Effectiveness of micro-exercises for managing neck/shoulder pain in sedentary workers: a systematic review and meta-analysis

Abstract

This systematic review and meta-analysis evaluated the effectiveness of micro-exercise interventions on neck/shoulder pain, neck disability, muscle activity, quality of life, and work-related outcomes among sedentary workers. PubMed, Scopus, and Web of Science were searched for randomized controlled trials (RCTs) from inception to July 2025. Nineteen studies (n = 2732) met the inclusion criteria. A meta-analysis was conducted using Comprehensive Meta-Analysis software. Micro-exercises significantly reduced combined neck/shoulder pain (4 studies; g = –0.79, 95% CI –0.98 to –0.60; I² = 39%), neck pain (11 studies; g = –1.36, 95% CI –2.09 to –0.64; I² = 98%), shoulder pain (2 studies; g = –0.81, 95% CI –1.59 to –0.03; I² = 96%), right shoulder pain (3 studies; g = –1.72, 95% CI –2.90 to –0.55; I² = 98%) and left shoulder pain (3 studies; g = –1.83, 95% CI –3.05 to –0.62; I² = 98%). Improvements were also found in neck disability (4 studies; g = –0.45, 95% CI –0.73 to –0.17; I² = 29%) and overall quality of life (2 studies; g = –0.57, 95% CI –0.94 to –0.20; I² = 0%). Negative values indicate benefit favoring the intervention. No significant effects were found for muscle activity (2 studies). Work-related outcomes were mixed, as work disability (1 study) measured by Disabilities of the Arm, Shoulder and Hand (DASH) indicated significant reductions in two of three intervention groups in one study (− 4 points, 95% CI − 8 to − 1; p < 0.05 and − 7 points, 95% CI − 10 to − 3; p < 0.01), while work ability and absenteeism showed no significant change (each 1 study). The certainty of evidence ranged from moderate (for combined neck/shoulder pain and neck disability) to very low (for shoulder pain, muscle activity, and work-related outcomes) according to the certainty of evidence (GRADE) approach. Substantial heterogeneity and potential publication bias (Egger’s p = 0.028; Begg’s p < 0.001) suggest that the pooled estimates should be interpreted with caution. Micro-exercises appear beneficial for reducing neck/shoulder pain and improving function among sedentary workers, with effect sizes exceeding minimal clinically important difference thresholds for pain. However, the certainty of evidence is moderate to very low due to high heterogeneity and potential publication bias. Evidence for work-related outcomes is mixed as work disability measured by Disabilities of the Arm, Shoulder and Hand (DASH) presented significant improvements in two of three intervention groups in one study, suggesting a possible benefit; however, work ability and absenteeism did not show significant results. Larger, well-designed trials are needed to strengthen the evidence for both clinical and occupational outcomes.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1038/s41598-026-56061-z.

Introduction

Work-related musculoskeletal disorders, especially those in the neck/shoulder regions, are prevalent in sedentary employees characterized by prolonged sitting, poor postures, and repetitive desk-based tasks^1–3. These disorders affect employees’ well-being, imposing a significant economic burden on employers and healthcare organizations⁴. Work-related musculoskeletal disorders are reported to contribute to disability, work absenteeism, and reduced productivity⁵. Due to the high prevalence of desk-based occupations and the musculoskeletal risks associated with prolonged sitting and repetitive work tasks, effective and feasible preventive strategies are needed to address neck/shoulder symptoms among working populations⁶. Conventional approaches such as medication, physical therapy, and ergonomic modification may provide benefit⁷. However, when implemented in isolation they may be limited by cost, time demands, access barriers, or suboptimal long-term adherence^8,9. These practical limitations highlight the need for scalable and workplace-feasible interventions that can be integrated into daily routines^10,11.

Micro-exercises are brief (2–20 min), targeted physical activities that can be performed at the workstation without special equipment, offering a low-cost, accessible alternative to time-intensive interventions for neck/shoulder pain^12–14. Unlike passive micro-breaks or general active breaks, micro-exercises involve purposeful therapeutic movements such as strengthening, mobility, or stretching exercises. In contrast, conventional workplace exercise programs are usually longer, less task-integrated, and may require dedicated exercise time or facilities.

Previous reviews suggest that workplace exercise interventions may reduce musculoskeletal symptoms among office workers. In particular, strengthening and stretching exercises have been associated with improvements in pain and functional outcomes^15–17. However, these reviews were not specifically focused on micro-exercises for desk-based workers and were limited by methodological issues such as risk of bias, variability in study quality, lack of standardization in exercise protocols, and inconsistent outcome measures. Moreover, reviews addressing micro-breaks have generally examined workplace breaks more broadly rather than specific micro-exercises targeting neck/shoulder symptoms¹⁸. Although therapeutic exercise interventions have been explored in the literature, a comprehensive systematic review specifically examining the role of micro-exercises in reducing pain and improving functional outcomes among sedentary employees remains lacking.

Therefore, a comprehensive and methodologically rigorous synthesis of the available evidence is warranted. This systematic review and meta-analysis were conducted to evaluate the effectiveness of micro-exercises on primary outcomes of neck/shoulder pain, and secondary outcomes including neck disability, quality of life, muscle activity, work ability, work disability, and absenteeism in desk-based workers.

Methods

The present systematic review with meta-analysis followed PRISMA guidelines¹⁹. This review was registered in the International Prospective Register of Systematic Reviews (PROSPERO) under the number CRD42024581513 in August 2024 and amended in September 2025.

Eligibility criteria

Included studies had to meet the following eligibility criteria:

• Population: Sedentary workers aged ≥ 18 years with neck and/or shoulder pain of musculoskeletal origin.

• Intervention: Micro-exercises: structured, targeted physical activities lasting 2–20 min, performed at the workstation without special equipment, and specifically engaging neck and/or shoulder musculature.

• Comparators: No intervention, alternative exercise programs, medication, or physical therapy.

• Outcomes: Primary: pain intensity (neck, shoulder, or both). Secondary: neck disability, quality of life, muscle activity patterns, work ability, work disability, absenteeism.

• Exclusion criteria: Non-English publications; populations not meeting the definition (e.g., non-sedentary workers, non-musculoskeletal pain); absence of specified outcomes; interventions not meeting the micro-exercise definition (e.g., > 20 min, gym-based, non-targeted)

Information source

An extensive search was performed to identify potentially relevant studies evaluating the efficacy of workplace micro-exercises among sedentary workers. We searched five major bibliographic databases, including, PubMed, Scopus, and Web of Science. All included studies were published in English, with no restriction on publication date. The search was conducted from database inception to October 2024 (45,019 records) and updated in July 2025 (7760 additional records), resulting in 52,779 total retrieved records prior to duplicate removal.

Search strategies used both controlled vocabulary (MeSH in PubMed) and free-text keywords. Additionally, reference lists of included studies and relevant reviews were manually screened. Trial registries (ClinicalTrials.gov, WHO ICTRP) and grey literature (OpenGrey, ProQuest Dissertations & Theses) were searched, but no additional eligible studies were identified.

Eligible studies were subsequently selected based on predefined inclusion criteria, which required an RCT design. All identified records were imported into EndNote v20 (Clarivate Analytics, Philadelphia, PA). Initial duplicate detection was performed using EndNote’s automated duplicate identification algorithm, with manual verification of potential matches (n = 273 duplicates removed). The remaining records were uploaded to Covidence (Veritas Health Innovation, Melbourne, Australia), which applied its built-in duplicate detection algorithm, removing an additional 5655 duplicates. Covidence also applied its machine learning-based automation tool to screen titles and abstracts against pre-defined exclusion criteria (animal studies, non-English language, non-RCT designs, clearly irrelevant populations). Records flagged as ineligible by the automation tool (n = 20,936) were reviewed by two authors (ZY and MB); all were confirmed as ineligible and excluded before the manual screening process. The remaining 25,915 records underwent traditional title/abstract screening by two independent reviewers. All removal decisions were documented in the Covidence audit trail.

Database-wise search results are presented in supplementary material. The search strategy was deliberately designed to be highly sensitive (prioritizing recall over precision) because the terminology for ‘micro-exercises’ is heterogeneous and not standardized in MeSH. The broad search yielded a large initial record count, which is justified by the need to avoid missing relevant RCTs. Microsoft Excel for Microsoft 365 was used for data extraction and management of study details.

Study selection and data collection

Two independent reviewers (ZY and MB) performed title/abstract screening and full-text assessment independently and without blinding to study authors, institutions, or results. All records were managed in Covidence and disagreements at either stage were resolved first by discussion between the two reviewers; if consensus could not be reached, a third reviewer (MG) was consulted. The same two reviewers independently performed data extraction using a pre-piloted standardized form, with disagreements resolved similarly.

