What is the contribution of latissimus dorsi to trunk movement and control? A systematic review and meta-analysis

Abstract

Background

Latissimus dorsi may contribute to trunk movement and control because of its extensive attachments to the trunk. However, electromyography studies have shown highly variable activity levels during trunk tasks.

Objective

To critically evaluate whether latissimus dorsi has a role in trunk movement and/or control.

Methods

Studies assessing the activation of latissimus dorsi using electromyography during trunk movements and/or trunk stability tasks were sourced (May 2022). Risk of biases and quality of evidence was assessed. Activation levels were pooled and meta-analysed where possible.

Results

Thirty nine of 6125 studies identified in the search met the inclusion criteria. The meta-analyses showed high latissimus dorsi activity levels (60% maximal voluntary contraction [MVC]) during ipsilateral trunk rotation and low levels (<20% MVC) during contralateral trunk rotation, extension and stability tasks. Considerable variability of activity levels existed between studies when using high loads. Quality of evidence was very low to moderate.

Conclusion

Although high activity levels were found during ipsilateral trunk rotation, there is very low confidence that these activity levels reflect the true levels. There is moderate confidence latissimus dorsi has a limited contribution to trunk control. The use of surface electrodes and non-validated normalisation processes were critical methodological issues that contributed to lower quality of evidence.

Keywords:

• Electromyography
• latissimus dorsi
• trunk
• control
• movement

Introduction

Latissimus dorsi is a broad, flat, triangular muscle with extensive attachments arising from the spine, ribs and pelvis that converge onto the intertubercular groove of the humerus [1]. Because of these expansive attachments to the axial skeleton, latissimus dorsi has the potential to play a significant role in trunk movement and/or control. As a result, treatment aimed at improving latissimus dorsi function has been incorporated into rehabilitation programs for people with back pain [2].

Efficient trunk control and movement are required to successfully interact with our environment by allowing optimal production and transfer of force to the lower limbs for locomotion and to the upper limbs for manipulation [3]. This is achieved by the muscular system working in a coordinated manner to create a stable yet adaptable platform from which the limbs can work, with many muscles contributing to this system depending on the requirements of the task [4–8]. Trunk control will be compromised if there is a non-optimal recruitment pattern of the trunk muscles, which will consequently impact task performance and can lead to increased abnormal loads on spinal structures potentially resulting in pain [9,10].

It is believed that latissimus dorsi contributes to trunk control via its attachment through the thoracolumbar fascia (TLF) to the transverse and spinous processes of the spine [11,12]. In conjunction with other muscles that attach to the TLF (gluteus maximus, internal oblique, transverse abdominus) latissimus dorsi can tension the TLF to stiffen the spine and allow forces to be transmitted between the upper and lower limbs [11–14]. Exercises targeting this group of muscles that tension the TLF are often incorporated into rehabilitation programs aimed at improving trunk control in people with low back pain [2,15].

The contribution of latissimus dorsi to trunk movement and control has been investigated by measuring the activity levels of latissimus dorsi as a percentage of maximal voluntary contraction (% MVC) during various trunk tasks. However, electromyographic (EMG) studies have shown wide variations in latissimus dorsi activity levels when investigating similar trunk tasks at similar loads. For example, levels range from 26% to 116% MVC [16,17] during ipsilateral trunk rotation with high resistance, and from 15% to 85% MVC [18,19] during ipsilateral trunk lateral flexion with high resistance. With such wide variations in activity levels, it is difficult to draw conclusions in regards to the contribution of this important muscle to trunk movement and control. Therefore, the aim of this systematic review was to synthesise and critically evaluate all EMG evidence investigating latissimus dorsi activity levels during trunk tasks to determine its contribution to trunk movement and control.

Methods

Protocol

This systematic review was conducted and reported according to the PRISMA statement [20]. The protocol was registered on the international prospective register of systematic reviews (PROSPERO), ID number: CRD42018089210 (February, 2018).

Eligibility criteria

All studies assessing the activation of latissimus dorsi using EMG during trunk movements and/or lower limb movements were included in this systematic review. All study designs were included if they reported on adults over 18 years of age with no current shoulder or back pain, were published in English and reported a muscle activation level as % MVC during cardinal plane trunk movements and exercises traditionally used for trunk control. Studies were excluded if they were conference abstracts, used unstable surfaces or investigated latissimus dorsi activity involving upper limb movements as it would be difficult to differentiate between the latissimus dorsi effect on the shoulder or trunk.

Search strategy

A search of all relevant literature prior to May 2022 was completed using the following databases: Medline, EMBASE, PubMed, CINAHL, SCOPUS and SPORTDiscus. In consultation with a librarian, the following search strategy was developed: (EMG OR electromyography) AND (‘latissimus dorsi’ OR ‘superficial back muscles’ OR trunk OR torso OR back OR pelvis OR thorax) AND (stability OR movement). Screening of the reference lists was also undertaken on all relevant articles.

Study selection

The studies were retrieved from the databases, collated in Endnote software (EndNote X9) and duplicates removed. Two authors (DP and DR) independently reviewed the titles and abstracts of all sourced articles. Relevant studies identified from this process were saved for further review of the full text. Full texts were sourced and independently reviewed by pairs of authors. Disagreements were discussed within the pairs and if an agreement could not be reached a third author was consulted until a consensus was formed. A list of relevant articles was finalised for data extraction.

