This review summarised the results of studies of lumbo-pelvic kinematics for people with and without LBP. Although the results will be unsurprising to most clinicians, it is the first review to meta-analyse and quantify the clinical observation that, on average, people with LBP have reduced lumbar ROM, move more slowly and have reduced proprioception compared to with those without LBP.
The review highlights the highly heterogenous nature of available studies, with six of nine meta-analyses indicating significant between study heterogeneity in results. Possible sources of heterogeneity between study outcomes include differences in definitions of back pain, control characteristics, LBP intensity, and instruments and methods for measuring movements. This heterogeneity confounds secondary analyses such as the influence of pain intensity on observed differences between people with and without LBP.
The lack of detail or standardized definition for control subjects is also problematic. For example, it is hypothetically possible that altered movement characteristics occur as a result of a LBP episode and persist after pain resolves. If this is the case, people that were pain free but with persistent altered movements, would have been eligible as control subjects for many of the included studies, provided the episode had been prior to the pain-free time period required for that study. This would have diluted differences between the groups. Similarly, it is not known if certain ‘aberrant’ movement characteristics exist prior to the onset of LBP and are risk factors for an episode of LBP, in which case these characteristics may have also been present in people classified in the included studies as control subjects.
No studies attempted to blind assessors to group type, and a general absence of procedural standardization, such as movement instruction or assessor consistency, exposes studies to the potential for random or systematic error. However, the relative consistency of the direction of results across studies adds credibility to the findings of this review, and observed effects appear large enough to be visible despite potential study limitations.
Lordosis angle does not differentiate people with and without LBP. A similarly wide range of group means were reported for those with LBP (23° to 56°) and without LBP (19° to 53°). This variability might be associated with the six different measurement methods, but may also reflect biological differences in sample ethnicity , age  and gender [49, 57, 58]. Increasing age has been associated with reduced lordosis in the sixth decade [78–80] and on average, females have a greater lordosis than males [49, 58, 80]. Four studies included only males [31, 32, 38, 47] and it is perhaps understandable that these studies found the four lowest average lordosis angles. However, this variability in lordosis appears similar for people with and without LBP. Therefore, lumbar lordosis when measured using surface techniques, does not, on average, appear to discriminate between people with and without LBP.
Range and speed of motion
Clinicians commonly use ROM  to assist in identifying patterns of dysfunction, and to monitor change. ROM has been extensively studied by invasive and non-invasive methods, but non-invasive measurement is better suited to routine clinical assessment. This review included 20 studies that compared ROM for those with and without LBP using skin-surface measurement. The pooled sample was large enough to be confident in the finding that people with LBP have reduced average lumbar ROM compared to those without LBP. The mean ROM reported for people without LBP is so variable that it has little reference value e.g. (considering all studies) flexion: min = 23°, max = 92°; extension: min = 15°, max = 56°, lateral flexion: min = 3°, max = 44°; rotation: min = 3°, max = 62°. Large variations between studies suggest differences beyond those explained by biological variation and implicate method differences. Using flexion ROM as an example, 14 studies used nine different measurement devices ranging in sophistication from simple handheld inclinometers and flexible rulers to opto-electronic devices. Youdas [57, 58] used a flexible rule measurement technique (mean lumbar flexion angle = 23 ± 10°) while Hidalgo  used an opto-electronic system (92 ± 15°); both studies used similar inclusion criteria, and the same starting position. Other method processes may also contribute to differences: two studies assessed range in sitting, 10 in relaxed standing, and two used some form of restricted movement (harness or fixed pelvic position). Based on these findings, normative data may have limited relevance to a clinical environment unless the same measurement methods used to obtain published data are also used in the clinical setting where they are applied. The lack of clarity about similarity between study populations and method details makes the use of pooled group-level estimates of movements, such as mean flexion ROM, unwise. However, these between-study differences did not obscure consistent within-study findings; eight of 14 studies of flexion demonstrated significantly less lumbar flexion for those with LBP and only one study found that lumbar flexion was significantly greater for those with LBP. These findings of large between study differences in measurements, and consistent within study differences between those with and without LBP, are similar for the other movements analysed in this review.
Lower movement speed is commonly seen in people with LBP, so it is unsurprising to observe in our review that those with LBP demonstrated significantly slower speeds when the eight included studies were pooled in meta-analysis. Reduced speed of lumbar movement has been linked to fear of movement and has also been shown to persist after recovery .
Lumbar versus hip contribution to movement
Clinicians have reported assessing the relative contribution of lumbar and hip joints (during flexion and extension movements) to assist in determining subgroups within the LBP population that require specific treatment strategies [83, 84]. This review identified six studies that measured patterns and relative contributions to trunk flexion from the lumbar spine and hip joints, often described as ‘lumbo-pelvic rhythm’. Data could be pooled for four studies (six comparisons) evaluating ROM of lumbar and hip contribution at end-range flexion. A typical pattern of lumbar versus hip movement for both groups showed less lumbar and greater hip ROM at end-range flexion, with small, non-significant differences of reduced lumbar contribution for the LBP group when compared to people without LBP.