Data items

Data extracted using a standardized form included: author, year, country, study design, measured outcomes (primary: neck, shoulder, or neck-shoulder pain intensity; secondary: muscle activity, quality of life, neck disability, work ability, work disability, absenteeism). Participant information (mean ± SD, 95% CI, age, weight, height, sample size) and intervention/comparator details (type, dosage, frequency, duration, outcome scale) were also extracted. Missing data were handled according to Cochrane recommendations as studies with insufficient data were excluded only from quantitative synthesis without imputation. To enhance conceptual clarity and reproducibility, a decision framework was applied during study selection: (1) structured, purposeful physical activity; (2) specifically targets neck/shoulder; (3) duration 2–20 min per session; (4) workplace-feasible without specialized facilities. Interventions described as “micro-breaks” were included only if they met all criteria. Detailed intervention characteristics were extracted descriptively (Table 5).

Table 5.

Conceptual framework and classification criteria for workplace exercise interventions.

Domain	Micro-exercise	Micro-break Exercise (Subtype)	Conventional Workplace Exercise
Core definition	Structured, targeted, short-duration exercise integrated into the work task	Brief exercise performed during work breaks that meets micro-exercise criteria	Structured exercise performed as separate sessions (before/after work or scheduled programs)
Primary purpose	Therapeutic (pain reduction, functional improvement)	Combined: interruption of sedentary behavior + therapeutic effect (if structured)	Fitness, rehabilitation, or general conditioning
Structure	Structured and protocol-driven	Semi-structured, included only if structured	Structured and supervised
Target specificity	Specific (neck/shoulder muscles)	Variable, included only if targeted	General or whole-body
Duration per bout	2–20 min	Typically 1–10 min	Typically 20–60 min
Workplace integration	Performed at the workstation without workflow disruption	Performed during short breaks	Requires dedicated time and often a separate setting
Equipment requirement	Minimal (e.g., bands, light dumbbells)	Minimal or none	May require equipment/facilities
Supervision	None or minimal	None	Often supervised
Examples	Resistance band neck strengthening, scapular activation	Stretching or strengthening during short breaks (if structured and targeted)	Gym-based training, organized workplace fitness programs
Classification rule applied in this review	Included if all criteria are met	Included only if ALL of the following are present: (1) structured protocol, (2) targeted neck/shoulder activity, (3) therapeutic intent	Excluded

Interventions were classified based on structure, target specificity, duration, and purpose. Micro-exercises were defined as structured, targeted activities lasting 2–20 min. Micro-break exercises were included only if they met these criteria. Conventional workplace exercise programs were excluded, characterized by longer duration and separation from work tasks.

Risk of bias within studies

Two independent reviewers assessed risk of bias using the Cochrane RoB 1.0 tool within Covidence, consistent with the pre-specified PROSPERO protocol (RoB 2.0 was not applied to avoid post-hoc methodological changes). Study risk-of-bias judgments were based on the domain-level assessments.

Domains evaluated: random sequence generation, allocation concealment, blinding of participants/personnel, blinding of outcome assessors, incomplete outcome data, selective outcome reporting, and other sources of bias (e.g., funding or conflicts of interest). Each domain was judged as low, high, or unclear risk using standard criteria. A domain was judged as “low” risk when sufficient methodological details were reported, indicating appropriate procedures (e.g., computer-generated randomization, adequate allocation concealment, low attrition with appropriate handling of missing data, or blinded outcome assessment when feasible). A domain was judged as “high” risk when clear methodological limitations were identified that could introduce bias (e.g., inadequate concealment, substantial attrition without appropriate statistical handling, or evidence of selective reporting). A domain was classified as “unclear” risk when insufficient methodological information was reported to allow a definitive judgment. Disagreements were resolved by discussion or by consulting a third reviewer (MG).

Risk of bias across studies

To assess potential reporting/publication bias, we visually inspected funnel plots for meta-analyses with ≥ 10 studies and performed Egger’s and Begg’s tests where applicable. Asymmetry was interpreted cautiously, considering small-study effects and heterogeneity. Formal bias-adjustment methods (e.g., trim-and-fill) were not performed because the number of studies for most pooled outcomes fell below the recommended minimum (≈10 studies) for reliable application (Cochrane Handbook for Systematic Reviews of Interventions). Consequently, the possibility that pooled effect estimates may overestimate the true intervention effect cannot be excluded.

Certainty of evidence

The certainty of evidence for each outcome was evaluated using the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) framework. Because all included studies were RCTs, evidence started at high and was downgraded based on five domains: risk of bias, inconsistency (heterogeneity), indirectness, imprecision, and publication bias. Downgrading occurred when substantial heterogeneity (I² > 75%), small numbers of studies, wide CIs, or evidence of publication bias was present. Certainty was categorized as high, moderate, low, or very low (Table 6).

Table 6.

Summary of findings and GRADE assessment of certainty of evidence.

Outcome	No. of Studies	Effect Size (SMD)	95% CI	Risk of Bias	Inconsistency (I2)	Imprecision	Publication Bias	Certainty (GRADE)	Reasons for Downgrading
Neck & Shoulder Pain	4	–0.79	–1.05 to –0.53	Not serious	Serious (I2 = 39%)	Not serious	Suspected	Moderate	Downgraded for inconsistency and suspected publication bias
Neck Pain	11	–1.36	–1.98 to –0.75	Not serious	Very serious (I2 = 98%)	Not serious	Suspected	Low	Downgraded for very serious inconsistency and suspected publication bias
Shoulder Pain (combined)	2	–0.81	–1.59 to –0.03	Some concerns	Very serious (I2 = 96%)	Serious	Not assessable	Very Low	Downgraded for risk of bias, very serious inconsistency, and imprecision
Shoulder Pain (right)	3	–1.72	–2.88 to –0.56	Some concerns	Very serious (I2 = 98%)	Serious	Not assessable	Very Low	Downgraded for very serious inconsistency and imprecision
Shoulder Pain (left)	3	–1.83	–3.04 to –0.62	Some concerns	Very serious (I2 = 98%)	Serious	Not assessable	Very Low	Downgraded for very serious inconsistency and imprecision
Quality of Life (overall)	2	–0.57	–0.93 to –0.21	Some concerns	Not serious (I2 = 0%)	Serious	Not assessable	Low	Downgraded for risk of bias and imprecision
Neck Disability	4	–0.45	–0.73 to –0.17	Not serious	Not serious (I2 = 29%)	Serious	Not assessable	Moderate	Downgraded for imprecision
Muscle Activity	2	–0.76	–1.60 to 0.08	Some concerns	Serious (I2 = 79%)	Serious	Not assessable	Very Low	Downgraded for inconsistency and imprecision
Work Disability	1	Not pooled (single study)	–	Some concerns	Not applicable	Very serious	Not assessable	Very Low	Downgraded for single-study evidence and very serious imprecision
Work Ability	1	Not pooled (single study)	–	Some concerns	Not applicable	Very serious	Not assessable	Very Low	Downgraded for single-study evidence and very serious imprecision
Absenteeism	1	Not pooled (single study)	–	Some concerns	Not applicable	Very serious	Not assessable	Very Low	Downgraded for single-study evidence and very serious imprecision

Effect sizes are presented as standardized mean differences (Hedges’ g), where negative values indicate improvement favoring micro-exercise interventions. Certainty of evidence was assessed using the GRADE approach. Evidence from randomized controlled trials started at high certainty and was downgraded based on risk of bias, inconsistency, imprecision, and publication bias. Publication bias was not formally assessed for outcomes with fewer than 10 studies.

Data synthesis and statistical analysis

All analyses used Comprehensive Meta-Analysis software (version 4). Meta-analysis was conducted when ≥ 2 studies reported a given outcome; single-study outcomes were described narratively. Effect sizes were calculated as Hedges’ g (standardized mean difference) with 95% CIs; negative values indicate improvement favoring micro-exercises (for quality of life, direction reversed). When multiple time points were reported, post-intervention outcomes were extracted. One cluster-randomized trial was included; effect estimates were used as reported (insufficient ICCs for adjustment). Random-effects models were used due to anticipated heterogeneity, which was quantified using I². Publication bias was assessed visually using funnel plots, together with Egger’s and Begg’s tests where appropriate. Effect sizes were interpreted using Cohen’s criteria (0.2 = small, 0.5 = moderate, 0.8 = large). For anatomically distinct but related outcomes from the same study (e.g., left and right shoulder pain), separate meta-analyses were conducted. Due to the small number of studies (n = 3), multivariate meta-analysis or robust variance estimation was not feasible; therefore, these pooled estimates are exploratory, and the statistical dependence is acknowledged as a limitation. Decision rules for multiple effect sizes from the same study (applied sequentially): Time point: post-intervention (shortest follow-up after intervention end) preferred over follow-up. Outcome scale: when multiple scales measured the same construct, the most commonly used scale across studies (VAS) was preferred; otherwise, the scale with highest face validity for sedentary workforces was selected.