Data extraction and synthesis

Descriptive data from relevant studies were extracted by DP and independently checked by DR with any discrepancies discussed by all authors. Data included mean age, sex, electrode type, electrode location, normalisation process and tasks.

The primary outcome measure of this study was the activation level of latissimus dorsi expressed as % MVC. The activation levels were extracted by DP, independently checked by DR and grouped based on the tasks performed and load employed. Load was divided into seven categories based on either the percentage of maximum voluntary exertion (MVE) (minimal = 0–9% MVE, low = 10–29% MVE, moderate = 30–59% MVE, high = 60–89% MVE and maximum = 90–100% MVE), use of body weight (unresisted) or other load (speed, N.m, kg). Where activation levels were presented in graphs a customised script in Matlab (R2019, the Mathworks, Inc., Natick, MA) was used to estimate the values.

Methodological quality assessment

Methodological quality of the studies was assessed using 11 items from the Downs and Black quality assessment tool [21]. The items retained from this tool were considered relevant for laboratory EMG studies and gave a total score out of 11. These included: clear hypothesis (Item 1), outcome measures clearly described (Item 2), participant characteristics clearly described (Item 3), experimental conditions clearly described (Item 4), main findings clearly described (Item 6), estimates of random variability provided (Item 7), lost data reported (Item 9), actual p values reported (Item 10), participants represent entire population (Item 12), appropriate statistical tests used (Item 18) and valid and reliable outcome measures used (Item 20). The descriptive explanation for item 9 was defined as the reporting of lost data from an EMG channel during the experiment, item 12 was defined as asymptomatic subjects but not elite or trained individuals and for item 20 the use of surface electrodes to record latissimus dorsi activity was considered invalid due to potential crosstalk from adjacent muscles [22]. The description of all other items was as per the original form. To standardise the methodological quality extraction process and clarify the meaning and interpretation of critical appraisal items, data were extracted from two studies independently by all authors and discussed. Two authors (DP and DR) then independently extracted data from the remaining studies and any discrepancies were resolved with a third author.

Overall quality of the evidence used in each individual meta-analysis was assessed using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) rating approach. Quality was downgraded for the following reasons [23,24]:

• High risk of bias: >60% of studies scoring <9/11 on the modified Downs and Black or the use of surface electrodes to record latissimus dorsi activity, which was considered to be a crucial limitation sufficient to lower confidence in the outcome [25].

• High publication bias: Visual inspection of funnel plot showing asymmetrical spread.

• Imprecision: Large 95% confidence intervals.

• Inconsistency: Heterogeneity between studies was large (I² > 50%).

• Indirectness was not considered as only studies with similar target populations, methodological conditions and outcome measures were included.

Statistical analysis

Separate meta-analyses to pool the means and standard deviations (SDs) across studies were performed using the Comprehensive Meta-Analysis software (CMA version 3, Biostat, Englewood, NJ) for each task where there were three or more homogenous studies reporting % MVC values for latissimus dorsi with a SD. A random effects model was used when heterogeneity among studies was significant as indicated by a significant Q-value. The results indicate that all Q values were significant; therefore, a random effects model was used for all meta-analyses. Studies were grouped together if they used the same load, same electrode type, similar normalisation processes and if they provided a measure of variability (standard error or SD). Studies that reported more than one outcome value for latissimus dorsi activation levels during each task had their values pooled together using the meta-analysis software to give one value per study per load range. If the values within the same study were unable to be pooled (no variability measure) then the values were averaged but not included in the meta-analyses.

Results

The results of the literature search are presented in Figure 1. The combined searches yielded a total of 6125 studies following the removal of duplicates. After reviewing the titles and abstracts and manually scanning the reference lists, 79 studies were selected for full text review. Thirty-nine studies met the criteria for inclusion in this systematic review.

Figure 1. PRISMA flowchart literature search.

A flowchart of the literature search showing an initial 14 508 studies identified. Removal of duplicates, screening of the titles, abstracts and full text reviews reduced the eligible studies to 39 studies to be included in the review.

Study characteristics

Characteristics of the included studies are summarised in Table 1. The overall age of the combined cohort was young (<30 years old) and mostly male (66%). The tasks and the number of studies investigating each task (note that some studies reported on multiple tasks) were as follows: ipsilateral trunk rotation (16 studies), contralateral trunk rotation (13 studies), trunk extension (18 studies), trunk flexion (14 studies), ipsilateral trunk lateral flexion (11 studies), contralateral trunk lateral flexion (8 studies), combined trunk movements (2 studies) and limb movements in four-point kneeling (4 studies).

Table 1. Study characteristics.

