sociodemographic risk factor that consistently demonstrated an association with LBP [24, 25]. MacGregor et al. concluded that females (OR 1.94, 95% CI 1.61–2.34) were more likely to report LBP compared to males , and Knox et al. concluded that being female (IRR 1.45, 95% CI 1.39–1.52) was associated with LBP .
Associations between sociodemographic risk factors and LBP
An association with LBP was reported for ‘single’ marital status being less likely to experience LBP (IRR 0.87, 95% CI 0.84–0.91) compared to individuals reporting ‘married’ as their marital status . Additionally, lower education level (HR 1.6, 95% CI 1.1–2.3) was associated with LBP . As these were only reported in single studies, these sociodemographic risk factors may be further studied.
Non-associations between sociodemographic risk factors and LBP
Based on a single study by Taanila et al., the following factors were not found to be associated with LBP: father’s occupation, urbanisation level of place of residence, smoking habits, use of alcohol, frequency of drunkenness before military service, agreeing that soldiers need good physical fitness, amount of time spent on sweating exercises, participation in individual aerobic sports, belonging to a sports club, participation in competitive sports, and last degree in school sports . Additionally, Knox et al. reported that race was not associated with LBP .
Inconsistent associations between sociodemographic risk factors and LBP
There was conflicting evidence on the association between age and LBP, with Knox et al. reporting that younger age (less than 20 years) was associated with LBP (IRR 1.24, 95% CI 1.15–1.36) , while Ernat et al. reported that among infantrymen, the incidence of LBP increased with age (from IRR 0.61, 95% CI 0.59–0.63 in those under the age of 20 to IRR 0.91, 95% CI 0.86–0.97 in those over the age of 40) . Two studies reported no association between age and LBP [24, 26].
Consistent associations between occupational risk factors and LBP
Among occupational risk factors, lower rank consistently demonstrated an association with LBP, with one study demonstrating that junior rank was associated with a higher risk for incident LBP compared to those with senior rank (IRR 1.60, 95% CI 1.52–1.70) , while another study demonstrated that mid-level ranks (compared to junior ranks) were associated with a lower risk for LBP (OR 0.73, 95% CI 0.64–0.83) .
Associations between occupational risk factors and LBP
Several risk factors demonstrating an association with LBP were studied in single studies. These included having a blast injury (OR 2.29, 95% CI 1.64–3.19) , job duties (e.g., lifting > 30 pounds (OR 1.30, 95% CI 1.06–1.60) or wearing body armour (OR 1.14–1.30, 95% CI 1.07–1.53)) , and service type (e.g., Army (IRR 2.74, 95% CI 2.60–2.89) and Air Force (IRR 1.98, 95% CI 1.84–2.14) compared to Marines) . In a study of U.S. military service members, no association with LBP was found for location of deployment and time deployed .
Inconsistent associations between occupational risk factors and LBP
There were no military occupations that were consistently found to be associated with developing LBP, as positive associations were found for many different occupations [24, 26, 30]. MacGregor et al. concluded that being in the service/supply occupation (compared to administrative/other occupations) (OR 1.33, 95% CI 1.12–1.59) and being in the electrical/mechanical/craftsworker occupation (compared to administrative/other occupations) (OR 1.31, 95% CI 1.12–1.53) were risk factors for LBP . Taanila et al. concluded that being part of the engineer company (HR 2.0, 95% CI 1.2–3.3) was associated with LBP compared to those working in the anti-tank company . Ernat et al. concluded that infantrymen had a lower risk of LBP compared to non-infantry soldiers (IRR 0.69, 95% CI 0.68–0.70) . There was also no consistent evidence for the association of driving and incident LBP [25, 31]. Knox et al. concluded that being a military vehicle operator was associated with an increased risk of LBP compared to those of other occupations (IRR 1.15, 95% CI 1.13–1.17) , while Zack et al. concluded that professional truck drivers were less likely to experience LBP compared to those working in administrative units (RR 0.49, 95% CI 0.40–0.60) .
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.