Multi-arm studies: if a study had multiple intervention arms meeting the micro-exercise definition, the arm with the most representative dose (mid-range duration/frequency) was selected. If a study included both a micro-exercise arm and a non-qualifying arm, only the micro-exercise arm was included. Comparators were combined using standard formulae.

Multiple outcomes within the same domain: for pain, intensity (most directly comparable) was selected over frequency.

Due to the small number of studies per outcome and inconsistent reporting of moderators, subgroup analyses, meta-regression, and formal sensitivity testing (e.g., leave-one-out) were not performed or considered robust.

Results

Study selection and characteristics

The database searches yielded a total of 52,779 records (initial search October 2024: n = 45,019; update search July 2025: n = 7760). Database-wise results are presented in Supplementary file. After removal of duplicates (n = 5928: 273 manual, 5655 Covidence) and exclusion of records marked as ineligible by Covidence’s automation tool (n = 20,936; confirmed by authors), 25,915 records proceeded to title/abstract screening. After title and abstract screening, 1116 full texts were assessed for eligibility, and 19 studies met the inclusion criteria and were included in the final analysis (Fig. 1, PRISMA 2020 flow diagram). The high initial yield reflects a sensitive search strategy designed to capture the diverse terminology used to describe micro-exercise interventions, as detailed in the Supplementary Material.

Fig. 1.

PRISMA 2020 flow diagram of study identification, screening, eligibility assessment, and inclusion. The diagram follows the PRISMA 2020 guidelines. Automation tools: Covidence’s machine-learning algorithm flagged records as ineligible based on title/abstract content.

The 19 included studies were published between 2007 and 2025. The majority were conducted in Denmark (n = 11)^20–30, with remaining studies from Iran (n = 2)^31,32, Greece (n = 1)³³, Japan (n = 1)³⁴, Malaysia (n = 1)³⁵, South Korea (n = 1)³⁶, Thailand (n = 1)³⁷, and the United States (n = 1)³⁸. Total participant number across all studies was 2732. Mean age of participants ranged from 27.1 ± 4.8 to 49.0 ± 1.4 years across studies. Interventions consisted of structured micro-exercises targeting the neck/shoulder regions, including resistance training, stretching, strengthening, and mixed protocols. Detailed study characteristics (design, participant demographics, intervention dose, frequency, duration, type, outcome measures, scales, and data at all time points) are presented in Table 1 (primary outcomes) and Table 2 (secondary outcomes).

Table 1.

Characteristics of included studies and data extraction for primary outcomes.