Author	Cohort size and gender	Mean age or range in years (±SD)	Electrode type	Electrode placement	EMG normalisation - MVC - Test position	Trunk task (position)	Movement type	Load/speed	Downs and Black Methodological quality (/11)
[26]	9 M	23.9 (2.8)	Surface	Muscle belly	MVC Trunk flexion, lateral flexion and rotation in modified sit-up position; reverse curl-up in supine	1- Flexion (side-lying) 2- Extension (side-lying) 3- Lateral flexion (supine)	Dynamic	Minimal	9
[27]	13 M	21.0 (1.0)	Surface	Lateral to T9	MVC Lateral pull-down	Four point kneel: 1- Ipsilateral leg extension 2- Contralateral leg extension 3- Ipsilateral leg extension with contralateral arm flexion 4- Contralateral leg extension with ipsilateral arm flexion	Dynamic	Body weight	9
[14]	14 (8 M, 6 F)	24.7 (3.2)	Surface	3 cm lateral and inferior to inferior angle of scapula	MVC Shoulder extension	Extension (prone)	Isometric and dynamic	Body weight	9
[15]	8 M	23.4 (1.8)	Surface	Lateral to T9	MVC Lateral pull down	1- Extension (prone) Four point kneel: 2- Ipsilateral leg extension 3- Contralateral leg extension 4 - Contralateral leg extension with ipsilateral arm flexion	Dynamic	Body weight	9
[28]	18 (9 M, 9 F)	27.7 (7.7)	Surface	4 cm below inferior angle of scapula and midway between the spine and lateral edge of the torso.	MVC Shoulder extension	Flexion (supine)	Dynamic	Body weight	8
[29]	11 M	28.2 (4.4)	Surface	Lateral to T9	MVC Trunk flexion, extension, lateral flexion and rotation	1- Extension (standing) 2- Flexion (standing)	Dynamic	22 N, 67 N, 156 N @ 2s, 1.5s, 1s	8
[30]	21 M	35.9 (6.6)	Surface	Not reported	MVC Trunk extension	Extension (kneeling and standing)	Isometric and dynamic	Maximal @ 30°/s, 60°/s, 90°/s	9
[22]	8 (5 M, 3 F)	19–49	Surface and indwelling	4 cm below inferior angle of scapula	MVC Shoulder internal rotation at 90° abduction, shoulder extension at 30° abduction	Extension (Prone)	Isometric	Maximal	10
[31]	12 M	26.1	Surface	Not reported	MVC Trunk flexion, extension, lateral flexion and rotation	Extension (standing)	Dynamic	13.6 kg, 27.3 kg @ preferred speed and faster	8
[32]	10 M	21 (3)	Surface	Not reported	MVC Trunk extension	Four-point kneeling: 1- Contralateral arm flexion and ipsilateral leg extension 2- Ipsilateral leg extension 3- Contralateral leg extension	Dynamic	Body weight	8
[33]	50 (27 M, 23 F)	M = 22.1 (3.8), F = 22.1 (4.3)	Surface	Superior lateral aspect where the muscle bands together	MVC Trunk rotation	Rotation (sitting)	Isometric Peak	Low, moderate, high, maximal	8
[34]	19 (7 M, 12 F)	M = 23 (4), F = 21 (3)	Surface	15 cm lateral to T9	MVC Trunk extension and rotation in flexed position	Combined rotation and extension (standing in flexed and rotated position)	Isometric Peak	Maximal	9
[35]	19 (7 M, 12 F)	M = 23 (4), F = 21 (3)	Surface	Not reported	MVC Trunk flexion from flexed and rotated trunk (task)	Combined rotation and flexion (standing in flexed and rotated position)	Isometric	Maximal	8
[36]	10 M	20–36	Surface	Muscle belly at T7, 13–15 cm lateral from midline	MVC Trunk flexion, extension and rotation	Flexion, extension and lateral flexion.	Isometric	10, 20, 30, 40 and 50 Nm	8
[37]	10 M	24–36	Surface	T7 over muscle belly	MVC Trunk flexion, extension and rotation in rotated position	Flexion, extension, lateral flexion (standing in rotated position)	Isometric	20 Nm, 40 Nm	5
[38]	19 M	19–41	Surface	T7 over muscle belly	MVC Normalised to moment direction and magnitude. Trunk flexion, extension and rotation in flexed position	Flexion, extension, lateral flexion (standing in flexed position)	Isometric	20 Nm, 40 Nm	5
[39]	15 (13 M, 2 F)	20–38	Surface	T7 over muscle belly	MVC Trunk flexion, extension, rotation while in laterally flexed position	Flexion, extension, lateral flexion (standing in laterally flexed position)	Isometric	20 Nm, 40 Nm	6
[40]	12 M	24–33	Surface	Not reported	MVC Movement not specified	Lateral flexion (standing)	Isometric and dynamic	13.6 kg, 27.3 kg @ 15, 30, 45°/s.	7
[41]	12 M	21–31	Surface	Lateral to T9	MVC as a function of velocity and trunk posture. Trunk rotation	Rotation (standing)	Isometric	40Nm	9
[42]	15 (10 M, 5 F)	M = 21.2 (8.5), F = 24 (1.9)	Surface	Lateral to T9	MVC Bent knee sit up, trunk flexion, lateral flexion and rotation	Rotation (standing)	Isometric	Maximal	8
[43]	11 M	22 (3.2)	Surface	Lateral to T9	MVC Trunk flexion, extension, lateral flexion and rotation	Rotation (standing)	Isometric	Maximal	8
[44]	12 (5 M, 7 F)	M = 68.8 (5); F = 69 (3.5)	Surface	Lateral to T9	MVC Lateral pull down	1- Lateral flexion (standing) 2- Rotation (standing) 3- Flexion (standing)	Isometric	Body weight	9
[9]	23 M	30.2 (7.9)	Surface	T12 level along a line connecting superior point of the axillary fold and S2	MVC Trunk flexion, extension, lateral flexion and rotation	Rotation (standing)	Isometric	Moderate, high, maximal	8
[45]	24 M (controls only = 12)	30 (7.6)	Surface	T12 level along a line connecting superior point of the axillary fold and S2	MVC Trunk flexion, extension, lateral flexion and rotation	Rotation (standing)	Isometric	Moderate, high, maximal	9
[16]	23 M	30.2 (7.9)	Surface	T12 level along a line connecting superior point of the axillary fold and S2	MVC Trunk flexion, extension, lateral flexion and rotation	Rotation (standing)	Isometric	Maximal	8
[46]	22 (11 M, 11 F)	F = 23.3 (2.2) M = 24.3 (2.8)	Surface	Lateral to T9	MVC Trunk rotation	Rotation (standing)	Isometric	Maximal	8
[27]	16 M	21–24	Surface	4 cm below inferior tip of scapula and midway between the spine and lateral edge of the torso.	MVC Peak level of: extension in prone, caudal depression, body lift, right and left upper trunk lateral flexion, lateral pull down	Lateral flexion (side-lying)	Isometric	Maximal	8
[48]	18 M	21.94 (2.24)	Surface	Lateral to T9	MVC Shoulder extension	Extension (Roman chair)	Dynamic	Body weight	8
[49]	16 M	23 (1.92)	Surface	Medial LD: lateral to T9 over the muscle belly. Lateral LD: 4 cm below inferior angle of scapula.	MVC Peak of either shoulder extension with adduction in prone or isometric lateral pull down.	1- Extension (prone) 2- Lateral flexion (side-lying)	Isometric	Body weight	9
[50]	24 (17 M, 7 F)	M = 22 (3.4), F = 20.4 (.5)	Surface	Lateral to T9	MVC Bent knee sit up, trunk extension, isometric exertions similar to body builder poses	1- Rotation (standing) 2- Extension (standing) 3- Flexion (standing) 4- Lateral flexion (standing)	Dynamic	Body weight	8
[19]	16 (8 M, 8 F)	M = 27 (6.93), F = 25 (2.51)	Surface	Lateral to T9	MVC Trunk extension, flexion, lateral flexion, rotation, and one armed brachiation	1- Rotation (standing) 2- Extension (standing) 3- Flexion (standing) 4- Lateral flexion (standing)	Isometric	Moderate	9
[51]	18 (13 M, 5 F)	M = 29.8 (4), F = 27.2 (8)	Surface	T12 -L1 level	MVC Trunk movement	1- Rotation (standing) 2- Extension (standing) 3- Flexion (standing)	Isometric	Moderate, high, maximal	8
[52]	30 (15 M, 15 F)	M = 25.0 (3.8), F = 22.8 (2.7)	Surface	Lateral to T9	MVC Isometric pull down with shoulder abducted to 90 degrees, with the arm externally rotated and the elbow flexed to 90 degrees	1- Rotation (standing) 2- Flexion (standing) 3- Lateral flexion (standing)	Dynamic	Body weight	8
[2]	30 (Gender not reported)	20.52 (1.74)	Surface	3 cm lateral and inferior to inferior angle of scapula	MVC Position not reported	Rotation (sitting)	Dynamic	Moderate, high	7
[17]	30 (15 M, 15 F)	19.6	Surface	3 cm lateral and inferior to inferior angle of scapula	MVC Shoulder extension	4 point kneel: 1- Ipsilateral leg extension 2- Contralateral leg extension 3- Ipsilateral leg extension with contralateral arm flexion 4- Contralateral leg extension with ipsilateral arm flexion	Dynamic	Body weight	9
[53]	12 M	25.9 (3.3)	Surface	T12 level along a line connecting the most superior point of the axillary fold and S2	MVC Trunk flexion, extension, lateral flexion, rotation and 4 biaxial exertions.	1- Extension (standing) 2- Flexion (standing)	Isometric Peak	Moderate, high, maximal	9
[54]	9 M	24 (3)	Surface	Not reported	Rectified and averaged	Extension (semi-seated position)	Isometric	Low, moderate	9
[55]	13 (4 M, 9 F)	22.6 (2.1)	Surface	3 cm lateral and inferior to inferior angle of scapula	MVC Shoulder extension	Extension (prone)	Dynamic	Body weight	9
[18]	8 F	26.0 (5.8)	Surface	Lateral to T9	MVC Upper and lower trunk flexion, extension and rotation; shoulder internal rotation and adduction; maximal effort abdominal hollowing and bracing. (Task)	1- Rotation (crook lying) 2- Extension (prone) 3- Flexion (crook lying) 4- Lateral flexion (side-lying)	Isometric	Maximal	7