However relative contributions of lumbar spine and hip to ROM may be less important than patterns of when and how movement takes place. Nelson-Wong et al.  recently reported that the relative timing of hip and lumbar movement when arising from a fully flexed position differentiated between people who do or do not develop back pain after two hours of standing. People who developed pain used a lumbar > hip initiation of movement (spine moves first followed by pelvic/hip movement) strategy on arising from the flexed position while non-pain developers used a hip > lumbar strategy (p = 0.03). This finding is supported by McClure et al. , Esola et al.  and Porter et al.  who all reported relatively greater lumbar through-range contribution in people with LBP on flexion movement. It may be that people with LBP can be subgrouped by lumbo-pelvic rhythm. For example, Kim et al.  examined lumbo-pelvic rhythm by comparing two subgroups of people with LBP to a group of people without LBP. One subgroup had pain provoked by flexion/rotation activities and the other by extension/rotation activity. The flexion-aggravated group had significantly greater lumbar contribution to flexion compared to the normal and extension groups. The extension-aggravated group on the other hand had a significant pattern of reduced lumbar contribution to flexion. Lumbar versus hip contributions to movement, particularly flexion, appear to have clinical relevance and warrant further exploration.
Pelvic tilt angle, position and range
Extreme (end-range) pelvic tilt angle in standing and sitting has been linked to back pain [85, 86] but with limited evidence. Clinical interventions aiming to modify pelvic tilt angle to achieve more neutral positions are based on the assumption that there is a relationship between position and pain. There are few studies that explore the relationship between LBP and typical pelvic tilt range (from full anterior to full posterior tilt) and the relative position of pelvic tilt angle during sitting and standing in people with and without LBP. This review found no differences when pooling data from three studies that compared standing pelvic tilt angle in people with and without LBP. Similarly, Astfalk et al.  found no differences in average lumbar flexion angle in sitting (reflecting pelvic tilt position) when comparing adolescents with and without LBP (125.3 ± 19.8° vs 130.6° ± 15.7 respectively). However significant differences were observed for lumbar flexion angle when adolescents with LBP were sub-grouped based on direction of movement that provoked pain. The flexion-provoked pain group had a significantly greater lumbar angle (135.6 ± 16.9°, p < 0.05) compared to those without LBP while the extension-provoked pain group had a significantly smaller lumbar angle (113.5 ± 16.3°, p < 0.05) when compared to those without LBP. Sub-grouping of a LBP population based on the relationship of aggravating activities and direction of painful movement may demonstrate associations between back pain and pelvic tilt angle/relative position.
Our meta-analysis of studies measuring one aspect of proprioception (absolute error during re-positioning trials) demonstrated a significant and large loss of re-positioning accuracy in the LBP group. The implications of reduced proprioception are that people with LBP are less ‘movement-aware’ with potentially reduced postural control. This is consistent with a recent systematic review on another aspect of proprioception, postural sway, by Ruhe et al.  who found that greater sway excursion and speed were present in people with LBP compared to people without back pain.
Differences in variability between people with and without LBP
Our assessment of differences in variability between people with and without LBP for nine movement characteristics demonstrated significantly greater variability for four movement characteristics: flexion, lateral flexion and rotation ROM, and speed of movement. There were no significant differences in variability for lordosis, extension ROM, lumbar versus hip contribution to movement or proprioception. It is not clear if the greater variability seen in the LBP group is clinically meaningful (10% difference in average variability estimates) but it raises a question of whether postures or activities performed using extremes of certain movement (e.g. excessive or restricted movement) may predispose people to LBP.
This review examined differences in group means for people with and without LBP. Given the high variability seen between studies, the small between-group differences compared with the high within-group differences, and the greater variability on some movement characteristics seen in the LBP group, these findings cast some doubt on whether an assessment of movements without reference to pain provides evidence of dysfunction at an individual patient level. The results neither endorse nor disqualify the role of movement assessment for (i) determining the relationship between movement and pain in individual patients, or (ii) monitoring changes in movement characteristics as a means of monitoring progress in individual patients and as an indication of the likelihood of their improvement . Key questions also remain, including (a) are deficits such as reduced proprioception, reduced ROM and speed of movement a result or a cause of LBP, and (b) are these deficits present prior to the development of LBP?
Strengths and limitations
The strengths of this systematic review are the comprehensive search, the breadth of the movement characteristics included in the analysis, and that screening and data extraction were independently performed by two reviewers. In addition, the review only included studies that assessed people with and without LBP using the same within-study method, thereby removing method differences as an explanation for observed within-study differences.
The review also has limitations. We treated the data for people with LBP as if they were measurements of a homogenous group. It is possible that sub-grouping by using the relationship of pain to movement may increase the clinical utility of particular measurements. The findings in this review do not inform clinicians about whether changes in ROM, movement speed or proprioception will produce better outcomes, or if changes in movement characteristics precede the onset of LBP or predispose to future recurrences. In addition, due to an absence of translation resources, only articles published in English were included and this may introduce a language, cultural and/or publication bias. To maximize the number of included studies, we did not place any restrictions on the criteria used to define pain cases versus pain-free controls. However, our broad inclusion criteria are likely to have weakened, rather than strengthened differences seen between people with and without LBP, and in the included studies, higher pain intensities had a weak correlation with increased differences between the these groups.