Authors/ Years/Country	Design	Primary Outcome	Group	N (allocated)	Timepoint	Mean ± SD	CI(Lower–Upper)	Mean age (years) ± SD	Mean Weight (kg) ± SD	Mean Height (cm) ± SD	Intervention Dose	Intervention Frequency	Intervention Duration	Intervention Type	Outcome Scale	Outcome Range/Units
Kietrys 2007, United States	RCT	Neck Pain	Intervention	24	Baseline	1.16 ± 1.86	0.8–1.97	not reported	not reported	not reported	~ 6–8 min/day	2 times/day	4 weeks	Resistance exercise - Isometric cervical rotation: 5 s hold, 5 reps each direction - Shoulder shrug: 12 reps with band - Scapular retraction: 12 reps with band	Visual Analog Scale (VAS)	0–10 (0 = no pain, 10 = worst pain imaginable)/cm
					Follow-up	1.89 ± 2.77	1.19–3.09
			Intervention	24	Baseline	1.24 ± 1.96	0.84–2.09				~ 4–6 min/day	2 times/day	4 weeks	Stretching exercise - Lateral cervical stretch: 5 s hold, 5 reps per side - Posterior neck stretch - Arm/forearm stretch
					Follow-up	1.33 ± 2.34	1.01–2.35
			Control	24	Baseline	1.16 ± 2.05	0.88–2.06				~ 4–6 min/day	2 times/day	4 weeks	Minimal movement (control activity) - Deep breathing: 5 reps - Seated ankle pumps: 10 reps
					Follow-up	2.45 ± 3.33	1.44–3.89
Andersen 2008, Denmark	RCT	Neck Pain	Intervention(SST)	18	Baseline	44 ± 25	31.6–56.4	44 ± 9	73 ± 12	165 ± 6	20 min per session	3 times per week	10 weeks	Dumbbell-based strength training of neck/shoulder muscles	Visual Analog Scale (VAS)	0–100 (mm)
					Follow-up	10 ± 10	15-May
			Control	8	Baseline	43 ± 27	20.5–65.5	42 ± 8	68 ± 12	166 ± 8	Up to 1 h per week	Irregular group/individual sessions	10 weeks	Health and lifestyle counseling (no physical activity)
					Follow-up	35 ± 29	10.8–59.2
Andersen 2008, Denmark	RCT	Shoulder Pain	Intervention (SRT)	180	Baseline	3.5 ± 0.3	3.2–3.8	f:44 ± 0.9 , m:49 ± 1.4	f :69 ± 1.1, m:83 ± 2.0	f:168 ± 0.01, m: 181 ± 0.01	20 min per session	3 times per week	12 months	Specific resistance training for neck/shoulder muscles, including dynamic and static exercises	0–9 scale (0 = no complaints, 9 = worst pain possible)	Pain intensity (scale 0–9)
					Follow-up	3.2 ± 0.2	3–3.4
			Control	182	Baseline	3.6 ± 0.5	3.1–4.1	f: 44 ± 0.9 ,m: 48 ± 1.4	f: 69 ± 1.5, m: 84 ± 2.7	f: 168 ± 0.01, m: 181 ± 0.01	Counseling and health-promoting activities	1–4 times per month	12 months	Health counseling and health-promoting activities
					Follow-up	4 ± 0.6	3.4–4.6
Andersen 2008, Denmark	RCT	Neck Pain	Intervention (SRT)	180	Baseline	5 ± 0.2	4.6–5.4	f:44 ± 0.9 , m:49 ± 1.4	f :69 ± 1.1, m:83 ± 2.0	m: 181 ± 0.01 , f:168 ± 0.01	20 min per session	3 times per week	12 months	Specific resistance training for neck/shoulder muscles, including dynamic and static exercises	0–9 scale (0 = no complaints, 9 = worst pain possible)	Pain intensity (scale 0–9)
					Follow-up	3.4 ± 0.2	3–3.8
			Control	182	Baseline	3.5 ± 0.3	3.2–3.8	f: 44 ± 0.9 , m: 48 ± 1.4	f: 69 ± 1.5 , m: 84 ± 2.7	f: 168 ± 0.01 , m: 181 ± 0.01	Counseling and health-promoting activities	1–4 times per month	12 months	Health counseling and health-promoting activities
					Follow-up	3.5 ± 0.3	3.2–3.8
Andersen 2010, Denmark	RCT	Neck Pain	Intervention(SRT)	180	Baseline	4.66 ± 0.1	4.64–4.67	f: 44.6 ± 10.2 , m: 45.7 ± 9.8	f: 68.2 ± 15.6 , m: 83.1 ± 14.6	f: 168 ± 6.2 , m:181 ± 6.8	20 min	3 times per week	12 months	Specific resistance training focused on neck and shoulder muscles	Nordic Musculoskeletal Questionnaire (NMQ) (0–9)	0–9 Pain Scale (0 = no pain, 9 = maximum pain)
					Follow-up	2.71 ± 1.9	2.43–2.98
			Control	182	Baseline	0.95 ± 0.05	0.94–0.95	f: 44.6 ± 10.2 , m: 45.7 ± 9.8	f: 68.2 ± 15.6, m: 83.1 ± 14.6	f: 168 ± 6.2 , m: 181 ± 6.8	No exercise, focused on health-promoting presentations	Not specified (focus on health-promoting presentations)	12 months	Participants self-organized activities related to health improvement without physical exercises
					Follow-up	1.05 ± 0.3	1–1.09
Andersen 2010, Denmark	RCT	Shoulder Pain / Left	Intervention(SRT)	180	Baseline	4.83 ± 0.16	4.51–5.15	f: 44.6 ± 10.2 , m: 45.7 ± 9.8	f: 68.2 ± 15.6, m: 83.1 ± 14.6	f: 168 ± 6.2 , m: 181 ± 6.8	20 min	3 times per week	12 months	Specific resistance training focused on neck and shoulder muscles	Nordic Musculoskeletal Questionnaire (NMQ) (0–9)	0–9 Pain Scale (0 = no pain, 9 = maximum pain)
					Follow-up	2.83 ± 1.85	2.43–3.23
			Control	182	Baseline	0.37 ± 0.07	0.23–0.51	f: 44.6 ± 10.2 ,m: 45.7 ± 9.8	f: 68.2 ± 15.6, m: 83.1 ± 14.6	f: 168 ± 6.2 , m: 181 ± 6.8	No exercise, focused on health-promoting presentations	Not specified (focus on health-promoting presentations)	12 months	Participants self-organized activities related to health improvement without physical exercises
					Follow-up	0.47 ± 0.28	0.36–0.58
Andersen 2010, Denmark	RCT	Shoulder Pain /Right	Intervention(SRT)	180	Baseline	4.79 ± 0.13	4.54–5.04	f: 44.6 ± 10.2 , m: 45.7 ± 9.8	f: 68.2 ± 15.6 , m: 83.1 ± 14.6	f: 168 ± 6.2 , m: 181 ± 6.8	20 min	3 times per week	12 months	Specific resistance training focused on neck and shoulder muscles	Nordic Musculoskeletal Questionnaire (NMQ) (0–9)	0–9 Pain Scale (0 = no pain, 9 = maximum pain)
					Follow-up	2.79 ± 1.85	2.45–3.13
			Control	182	Baseline	0.59 ± 0.04	0.51–0.67	f: 44.6 ± 10.2 , m: 45.7 ± 9.8	f: 68.2 ± 15.6, m: 83.1 ± 14.6	f: 168 ± 6.2 , m: 181 ± 6.8	No exercise, focused on health-promoting presentations	Not specified (focus on health-promoting presentations)	12 months	Participants self-organized activities related to health improvement without physical exercises
					Follow-up	0.67 ± 0.28	0.58–0.76
Andersen 2011, Denmark	RCT	Neck-Shoulder Pain	Intervention(2-min)	66	Baseline	5.2 ± 2.1	4.5–6.7	44 ± 11	72 ± 14	171 ± 8	2 min of progressive resistance training per day	5 times per week	10 weeks	Progressive resistance training with elastic tubing	0–10(Pain intensity scale)	0 to 10 (where 0 is no pain and 10 is worst imaginable pain)
					Follow-up	3.8 ± 2.2	2.5–5.2
			Intervention(12 min)	66	Baseline	5 ± 1.9	4.1–6	42 ± 11	68 ± 15	170 ± 8	12 min of progressive resistance training per day	5 times per week	10 weeks	Progressive resistance training with elastic tubing
					Follow-up	3.1 ± 2.2	2.5–4.6
			Control	66	Baseline	5.2 ± 2.1	4.5–6.7	43 ± 10	67 ± 11	169 ± 7	Weekly e-mailed information	Weekly	10 weeks	Information on general health
					Follow-up	4.9 ± 2.1	4.3–5.7
Andersen 2012, Denmark	RCT	Neck Pain	Intervention (3WS)	126	Baseline	3.13 ± 2.41	2.7–3.56	46 ± 10	75 ± 18	173 ± 10	20 min per session	3 times per week	20 weeks	High-intensity strength training for neck and shoulder muscles (with dumbbells)	0–9 (Pain intensity scale)	0 (no pain) to 9 (worst possible pain)
					Follow-up	2.41 ± 1.61	0.9–2.87
			Intervention(9WS)	106	Baseline	3.05 ± 2.3	2.61–3.49	45 ± 10	78 ± 15	175 ± 9	7 min per session	9 times per week	20 weeks	High-intensity strength training for neck and shoulder muscles (with dumbbells)
					Follow-up	2.3 ± 1.66	0.24–2.46
			Control	101	Baseline	3.24 ± 2.26	2.79–3.69	46 ± 10	80 ± 16	175 ± 9	No training	No training	20 weeks	No intervention
					Follow-up	2.26 ± 2.26	1.81–2.71
Andersen 2012, Denmark	RCT	Shoulder Pain/Right	Intervention (3WS)	126	Baseline	2.32 ± 2.38	1.9–2.74	46 ± 10	75 ± 18	173 ± 10	20 min per session	3 times per week	20 weeks	High-intensity strength training for neck and shoulder muscles (with dumbbells)	0–9 (Pain intensity scale)	0 (no pain) to 9 (worst possible pain)
					Follow-up	1.71 ± 4.11	0.99–2.43
			Intervention(9WS)	106	Baseline	1.88 ± 2.22	1.45–2.31	45 ± 10	78 ± 15	175 ± 9	7 min per session	9 times per week	20 weeks	High-intensity strength training for neck and shoulder muscles (with dumbbells)
					Follow-up	1.05 ± 4.07	0.27–1.83
			Control	101	Baseline	2.01 ± 2.39	1.54–2.48	46 ± 10	80 ± 16	175 ± 9	No training	No training	20 weeks	No intervention
					Follow-up	1.82 ± 2.39	1.35–2.29