F: female; LD: latissimus dorsi; M: male; MVC: maximal voluntary contraction; MVE: maximal voluntary exertion; SP: spinous process

Loads expressed as absolute loads or as a % of MVE: minimal = 0–9% MVE, low = 10–29% MVE, moderate = 30–59% MVE; high = 60–89% MVE; maximal = 90–100% MVE.

Surface electrodes were used in all but one included study. The placement of electrodes varied slightly with 21 studies (54%) placing the electrodes lateral to T9, about 4 cm below inferior angle of the scapula, five studies (13%) placing them lateral to T12, four studies (10%) placing them lateral to T7, one study (3%) placing them 15 cm lateral to T9 and eight studies (21%) not mentioning a specific location.

Out of the 39 studies, 38 (97%) used a MVC to normalise the EMG data. Twelve studies (31%) used shoulder movements (extension and/or lateral pull down) and 21 studies (54%) used a combination of trunk movements (flexion, extension, lateral flexion and/or rotation) as the MVC task. It was unclear in five studies (13%) how normalisation was completed. Direct comparison could only be made between latissimus dorsi activity normalised to a trunk or a shoulder task during trunk extension and ipsilateral trunk lateral flexion [18,19,22]. Large variations in latissimus dorsi activity were reported, ranging from 10% to 79% MVC during trunk extension and 15–85% MVC during ipsilateral trunk lateral flexion, dependant on the chosen normalisation task (Figure 2).

Figure 2. Comparison of studies normalised to trunk vs. shoulder tasks. Mean and (SD). Solid black columns = EMG data normalised to trunk movements, white columns = EMG data normalised to shoulder movements, diagonal pattern = normalised to shoulder movements and using indwelling electrodes. MVC = maximum voluntary contraction.

A column graph showing the high variability of latissimus dorsi activity when normalised to the trunk versus the shoulder using maximal levels during trunk extension and ipsilateral flexion.

Methodological quality

Most studies (n = 33, 85%) included in this review scored highly (≥8/11) using the modified Downs and Black checklist (Appendix 1). All studies scored well with the description of their hypothesis (Item 1) and outcome measures (Item 2), whereas all studies failed to report any lost data or that there was no lost data (Item 9). All but one study scored poorly for the use of valid outcome measures (Item 20) based on the use of surface electrodes.

Meta-analyses

Fourteen of the thirty nine studies (36%) were included in the meta-analyses representing seven trunk tasks (Figure 3) with only three of the meta-analyses including more than three studies. The results of the meta-analyses showed high latissimus dorsi activation levels during ipsilateral trunk rotation and low levels during contralateral trunk rotation, prone trunk extension and the trunk control tasks performed in four-point kneeling.

Figure 3. Forest plot of meta-analyses. 4 pt: Four-point kneeling; contra: contralateral; ipsi: ipsilateral.

A forest plot of the meta-analyses showing high activity during ipsilateral rotation and low levels during trunk extension and four point kneeling tasks.

GRADE

Overall quality of evidence for ipsilateral rotation was ‘very low’ due to the use of surface electrodes, high risk of publication bias, imprecision due to large confidence intervals and inconsistency (I² > 50%) (Table 2). Consequently, there is limited confidence in the latissimus dorsi activation estimate during ipsilateral trunk rotation as the true activation level may be substantially different. Overall quality of evidence for contralateral trunk rotation, prone trunk extension and each control (four-point kneeling) task was downgraded to ‘moderate’ due to high risk of methodological bias due to the use of surface electrodes and inconsistency (I² > 50%). There is moderate confidence that the true level of latissimus dorsi activation could be close to or substantially different to this estimate.

Table 2. Heterogeneity and tau-squared data for meta-analyses.

	Heterogeneity				Tau-squared
	Q-value	df (Q)	p Value	I-squared	Tau-squared	Standard error	Variance	Tau
Ipsilateral rotation	457.57	7	0	98.47	842.95	720.26	518776.55	29.03
Contralateral rotation	54.23	6	6.63E-10	88.94	41.1	38.26	1463.5	6.41
Prone extension	124.49	4	0	96.79	65.47	72.51	5257.42	8.09
4 pt kneeling – ipsi leg	8.76	2	0.01	77.18	12.05	15.97	254.89	3.47
4 pt kneeling – contra leg	16.71	2	2.35E-04	88.03	29.68	37.97	1441.61	5.45
4 pt kneeling – ipsi leg, contra arm	13.89	2	9.64E-04	85.6	34.65	43.65	1904.88	5.87
4 pt kneeling – contra leg, ipsi arm	11.67	2	2.93E-03	82.86	58.7	75.73	5735.19	7.66

4 pt: four-point kneeling; contra: contralateral; ipsi: ipsilateral

Activation levels

Latissimus dorsi activation levels stratified under task and load, and summarised as a pooled range of % MVC are shown in Appendices 2 and 3 with levels at maximum load during all trunk movements also illustrated in Figure 4. Single studies not included in the meta-analyses but investigating similar tasks, reported: very high levels of latissimus dorsi activity (73–116% MVC) during ipsilateral trunk rotation at high to maximal load [17,51], low activity (10% MVC) during contralateral trunk rotation at high to maximal load [51], and low activity (11–22% MVC) during limb movements in the four-point kneeling position [8]. For other trunk movements for which not enough studies were available to complete a meta-analysis, single studies investigating trunk movements at maximal load reported high to very high activation levels during ipsilateral lateral flexion (85 ± 25% MVC) [18], extension (79 ± 20% MVC) [18] and flexion (41 ± 14% MVC) (Figure 4) [18].

Figure 4. Latissimus dorsi activity across tasks at maximum load. Solid black bars represent surface electrodes being used to record latissimus dorsi activity and the diagonal pattern represents indwelling electrodes being used to record latissimus dorsi activity. Contra: contralateral; Ipsi: ipsilateral; LF: lateral flexion; MVC: maximal voluntary contraction.

A column graph showing latissimus dorsi activity from all included studies during trunk rotation, lateral flexion, flexion and extension. High variability is shown between individual studies, with both high and low activity levels reported during ipsilateral trunk rotation, trunk extension and ipsilateral lateral flexion.

Discussion

There is considerable variability in the contribution of latissimus dorsi to trunk movements across previous EMG studies. Therefore, one of the aims of this study was to conduct a meta-analysis to synthesise the evidence in order to clarify the contribution of latissimus dorsi to trunk movement. Only three trunk movements (ipsilateral and contralateral rotation, and extension) could be included in the meta-analyses with ipsilateral rotation being the only movement where mean latissimus dorsi activity was moderate to high. Although this result was supported by two studies not able to be included in the meta-analysis [17,51], the overall quality of evidence, as assessed using the GRADE rating, was ‘very low’. Therefore, the result of moderate to high latissimus dorsi activity during ipsilateral trunk rotation needs to be interpreted with caution as there is limited confidence that this activity is representative of its true levels.