Andersen 2012, Denmark	RCT	Shoulder Pain/Left	Intervention (3WS)	126	Baseline	1.75 ± 2.28	1.35–2.15	46 ± 10	75 ± 18	173 ± 10	20 min per session	3 times per week	20 weeks	High-intensity strength training for neck and shoulder muscles (with dumbbells)	0–9 (Pain intensity scale)	0 (no pain) to 9 (worst possible pain)
					Follow-up	1.56 ± 2.43	1.13–1.99
			Intervention(9WS)	106	Baseline	1.75 ± 2.18	1.33–2.17	45 ± 10	78 ± 15	175 ± 9	7 min per session	9 times per week	20 weeks	High-intensity strength training for neck and shoulder muscles (with dumbbells)
					Follow-up	1.43 ± 2.32	0.99–1.87
			Control	101	Baseline	1.5 ± 1.92	1.12–1.88	46 ± 10	80 ± 16	175 ± 9	No training	No training	20 weeks	No intervention
					Follow-up	1.57 ± 1.92	1.19–1.95
Lidegaard 2013, Denmark	RCT	Neck-Shoulder Pain	Intervention	15	Baseline	3.44 ± 1.4	2.66–4.21	41.7 ± 10.8	66.5 ± 9.07	168.8 ± 6.68	max 2 min	Daily on workdays (average 4.3 sessions/week over 10 weeks)	10 weeks	High-intensity resistance training using elastic tubing (“lateral raise” exercise)	0–10 (Pain intensity scale)	0–10 (numeric pain rating scale)
					Follow-up	2.04 ± 1.6	1.15–2.92
			Control	15	Baseline	3.24 ± 1.37	2.48–3.99	40.5 ± 7.27	65.2 ± 10.1	166.1 ± 4.44	No exercise; weekly health-related email	One email per week for 10 weeks	10 weeks	Health information via weekly emails
					Follow-up	3.45 ± 1.99	2.34–4.55
Gram 2014, Denmark	Cluster RCT	Neck Pain	Intervention(3WS)	126	Baseline	3.13 ± 2.41	2.71–3.56	46 ± 10	74.6 ± 17.3	173.1 ± 9.7	20 min each session	3 times per week	20 weeks	Supervised strength training with instructors throughout the intervention	0–9 (Pain Intensity)	(0: No pain, 9: Worst pain)
					Follow-up	1.5 ± 1.63	1.13–1.87
			Intervention(3MS)	124	Baseline	3.24 ± 2.36	2.82–3.66	45 ± 11	78.6 ± 15.9	174.6 ± 10.2	20 min each session	3 times per week	20 weeks	Minimal supervision (only at the start for instructions)
					Follow-up	1.78 ± 2.08	1.26–2.3
			Control	101	Baseline	3.24 ± 2.26	2.79–3.68	46 ± 10	80.2 ± 15.5	175.3 ± 9.4	No exercise intervention	No exercise intervention	20 weeks	No exercise intervention
					Follow-up	2.64 ± 2.33	2.13–3.16
Gram 2014, Denmark	Cluster RCT	Shoulder Pain	Intervention(3WS)	126	Baseline	2.32 ± 2.38	1.9–2.74	46 ± 10	74.6 ± 17.3	173.1 ± 9.7	20 min each session	3 times per week	20 weeks	Supervised strength training with instructors throughout the intervention	0–9 (Pain intensity scale)	(0: No pain, 9: Worst pain)
					Follow-up	0.95 ± 1.51	0.6–1.29
			Intervention(3MS)	124	Baseline	1.99 ± 2.36	1.57–2.41	45 ± 11	78.6 ± 15.9	174.6 ± 10.2	20 min each session	3 times per week	20 weeks	Minimal supervision (only at the start for instructions)
					Follow-up	1.11 ± 1.84	0.65–1.57
			Control	101	Baseline	2.01 ± 2.39	1.54–2.48	46 ± 10	80.2 ± 15.5	175.3 ± 9.4	No exercise intervention	No exercise intervention	20 weeks	No exercise intervention
					Follow-up	1.69 ± 2.08	1.23–2.15
Andersen 2014, Denmark	RCT	Neck-Shoulder Pain	Intervention	24	Baseline	5.7 ± 1.9	4.94–6.46	44 ± 13	72 ± 13	171 ± 7	20 min per session	3 sessions per week	10 weeks	Scapular Function Training (SFT) involving intensive training of the lower trapezius and serratus anterior muscles while minimizing activation of the upper trapezius	0–9 (Pain intensity scale)	0 (no pain) to 9 (worst possible pain)
					Follow-up	3.7 ± 1.7	3.02–4.38
			Control	23	Baseline	5.4 ± 1.5	4.79–6.01	45 ± 11	72 ± 12	171 ± 8	No physical training (encouraged to stay active)	No physical training (encouraged to stay active)	10 weeks	No physical training (the participants were encouraged to stay active as usual)
					Follow-up	5.6 ± 1.6	4.99–6.21
Tunwattanapong 2016, Thailand	RCT	Neck Pain	Intervention	48	Baseline	6.7 ± 1.2	6–7.4	34.2 ± 9.0	not reported	not reported	10–15 min per session	Two sessions per day. 5 days per week for 4 weeks	4 weeks	Neck and shoulder stretching exercises (20–30 repetitions/session), including neck stretching, shoulder stretching, shoulder rolling, trunk stretching, and back extension exercises	Visual Analogue Scale (VAS)	0–10 (cm)
					Follow-up	4.5 ± 1.8	3.5–5.5
			Control	48	Baseline	6.2 ± 1	5.5–6.8	36.5 ± 8.7			No exercise, only educational brochure on ergonomics	Informative brochure on ergonomics	4 weeks	Informative brochure on ergonomics
					Follow-up	5.6 ± 1.8	4–7.1
Lee 2017, South Korea	RCT	Neck Pain	Intervention	11	Baseline	5.2 ± 2.19	3.91–6.49	27.09 ± 4.83	63.73 ± 15.62	168 ± 10.15	10–15 min per session	2 days/week	8 weeks	McKenzie neck exercise via mobile app	Visual Analog Scale (VAS)	0 (no pain) to 100 (unbearable pain)/ mm
					Follow-up	2.73 ± 1.99	1.55–3.91
			Control	9	Baseline	4.02 ± 1.75	2.88–5.16	27.56 ± 4.67	73.11 ± 13.29	172.56 ± 7.09	Brochure	Single education session	8 weeks	Posture correction education (brochure)
					Follow-up	3.69 ± 2.02	2.37–5.01
Shariat 2018, Malaysia	RCT	Neck- Pain	Intervention	43	Baseline	13.46 ± 2.57	12.67–14.25	29.41 ± 1.16	73.26 ± 3.75	163.29 ± 1.46	10–15 min per session	3 times per week	6 months	Stretching exercises (neck, shoulder, back)	Cornell Musculoskeletal Disorders Questionnaire (CMDQ)	Pain scores (Likert scale)
					Follow-up	1.88 ± 0.35	1.77–1.99
			Control	28	Baseline	15.65 ± 2.8	14.56–16.74	28.74 ± 0.82	72.61 ± 2.60	161.25 ± 1.57	No intervention		No intervention	No treatment
					Follow-up	12.55 ± 2.24	11.68–13.42
Shariat 2018, Malaysia	RCT	Shoulder Pain/Right	Intervention	43	Baseline	15.34 ± 2.45	14.59–16.09	29.41 ± 1.16	73.26 ± 3.75	163.29 ± 1.46	10–15 min per session	3 times per week	6 months	Stretching exercises (neck, shoulder, back)	Cornell Musculoskeletal Disorders Questionnaire (CMDQ)	Pain scores (Likert scale)
					Follow-up	1.41 ± 0.22	1.34–1.48
			Control	28	Baseline	18.24 ± 3.42	16.91–19.57	28.74 ± 0.82	72.61 ± 2.60	161.25 ± 1.57				No treatment
					Follow-up	13.05 ± 2.44	12.1–13.99
Shariat 2018, Malaysia	RCT	Shoulder Pain/Left	Intervention	43	Baseline	13.41 ± 2.08	12.77–14.05	29.41 ± 1.16	73.26 ± 3.75	163.29 ± 1.46	10–15 min per session	3 times per week	6 months	Stretching exercises (neck, shoulder, back)	Cornell Musculoskeletal Disorders Questionnaire (CMDQ)	Pain scores (Likert scale)
					Follow-up	1.65 ± 0.25	1.57–1.73
			Control	28	Baseline	15.21 ± 2.8	14.12–16.29	28.74 ± 0.82	72.61 ± 2.60	161.25 ± 1.57				No treatment
					Follow-up	12.74 ± 2.34	11.83–13.65
Anan 2021, Japan	RCT	Neck-Shoulder Pain	Intervention	48	Baseline	4 ± 1.1	3.69–4.31	41.8 ± 8.7	not reported	not reported	Short exercise routine (1 min per day)	7 days per week	12 weeks	AI-assisted health program via mobile app (LINE)	1 to 5 (Pain Scale)	1 to 5 (Pain level)
					Follow-up	3 ± 1.1	2.69–3.31
			Control	46	Baseline	4 ± 0.8	3.77–4.23	42.4 ± 8.0			Regular exercise routine (3 min per day)	7 days per week		No additional intervention, regular exercises at work
					Follow-up	4 ± 0.8	3.77–4.23
Tabanfar 2022, Iran	RCT	Neck Pain	Intervention	42	Baseline	3.83 ± 2.12	3.16–4.49	41.74 ± 7.55	65.94 ± 8.00	166.33 ± 9.38	15 min per session	5 sessions per week	12 weeks	Corrective exercises targeting neck muscles/Neck muscle strengthening exercises targeting flexors and scapulothoracic muscles (e.g., rhomboid, trapezius, pectoralis)	Visual Analog Scale (VAS)	0–10 (0 = no pain, 10 = worst pain imaginable)
					Follow-up	2.38 ± 1.82	1.81–2.94
			Control	43	Baseline	3 ± 1.78	2.45–3.54	42.53 ± 8.04	66.00 ± 8.17	168.49 ± 7.99	No intervention	No intervention	12 weeks	No intervention
					Follow-up	3.09 ± 1.75	2.55–3.62
Michaelidis 2025, Greece	RCT	Neck Pain	Intervention	34	Baseline	4.2 ± 2.1	3.49–4.91	38.4 ± 8.0	not reported	not reported	10 min per session	5 days per week, 20 sessions over 4 weeks	4 weeks	Static stretching of neck musculature	Numerical Pain Rating Scale (NPRS)	0–10 (from no pain to worst imaginable pain)
					Follow-up	3.2 ± 1.9	2.56–3.84
			Control	32	Baseline	4.3 ± 1.8	3.68–4.92	38.1 ± 8.8	not reported	not reported	No active intervention	No active intervention	4 weeks	Participants received a booklet and training diary
					Follow-up	3.5 ± 1.7	2.91–4.09