The meta-analyses for contralateral trunk rotation and trunk extension resulted in low mean latissimus dorsi activity levels. A single study not able to be included in the meta-analysis showed similarly low levels during contralateral trunk rotation [51]. As the quality of studies assessing both these trunk movements obtained a ‘moderate’ GRADE rating, there is some confidence the reported latissimus dorsi activity levels are representative of their true value. These results indicate that the actions of latissimus dorsi are unlikely to include contralateral trunk rotation and trunk extension.

The findings from the meta-analyses of studies investigating the role of latissimus dorsi during trunk control tasks support previous understanding that it is active at low levels. With a ‘moderate’ GRADE rating there is some confidence the reported activity levels may be close to their true values. Although this low level of activity suggests latissimus dorsi has a limited contribution to trunk control, this still may be an important contribution as part of the normal trunk muscle recruitment required to achieve optimal trunk control.

This systematic review revealed considerable variability in latissimus dorsi activity between studies during trunk movements. This was most evident when latissimus dorsi was working at high to maximum loads (Figure 4). For example, mean latissimus dorsi activity levels during ipsilateral rotation ranged from 26% to 116% MVC, during extension from 10% to 79% MVC and during ipsilateral lateral flexion from 15% to 85% MVC. Although these individual studies were of similarly high methodological quality (≥7/11 on the modified Downs and Black quality assessment tool) they differed with respect to EMG methodology. These methodological differences could account for the large variability in latissimus dorsi activity in previous studies.

All but one of the included studies in this review recorded latissimus dorsi activity using surface electrodes. The only study that has directly compared latissimus dorsi activity during trunk movements, using surface and indwelling electrodes [22], concluded that surface electrode recording greatly overestimated latissimus dorsi activity due to crosstalk from the underlying erector spinae (Figures 2 and 4). As erector spinae is active during many trunk movements the vast majority of studies in this review using surface electrodes to record latissimus dorsi activity are likely to be overestimating its contribution to trunk movement.

The majority of studies in this review (54%) used maximally resisted trunk movements to obtain maximal latissimus dorsi activation for EMG normalisation. Two studies [18,47] using surface electrodes have compared latissimus dorsi activity levels during maximally resisted trunk tasks to maximally resisted shoulder tasks, which have been shown to generate high levels of activity in latissimus dorsi [56]. Similar high latissimus dorsi activity levels were reported during maximal trunk extension and rotation when compared to shoulder adduction [18]. However, for maximally resisted trunk lateral flexion, results are conflicting with one study reporting similarly high levels of latissimus dorsi activity when compared to shoulder adduction [18] and another study reporting statistically significantly lower levels [47]. The use of surface electrodes in these studies may have resulted in an overestimation of latissimus dorsi activity due to potential cross talk from adjacent muscles. Support for such overestimation can be found in the only study that has compared latissimus dorsi activity during a maximally resisted trunk task to a shoulder task using indwelling electrodes [22]. In this study, significantly lower activity was reported during trunk extension compared to shoulder extension. If trunk tasks used to normalise the EMG data do not result in maximal latissimus dorsi activity, the reported % MVC levels will be higher than true % MVC levels, leading to an overestimation of the potential contribution of latissimus dorsi to trunk movement and control.

Limitations

Literature searches were conducted in six major databases: Medline, EMBASE, PubMed, CINAHL, SCOPUS and SPORTDiscus; however, there is a possibility relevant studies in journals not indexed in these databases may have been missed. In addition, no grey literature was included and only studies published in English were included precluding any potential studies in different languages. Finally, the utilisation of meta regression models to explore between-study heterogeneity was not possible due to the small number of studies evaluating each task.

Conclusion

The results of this systematic review provide little evidence that trunk movements will result in significant strength increases in latissimus dorsi. Even though relatively high latissimus dorsi levels were found during ipsilateral trunk rotation there is very low confidence this represents the true estimate of its activity level. There is moderate confidence in the finding that latissimus dorsi is only activated at low levels during trunk extension and contralateral trunk rotation. The results also indicate, with moderate confidence, that latissimus dorsi has a limited contribution to trunk control. This suggests that any therapeutic benefit from trunk control exercises in the treatment of people with low back pain is potentially due to their impact on other trunk muscles.

However, the critical EMG methodological issues identified in this review, that is the use of surface electrodes and non-validated EMG normalisation processes, challenges the validity and reliability of the reported latissimus dorsi activity levels. Further research using indwelling electrodes and a validated EMG normalisation procedure is needed to confirm the contribution of latissimus dorsi to trunk movement and control.

PROSPERO registration number

CRD42018089210

Acknowledgements

Elaine Tam (librarian) for assistance with search strategy.

Disclosure statement

The authors declare that there is no conflict of interest.

Funding

This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.

Correction Statement

This article has been corrected with minor changes. These changes do not impact the academic content of the article.

Additional information

Notes on contributors

Declan Price

Declan Price is a musculoskeletal physiotherapist and teaches functional musculoskeletal anatomy in the School of Medical Sciences, Faculty of Medicine and Health, The University of Sydney, Australia. He is currently enrolled in a Doctorate in Philosophy using electromyography to investigate muscle activation patterns in normal subjects during rehabilitation exercises and functional movements.