Summary of study characteristics, including study identification (author, year, country), study design, outcome type (neck/shoulder pain), group allocation, sample size, assessment time points, outcome data (mean ± SD and 95% CI), participant characteristics (age, weight, height), intervention characteristics (dose, frequency, duration, and type), and outcome measurement details (assessment scale and units).

Table 2.

Characteristics of included studies and data extraction for secondary outcomes.

Authors/Years/Country	Design	Secondary Outcome	Group	N (allocated)	Timepoint	Mean ± SD	CI (Lower–Upper)	Mean age (years) ± SD	Mean Weight (kg) ± SD	Mean Height (cm) ± SD	Intervention Dose	Intervention Frequency	Intervention Duration	Intervention Type	Outcome Scale	Outcome Range/Units
Kietrys 2007, United States	RCT	Neck Disability	Intervention	24	Baseline	4.65 ± 3.89	1.68–6.33	not reported	not reported	not reported	~ 6–8 min/day	2 times/day	4 weeks	Resistance exercise - Isometric cervical rotation: 5 s hold, 5 reps each direction - Shoulder shrug: 12 reps with band - Scapular retraction: 12 reps with band	Neck Disability Index (NDI)	0–50 (questionnaire)/points
					Follow-up	5.35 ± 4.9	2.12–7.47
			Intervention	24	Baseline	4.87 ± 3.42	1.47–6.34				~ 4–6 min/day	2 times/day	4 weeks	Stretching exercise - Lateral cervical stretch: 5 s hold, 5 reps per side - Posterior neck stretch - Arm/forearm stretch
					Follow-up	4.3 ± 2.67	1.15–5.45
			Control	24	Baseline	5.35 ± 4.9	2.12–7.47				~ 4–6 min/day	2 times/day	4 weeks	Minimal movement (control activity) - Deep breathing: 5 reps - Seated ankle pumps: 10 reps
					Follow-up	4.67 ± 3.96	1.71–6.38
Andersen 2008,Denmark	RCT	Muscle Activation (UT)	Intervention(SST)	18	Baseline	562 ± 56	510–614	44 ± 8	72 ± 15	165 ± 6	20 min per session	3 times per week	10 weeks	High-intensity strength training (5 dumbbell exercises)	Electromyography (EMG)	0 μV (no muscle activity) to higher EMG values (maximal)/Microvolts (μV)
					Follow-up	795 ± 72	723–867
			Intervention(GFT)	16	Baseline	464 ± 52	412–516	43 ± 7	71 ± 14	166 ± 5	20 min per session	3 times per week	10 weeks	General fitness training (leg bicycling)
					Follow-up	862 ± 78	723–1001
			Control	8	Baseline	365 ± 57	308–422	45 ± 9	70 ± 13	165 ± 6	1 h per week (education)	1 h per week (education only)	10 weeks	Health-promoting information (no physical training)
					Follow-up	570 ± 77	416–724
Blangsted 2008,Denmark	RCT	Absenteeism	Intervention(SRT)	180	Baseline	5 ± 1.2	4.76–5.24	f: 45.5 ± 10.4 , m: 47.3 ± 9.3	f: 68.9 ± 12.9, m: 82.5 ± 8.6	f: 168 ± 6, m: 181 ± 6	20 min	3 sessions per week	12 months	Resistance training targeting neck and shoulder muscles, dynamic and static exercises, rowing, and kayaking ergometer	Number of sick leave days taken	0–30 days (depending on number of sick leave days)
					Follow-up	5 ± 1.3	4.74–5.26
			Control	182	Baseline	5 ± 1.3	4.77–5.23	f: 43.9 ± 9.7, m: 46.3 ± 9.0	f: 70.4 ± 12.6, m: 81.6 ± 10.3	f: 169 ± 6,m: 181 ± 6	General health-promoting activities (no physical activity intervention)	No specific physical activity intervention	12 months	No physical activity; focused on improving work conditions and general health
					Follow-up	5 ± 1.4	4.75–5.25
Blangsted 2008,Denmark	RCT	Work Ability	Intervention(SRT)	180	Baseline	37.4 ± 3.7	36.67–38.13	f: 45.5 ± 10.4, m: 47.3 ± 9.3	f: 68.9 ± 12.9 , m: 82.5 ± 8.6	f: 168 ± 6, m: 181 ± 6	20 min	3 sessions per week	12 months	Resistance training targeting neck and shoulder muscles, dynamic and static exercises, rowing, and kayaking ergometer	Work ability (0–10 scale, WAI score from 6–42)	0–10 (higher = better) Points (WAI score from 6–42)
					Follow-up	36.9 ± 4.2	36.08–37.72
			Control	182	Baseline	36.7 ± 4	35.99–37.41	f: 43.9 ± 9.7 , m: 46.3 ± 9.0	f: 70.4 ± 12.6 , m: 81.6 ± 10.3	f: 169 ± 6, m: 181 ± 6	General health-promoting activities (no physical activity intervention)	No specific physical activity intervention	12 months	No physical activity; focused on improving work conditions and general health
					Follow-up	36.5 ± 4.8	35.64–37.36
Andersen 2012,Denmark	RCT	Work Disability	Intervention (3WS)	126	Baseline	13 ± 18	9.83–16.17	46 ± 10	75 ± 18	173 ± 10	20 min per session	3 times per week	20 weeks	High-intensity strength training for neck and shoulder muscles (with dumbbells)	0–9 (Pain intensity scale)	0 (no pain) to 9 (worst possible pain)
					Follow-up	4 ± 9	0–9.35
			Intervention(9WS)	106	Baseline	10 ± 16	6.93–13.07	45 ± 10	78 ± 15	175 ± 9	7 min per session	9 times per week	20 weeks	High-intensity strength training for neck and shoulder muscles (with dumbbells)
					Follow-up	8 ± 18	1.8–14.2
			Control	101	Baseline	11 ± 14	8.26–13.74	46 ± 10	80 ± 16	175 ± 9	No training	No training	20 weeks	No intervention
					Follow-up	8 ± 18	4.46–11.54
Andersen 2014,Denmark	RCT	Muscle Activation (UT)	Intervention(SST)	18	Baseline	244 ± 139	174.88–313.12	44 ± 8	72 ± 15	165 ± 6	20 min per session	3 times a week	10 weeks	High-intensity strength training targeting the neck and shoulder muscles. 8–12 repetitions maximum (RM) for each exercise	Microvolts (µV) for peak EMG and integrated EMG	µV (microvolts)
					Follow-up	320 ± 139	250.88–389.12
			Intervention(GFT)	16	Baseline	244 ± 139	169.93–318.07	44 ± 8	72 ± 15	165 ± 6	20 min per session	3 times a week	10 weeks	Leg cycling on a stationary ergometer with a relative load of 50–70% of their maximal oxygen uptake
					Follow-up	248 ± 139	173.93–322.07
			Control	20	Baseline	320 ± 120	127.79–360.21	45 ± 9	70 ± 11	167 ± 6	no intervention	None (healthy control group)	10 weeks	No intervention; used for comparison with symptomatic participants
					Follow-up	320 ± 120	128.79–361.21
Tunwattanapong 2016,Thailand	RCT	Quality of Life	Intervention	48	Baseline	57.05 ± 19.6	50.86–63.23	34.2 ± 9.0	not reported	not reported	10–15 min per session	Two sessions per day. 5 days per week for 4 weeks	4 weeks	Neck and shoulder stretching exercises (20–30 repetitions/session), including neck stretching, shoulder stretching, shoulder rolling, trunk stretching, and back extension exercises. 10–15 min per session, two times a day, five days a week for 4 weeks	SF-36 Physical & Mental Dimensions	Physical (0–100), Mental (0–100)
					Follow-up	67.65 ± 19.4	61.52–73.77
			Control	48	Baseline	64.7 ± 17.95	59.36–70.03	36.5 ± 8.7			No exercise, only educational brochure on ergonomics	Informative brochure on ergonomics	4 weeks	Informative brochure on ergonomics
					Follow-up	62.8 ± 18.95	57.17–68.42
Tunwattanapong 2016,Thailand	RCT	Neck Disability	Intervention	48	Baseline	28 ± 12.1	15.9–40	34.2 ± 9.0	not reported	not reported	10–15 min per session	Two sessions per day. 5 days per week for 4 weeks	4 weeks	Neck and shoulder stretching exercises (20–30 repetitions/session), including neck stretching, shoulder stretching, shoulder rolling, trunk stretching, and back extension exercises. 10–15 min per session, two times a day, five days a week for 4 weeks	Northwick Park Neck Pain Questionnaire (NPNPQ)	0–100
					Follow-up	21 ± 10.9	14.2–28.7
			Control	48	Baseline	28.2 ± 11.9	15.3–41.5	36.5 ± 8.7			No exercise, only educational brochure on ergonomics	Informative brochure on ergonomics	4 weeks	Informative brochure on ergonomics
					Follow-up	25.9 ± 13.9	17.3–34.5
Lee 2017,South Korea	RCT	Quality of Life	Intervention	11	Baseline	59.09 ± 21.8	43.49–74.69	27.09 ± 4.83	63.73 ± 15.62	168 ± 10.15	10–15 min per session	2 days/week	8 weeks	McKenzie neck exercise via mobile app	Short-Form 36 Health Survey (SF-36)	0–100 (higher score indicates better quality of life)
					Follow-up	65.61 ± 23.22	49–82.23
			Control	9	Baseline	63.41 ± 24.08	46.18–80.64	27.56 ± 4.67	73.11 ± 13.29	172.56 ± 7.09	Brochure	Single education session	8 weeks	Posture correction education (brochure)
					Follow-up	64.58 ± 19.26	50.8–78.36
Lee 2017,South Korea	RCT	Neck Disability	Intervention	11	Baseline	26.8 ± 9.68	21.08–32.52	27.09 ± 4.83	63.73 ± 15.62	168 ± 10.15	10–15 min per session	2 days/week	8 weeks	McKenzie neck exercise via mobile app	Neck Disability Index (NDI)	0% (no disability) to 100% (full disability)/ Percentage
					Follow-up	17.25 ± 8.34	12.32–22.18
			Control	9	Baseline	17.7 ± 9.2	11.69–23.71	27.56 ± 4.67	73.11 ± 13.29	172.56 ± 7.09	Brochure	Single education session	8 weeks	Posture correction education (brochure)
					Follow-up	15.92 ± 8.67	10.26–21.58
Tabanfar 2023,Iran	RCT	Neck Disability	Intervention	42	Baseline	6.64 ± 5.89	4.86–8.42	41.74 ± 7.55	65.94 ± 8.00	166.33 ± 9.38	15 min per session	5 sessions per week	12 weeks	Corrective exercises targeting neck muscles/Neck muscle strengthening exercises targeting flexors and scapulothoracic muscles (e.g., rhomboid, trapezius, pectoralis)	Neck Disability Index (NDI)	0–50 /Percentage
					Follow-up	3.4 ± 3.3	2.4–4.4
			Control	43	Baseline	6.35 ± 6.14	4.51–8.19	42.53 ± 8.04	66.00 ± 8.17	168.49 ± 7.99	No intervention	No intervention	12 weeks	No intervention
					Follow-up	6.77 ± 5.67	5.08–8.46

Summary of study characteristics, including study identification (author, year, country), study design, outcome type (quality of life, neck disability, muscle activation, absenteeism, work ability, and work disability), group allocation, sample size, assessment time points, outcome data (mean ± SD and 95% CI), participant characteristics (age, weight, height), intervention characteristics (dose, frequency, duration, and type), and outcome measurement details (assessment scale and units).

Effect sizes for study outcomes

Table 3 presents the meta-analysis results for outcomes with at least two studies along with the corresponding forest plots (Fig. 2). A total of eight outcomes were included in the meta-analysis, while additional outcomes were assessed descriptively. The number of effect sizes reported in Table 3 reflects the number of comparisons entered into each meta-analysis. For any given outcome domain (e.g., neck pain), each study contributed at most one effect size. For anatomically distinct but related outcomes (left and right shoulder pain), separate meta-analyses were conducted; these involve the same set of studies, creating statistical dependence that was not adjusted for due to the small number of studies.

Table 3.

Effect sizes and statistical synthesis of included studies.

Row	Outcomes	Frequency	Heterogeneity Status		Model Type	Effect Size	95% Confidence Interval		Significant Values		Status
			I2 value	Q value			Lower	Upper	z-value	p-value
1	Neck-Shoulder pain	5	39.007	6.558	Fixed	0.791	0.596	0.985	7.975	0.0001	Acceptance
2	Neck Pain	14	98.020	656.597	Random	1.364	0.642	2.087	3.700	0.0001	Acceptance
3	Shoulder Pain/Right	4	98.232	169.718	Random	1.722	0.546	2.898	2.871	0.004	Acceptance
4	Shoulder Pain/Left	4	98.314	177.986	Random	1.834	0.623	3.046	2.968	0.003	Acceptance
5	Shoulder Pain	3	96.448	56.3	Random	0.812	0.033	1.592	2.042	0.041	Acceptance
6	Quality of Life	2	0	0.727	Fixed	0.57	0.203	0.937	3.043	0.002	Acceptance
7	Neck Disability	5	29.042	5.637	Fixed	0.449	0.166	0.726	3.125	0.002	Acceptance
8	Muscle Activation	4	79.009	14.292	Random	0.763	-0.072	1.597	1.791	0.073	Rejection
9	Work Disability	1	Not pooled (single study)
10	Work Ability	1	Not pooled (single study)
11	Absenteeism	1	Not pooled (single study)

Standardized mean differences (SMD), 95% CI, heterogeneity statistics (I²), and model estimates derived from random-effects meta-analysis. For left and right shoulder pain, the same three studies contributed to both meta-analyses; thus, these effect sizes are statistically dependent. Pooled estimates should be interpreted as exploratory.

Fig. 2.

Forest plot of study effect sizes and pooled estimates. Forest plot showing SMD with 95% CI for included studies. Study weights were calculated using inverse-variance random-effects models. Negative effect sizes indicate outcomes favoring the micro-exercise intervention. Overall, pooled effects and heterogeneity statistics (I²) are presented.

Neck/shoulder Pain: Neck/shoulder pain was reduced in the intervention groups compared with controls (4 studies; g = –0.79, 95% CI –0.98 to –0.60; I² = 39%).

Neck Pain: Neck pain revealed a large pooled effect (11 studies; g = –1.36, 95% CI –2.09 to –0.64; I² = 98%; large effect by Cohen’s criteria [≥ 0.8]).

Shoulder Pain: Shoulder Pain (combined) was reduced in 2 studies that reported a single shoulder pain score (g = – 0.81, 95% CI – 1.59 to – 0.03; I² = 96%). When stratified by side: right and left shoulder pain was reported in 3 studies (all of which reported both sides), yielding a pooled effect of g = – 1.72 (95% CI – 2.90 to – 0.55; I² = 98% and g = – 1.83 (95% CI – 3.05 to – 0.62; I² = 98%, respectively. The difference in the number of studies between overall and stratified analyses reflects the availability of side-specific data.

Quality of Life: Overall quality of life improved modestly (2 studies; g = – 0.57, 95% CI – 0.94 to – 0.20; I² = 0%).

Neck Disability: Similarly, neck disability improved (4 studies; g = – 0.45, 95% CI – 0.73 to – 0.17; I² = 29%).

Muscle activity

The pooled effect size was moderate to large in 2 studies (g = – 0.76) but not statistically significant (95% CI – 1.60 to 0.08; I² = 79%; p = 0.076). The confidence interval crosses zero, indicating that the true effect could range from a large beneficial effect (negative) to a trivial effect (close to zero) or even a small harmful effect (positive). The non-significance is likely due to the small number of studies (n = 2), high heterogeneity, and small sample sizes, which result in a wide confidence interval. Therefore, evidence for improved muscle activity from micro-exercises remains inconclusive.

Work-related outcomes

Work-related outcomes (work ability, work disability, and absenteeism), each reported by a single study. Therefore, meta-analysis was not possible and were presented descriptively.

Blangsted et al. (2008) assessed work ability using a modified Work Ability Index (WAI), scale 6–42, higher = better. Absenteeism measured using personnel file records. At baseline, mean WAI scores were close to 90% of the maximum across all groups (mean ~ 37/42), and mean annual absenteeism was 5.3 days. Neither outcome changed significantly following one year of intervention: WAI changes were non-significant versus intervention groups (p = 0.31), (p = 0.42), and absenteeism presented no significant change across groups (p = 0.45). The authors concluded that the high baseline WAI and low baseline absenteeism left insufficient room for measurable improvement²².

Andersen et al. (2012) measured work disability using the DASH work module (scale 0–100, higher = worse). In the Intention to Treat analysis, significant reductions in DASH scores compared to the reference group were found in two groups (− 4 points, 95% CI − 8 to − 1; p < 0.05) and (− 7 points, 95% CI − 10 to − 3; p < 0.01), but not in one group (− 2 points, 95% CI − 6 to 1; p > 0.05)27.

Due to the very high heterogeneity (I² ≥ 96% for neck pain, shoulder pain, and left/right shoulder pain) and the inability to perform subgroup analyses, these pooled effect sizes should not be interpreted as consistent or generalizable estimates. They primarily reflect the average of highly variable study results.

Risk of bias within studies

Risk of bias was assessed using the Cochrane Risk of Bias tool (RoB 1.0). Table 4 presents the Risk-of-bias assessments. Generally, 10 studies (52.63%) were rated as low risk of bias, 9 studies (47.37%) as some concerns/unclear, and no study (0%) as high risk. The most frequent issues were related to blinding of participants/personnel, blinding of outcome assessors, and incomplete outcome data.

Table 4.

Risk of bias assessment of included studies.

Author/ Year	Country	Design	Seq. Gen	Alloc. Conceal	Blinding/Part. & Pers	Blinding/Assessors	Incomp. Outcome Data	Select. Report	Other Bias	Risk of Bias
Kietrys 2007	United States	RCT	Low	Unclear	High	Unclear	Low	Low	Unclear	Moderate
Andersen 2008	Denmark	RCT	Low	Unclear	Unclear	Low	Low	Low	Unclear	Low
Andersen 2008	Denmark	RCT	Unclear	Unclear	High	Low	Low	Low	Low	Moderate
Andersen 2008	Denmark	RCT	Low	Unclear	Low	Low	Low	Low	Low	Low
Blangsted 2008	Denmark	RCT	Low	Unclear	High	Low	Low	Low	Low	Moderate
Andersen 2010	Denmark	RCT	Low	Unclear	Low	Low	Low	Low	Low	Moderate
Andersen 2011	Denmark	RCT	Low	Low	Low	Low	Low	Low	Low	Low
Andersen 2012	Denmark	RCT	Low	Unclear	Unclear	Low	Low	Low	Unclear	Moderate
Lidegaard 2013	Denmark	RCT	Low	Low	Low	Low	Low	Low	Low	Low
Gram 2014	Denmark	Cluster RCT	Low	Unclear	High	High	Unclear	Low	Low	Moderate
Andersen 2014	Denmark	RCT	Low	Unclear	Low	Low	Low	Low	Low	Low
Andersen 2014	Denmark	RCT	Low	Low	Unclear	Low	Low	Low	Low	Low
Tunwattanapong 2016	Thailand	RCT	Low	Low	High	Low	Low	Unclear	Low	Low
Lee 2017	South Korea	RCT	Low	Unclear	High	High	Low	Low	Low	Moderate
Shariat 2018	Malaysia	RCT	Low	Low	Unclear	High	Low	Low	Low	Low
Anan 2021	Japan	RCT	Low	Unclear	High	High	Low	High	Moderate	Moderate
Tabanfar 2022	Iran	RCT	Unclear	Unclear	High	Unclear	Low	Low	Low	Moderate
Tabanfar 2023	Iran	RCT	Low	Unclear	High	Low	Low	Low	Low	Low
Michaelidis 2025	Greece	RCT	Low	Low	High	Low	Unclear	Low	Low	Low

Sequence Generation → Seq. Gen., Allocation Concealment → Alloc. Conceal., Blinding of , Participants and Personnel → Blinding (Part. & Pers.), Blinding Outcome Assessors → Blinding (Assessors), Incomplete Outcome Data → Incomp. Outcome Data, Selective Reporting → Select. Report., Other Sources of Bias → Other Bias, Risk of Bias Category → Risk of Bias.