Karen A. Ginn

Karen A. Ginn is a Research Affiliate in the School of Health Sciences in the Faculty of Medicine and Health at the University of Sydney and is a musculoskeletal physiotherapist in part time private practice. For many years she taught functional, applied anatomy to various health professional groups at undergraduate and postgraduate level and currently conducts professional development courses related to the assessment and treatment of shoulder dysfunction nationally and internationally. Her main research interests relate to the assessment and treatment of shoulder dysfunction including EMG studies, clinical trials, shoulder stiffness and cortical changes associated with shoulder pain & prevention of shoulder pain in the elderly and at-risk professional groups. She has more than 70 publications in peer-reviewed journals and is currently a member of the Board of the International Confederation of Societies for Shoulder and Elbow Therapy.

Mark Halaki

Mark Halaki is a Professor in the Sydney School of Health Sciences, Faculty of Medicine and Health, The University of Sydney. His research interests are in movement coordination and muscle function in healthy people and people with injury or pathology. With injury and pathology, movement control is disturbed, resulting in pain or inability to move. By understanding differences in movement and muscle coordination between healthy people and people with pathology, our ability to treat patients with pathology is enhanced. His main research area is studying these differences with the ultimate goal of identifying methods, validated using large randomised control studies, of preventing and treating injuries effectively. He co-leads the Shoulder Laboratory and is an Associate Director of Sound Practice: Australian Musicians Health in Performance at The University of Sydney.

Darren Reed

Darren Reed is a musculoskeletal physiotherapist and senior lecturer in functional musculoskeletal anatomy in the School of Medical Sciences, Faculty of Medicine and Health, The University of Sydney. His main research interests are in the investigation of shoulder muscle function using electromyography in common rehabilitation exercises and functional movements. He is also interested in the development, translation and testing of musculoskeletal and pain related patient reported outcome measures and their use in clinical practice.

Appendix 1.

Methodological quality of included studies. Criteria items are numbered based on items adapted from the Downs and Black checklist [21]

 

Table

	1. Clear hypothesis	2. Outcome measures clearly described	3. Participant characteristics clearly described	4. Experimental conditions clearly described	6. Main findings clearly described	7. Estimates of random variability provided	9. Lost data reported	10. Actual p values reported	12. Participants represent entire population	18. Appropriate statistical tests used	20. Valid and reliable outcome measures	Score /11
[26]	1	1	1	1	1	1	0	1	1	1	0	9
[27]	1	1	1	1	1	1	0	1	1	1	0	9
[14]	1	1	1	1	1	1	0	1	1	1	0	9
[15]	1	1	1	1	1	1	0	1	1	1	0	9
[28]	1	1	1	1	1	1	0	0	1	1	0	8
[29]	1	1	1	1	1	1	0	0	1	1	0	8
[30]	1	1	1	1	1	1	0	1	1	1	0	9
[22]	1	1	1	1	1	1	0	1	1	1	1	10
[31]	1	1	1	1	1	1	0	0	1	1	0	8
[32]	1	1	1	1	1	1	0	0	1	1	0	8
[33]	1	1	1	1	1	0	0	1	1	1	0	8
[34]	1	1	1	1	1	1	0	1	1	1	0	9
[35]	1	1	1	1	1	1	0	1	1	0	0	8
[36]	1	1	1	1	1	0	0	1	1	1	0	8
[37]	1	1	0	1	1	0	0	0	0	1	0	5
[38]	1	1	0	1	1	0	0	0	0	1	0	5
[39]	1	1	0	1	1	0	0	1	0	1	0	6
[40]	1	1	0	1	1	1	0	0	1	1	0	7
[41]	1	1	1	1	1	1	0	1	1	1	0	9
[42]	1	1	1	1	1	1	0	1	1	0	0	8
[43]	1	1	1	1	1	1	0	0	1	1	0	8
[44]	1	1	1	1	1	1	0	1	1	1	0	9
[9]	1	1	1	1	1	1	0	0	1	1	0	8
[45]	1	1	1	1	1	1	0	1	1	1	0	9
[16]	1	1	1	1	1	1	0	0	1	1	0	8
[46]	1	1	1	1	1	1	0	0	1	1	0	8
[47]	1	1	1	1	1	1	0	0	1	1	0	8
[48]	1	1	1	1	1	1	0	0	1	1	0	8
[49]	1	1	1	1	1	1	0	1	1	1	0	9
[50]	1	1	1	1	1	1	0	0	1	1	0	8
[19]	1	1	1	1	1	1	0	1	1	1	0	9
[51]	1	1	1	1	1	0	0	1	1	1	0	8
[52]	1	1	1	1	1	1	0	0	1	1	0	8
[2]	1	1	1	1	1	0	0	0	1	1	0	7
[17]	1	1	1	1	1	1	0	1	1	1	0	9
[53]	1	1	1	1	1	0	0	1	1	1	0	8
[54]	1	1	1	1	1	1	0	1	1	1	0	9
[55]	1	1	1	1	1	1	0	1	1	1	0	9
[18]	1	1	1	1	1	1	0	0	0	1	0	7

Appendix 2.

Mean latissimus dorsi activation % MVC (±SD) during cardinal plane trunk movements of individual studies and pooled results

 

Table

Study	Body weight	Minimal load (0–9% MVE)	Low load (10–29% MVE)	Moderate load (30–59% MVE)	High load (60–89% MVE)	Maximal load (90–100% MVE)	Nm/kg (Load)	Speed
Ipsilateral trunk rotation (ipsilateral latissimus dorsi to movement direction)
[33]			22 (peak)	42 (peak)	73 (peak)
[41]							19 (34) (40 Nm)
[42]						69 (13)
[43]						48 (17)
[44]		34 (7)
[9]				21 (16)	51 (21)	82 (21)
[45]				17 (14)	45 (19)	80 (24)
[16]						56 (23)
[46]						26 (4)
[50]	22 (13)
[19]				27 (19)		42 (27)
[51]				10	30	73
[52]	15 (10)
[2]				88	116
[53]				88 (peak)
[18]						78 (22)
Range	15–22	34	22	10–88	30–116	26–82	19	–
Contralateral trunk rotation (contralateral latissimus dorsi to movement direction)
[33]			25 (peak)	46 (peak)	68 (peak)
[42]						17 (11)
[43]						27 (12)
[9]				1 (6)	4 (9)	6 (12)
[45]				4 (8)	7 (12)	10 (16)
[16]						6 (14)
[46]						7 (1)
[50]	5 (5)
[19]				16 (15)		20 (17)
[51]				2	6	10
[52]	8 (3)
[2]				22		27
[53]				18 (peak)
Range	5–8	–	25	1–46	4–68	6–27	–	–
Trunk extension (unilateral left or right latissimus dorsi)
[26]		2 (3)	4(6)
[27]	6 (4)
[14]	19 (2)
[15]	26 (11)
[22]						24 (10), 45 (12)
[31]							13 (17) (13.6 kg) 14 (19) (27.3 kg)
[36]							1 (10 Nm) 1 (20 Nm) 1 (30 Nm) 4 (40 Nm) 5 (50 Nm)
[37]							6,6 (20 Nm) 2,2 (40 Nm)
[38]							6,5 (20 Nm0 2,2 (40 Nm)
[39]							3,3,3 (40 Nm)
[48]	18 (10)
[49]	17 (10)
[50]	5 (12)
[19]				9 (13)		10 (11)
[26]				7	17	34
[54]			0 (1)
[55]	26 (12)
[18]						79(20)
Range	5–26	2	0–4	7–9	17	10–79	2–18	–
Trunk flexion (unilateral left or right latissimus dorsi)
[26]		3 (3)	5 (8)
[28]	8 (3)
[29]							5–20 (22–156 Nm)
[36]							1 (10 Nm) 2 (20 Nm) 3 (30 Nm) 2 (40 Nm) 5 (50 Nm)
[37]							2,2 (20 Nm) 6,5 (40 Nm)
[38]							3 (20 Nm) 4 (40 Nm)
[39]							3,3 (40 Nm)
[44]	11 (6)
[50]	5 (5)
[19]				12 (16)		24 (20)
[51]				4	6	25
[53]						23 (peak)
[52]	9 (5)
[18]						41 (14)
Range	5–11	3	5	4–12	6	23–41	2–20	–
Ipsilateral trunk lateral flexion (ipsilateral latissimus dorsi to movement direction)
[36]							1 (10 Nm) 2 (20 Nm) 4 (30 Nm) 6 (40 Nm) 12 (50 Nm)
[37]							9,9 (20 Nm) 2,3 (40 Nm)
[38]							4 (20 Nm) 6 (40 Nm)
[39]							3,4 (40 Nm)
[40]								43,46 (0°/s) 70,74 (45°/s)
[47]						35 (15)
[49]	40 (15)
[50]	7 (3)
[19]				10 (15)		15 (17)
[52]	6 (3)
[18]						85 (25)
Range	6–40	–	–	10	–	15–85	1–12	43–74
Contralateral trunk lateral flexion (contralateral latissimus dorsi to movement direction)
[37]							0,9 (20 Nm) 2,3 (40 Nm)
[39]							6,7 (40 Nm)
[40]								3,4 (0°/s) 7,9 (45°/s)
[44]	12 (10)
[50]	5 (2)
[19]				10 (11)		14 (11)
[52]	11 (7)
Range	5–12	–	–	10	–	14	0–9	– 3–9

Appendix 3.

Latissimus dorsi activation levels % MVC (SD) during combined trunk movements (peak) and trunk stability tasks (mean)

 

Table

Peak latissimus dorsi activation % MVC (±SD) during combined movements (maximal load)
Study	Ipsilateral trunk rotation and flexion	Contralateral trunk rotation and flexion	Ipsilateral trunk rotation and extension	Contralateral trunk rotation and extension
Kumar, 2010	44 (22) – 238 (449)	61 (36) – 171 (185)
Kumar, 2006			78 (41) – 204 (198)	78 (37) – 161 (112)
Mean latissimus dorsi activation % MVC (±SD) during trunk stability tasks (body weight)
Study	Ipsilateral leg extension	Contralateral leg extension	Ipsilateral leg extension, contralateral arm flexion	Contralateral leg extension, ipsilateral arm flexion
Callaghan et al, 1998	7 (5)	6 (4)	9 (8)	12 (13)
Drake et al, 2006	12 (7)	17 (10)	25 (11)	33 (15)
Kavcic et al, 2004	11 (7)	15 (5)	15 (13)	22 (16)
Stevens et al, 2007b	13 (11)	13 (11)	15 (10)	15 (11)
Range	7-13	6-17	9-25	12-33