Risk of bias evaluation of included studies based on the Cochrane Risk of Bias tool (ROB-1), including domains such as randomization, allocation concealment, blinding, incomplete outcome data, and selective reporting.

Domain-level findings: Blinding of participants and personnel was the most frequently problematic domain, with 12 studies (63.2%) judged as unclear or high risk because exercise interventions cannot easily blind participants or trainers. Blinding of outcome assessors was judged as unclear or high risk in 8 studies (42.1%) due to lack of explicit reporting. Incomplete outcome data (attrition) was problematic in 6 studies (31.6%), where dropout rates exceeded 20% without adequate handling (e.g., intention-to-treat analysis).

Potential influence on outcomes: Lack of blinding is particularly likely to bias self-reported outcomes (e.g., pain, quality of life, work ability) toward larger perceived benefits in the intervention group. In contrast, objective outcomes (e.g., muscle activity measured by electromyography) are less susceptible to this bias. Incomplete outcome data may overestimate effects if dropouts are systematically different (e.g., participants with no improvement more likely to withdraw). The overall risk of bias suggests that pooled effect sizes for pain and disability may be modestly overestimated, whereas findings for muscle activity and work-related outcomes (based on fewer studies with unclear bias) should be interpreted more cautiously.

Risk of bias across studies

Funnel plot inspection revealed asymmetry (Fig. 3), with smaller studies reporting larger effects. Egger’s test (p = 0.028) and Begg’s test (p < 0.001) confirmed small-study effects and potential publication bias (Fig. 4). Trim-and-fill correction was not performed. Based on the high heterogeneity observed for neck pain (I² = 98.0%) and shoulder pain outcomes (I² = 96.4%–98.3%), the pooled estimates should be viewed in the context of the high between-study variability. Although the pooled effects were statistically significant for neck pain (g = 1.36) and shoulder pain (g = 0.81–1.83), the substantial between-study variability suggests that the true intervention effects may differ across studies and could be smaller than the pooled estimates indicate  .

Fig. 3.

Funnel plot for assessment of publication bias. Funnel plot showing effect sizes plotted against their standard errors to assess potential publication bias and small-study effects. The vertical line represents the pooled effect estimate.

Fig. 4.

Begg’s and Egger’s tests for publication bias. Statistical assessment of publication bias using Begg’s rank correlation test and Egger’s regression asymmetry test.

Certainty of evidence

The certainty of evidence ranged from moderate to very low across outcomes. Evidence for neck/shoulder pain and neck disability was rated as moderate certainty. Evidence for neck pain, shoulder pain, muscle activity, and work-related outcomes was rated as low to very low certainty due to very high heterogeneity (I² ≥ 96% for neck/shoulder pain), small numbers of studies, and potential publication bias.

Although the pooled effect sizes for neck pain (g = – 1.36), right shoulder pain (g = – 1.72), and left shoulder pain (g = – 1.83) are large by Cohen’s criteria (≥ 0.8), the certainty is low because these estimates are derived from studies with substantial unexplained heterogeneity, risk of bias (especially lack of blinding), and evidence of publication bias. Table 6 displays the detailed GRADE assessments.

Discussion

Principal findings

This systematic review and meta-analysis suggest that workplace micro-exercises (2–20 min, structured therapeutic movements) may reduce neck/shoulder pain in sedentary employees. Pooled effect sizes ranged from moderate (g = – 0.79 for combined neck/shoulder pain) to large (g = – 1.36 for neck pain). However, GRADE is low to very low for most pain outcomes, and the large effects should not be misinterpreted as high-quality evidence, as they are likely inflated by methodological limitations (lack of blinding, high heterogeneity) and probable publication bias. These estimates are best viewed as hypothesis-generating, and future research is likely to revise them.

To assess whether the observed reductions are clinically meaningful, we compared the pooled effect sizes with established benchmarks. For neck pain, the pooled standardized mean difference was large (SMD = – 1.36; Table 3). Based on the weighted pooled standard deviation derived from the included neck pain studies using comparable 0–10 pain scales (SD ≈ 1.8–2.2), this corresponds to an estimated absolute reduction of approximately 2.5–3.0 points on a 10-point Visual Analogue Scale (VAS). As the minimal clinically important difference (MCID) for neck pain is commonly reported as 1.0–2.0 points, the observed improvement appears clinically meaningful^39,40.

For shoulder pain, pooled effects ranged from SMD = – 0.81 to – 1.83 (Table 3), indicating moderate-to-large treatment benefits. Using similar pain-scale assumptions, these effects may correspond to an approximate reduction of 1.5–4.0 points, suggesting potential clinical relevance.

For neck disability, the pooled effect was SMD = – 0.45 (Table 3). Based on the standard deviations reported across the included studies, this likely reflects an improvement of approximately 4–5 points on the Neck Disability Index (NDI), which approaches commonly reported MCID thresholds of 5–7 points. Therefore, while statistical uncertainty exists (low GRADE certainty), the magnitude of pain reduction is likely clinically noticeable for many sedentary staff. These conversions are indirect, as individual patient data were not available, and MCID values vary by population and baseline severity.

Comparison with previous literature

Our findings align with earlier systematic reviews of workplace exercise interventions, which reported beneficial effects on musculoskeletal pain among office workers^15–17. However, those reviews did not specifically focus on micro-exercises as defined in the present study. This review extends the evidence by quantifying effects specifically for brief, low-burden interventions that can be integrated into the workday without special equipment. The magnitude of pain reduction observed is comparable to or larger than that reported in reviews of longer-duration workplace exercise interventions, suggesting that even short bouts may be clinically meaningful^{18,30,33,34,37}.

Mechanisms and interpretation

The pain-reducing effects of micro-exercises may be explained by neuromuscular relaxation, improved motor control, and increased local blood flow to the neck/shoulder musculature. Repetitive low-dose contractions may reduce muscle tension and modulate pain perception. Two included studies reported favorable electromyographic changes, such as increased relaxation time and improved scapular muscle balance^27,29. However, the pooled effect for muscle activity was not statistically significant, likely due to the small number of studies, short intervention durations, and methodological variability.

In contrast, work-related outcomes such as absenteeism and work ability are governed more strongly by psychosocial determinants including job stress, workload, organizational culture, and mental health than by physical pain levels alone^41–43. Consequently, even substantial pain relief may be insufficient to improve work performance or reduce sickness absence if workplace barriers remain. This is consistent with previous research showing that stretching combined with ergonomic modifications reduced discomfort but produced less pronounced improvements in work-related outcome³⁵. particularly among employees with chronic pain. Micro-exercises may therefore be most effective as part of a broader workplace strategy that combines exercise with ergonomic adjustment, stress management, and psychosocial support.

Another crucial factor is feasibility and adherence in real-world workplace settings. While the brevity and simplicity of micro-exercises are advantageous, ensuring consistent participation can be challenging due to time constraints, lack of supervision, and competing workplace priorities. Adherence may be compromised when exercise performance relies solely on individual motivation during busy work periods. Gram et al. (2014) highlighted that both supervised and minimally supervised workplace training significantly reduced neck pain and headaches, reinforcing the value of on-site implementation²⁸. Future research should examine extended-duration or progressive resistance protocols, employ objective electromyography-based assessments to clarify the conditions under which micro-exercises influence muscle activity and work-related outcomes, and explore strategies to better integrate micro-exercises into the workday⁴⁴.

Heterogeneity

Substantial heterogeneity was observed for several pain outcomes (I² ≥ 96%). Based on narrative assessment, likely contributors include variation in intervention characteristics (duration, frequency, supervision, and exercise type), participant factors (age, baseline symptoms, occupational demands), outcome measurement methods, and small-study effects. Differences in adherence and compliance monitoring may also have contributed. Formal subgroup analyses and meta-regression were not feasible because most outcomes included too few studies and potential moderators were inconsistently reported. Accordingly, the pooled estimates should be interpreted as average effects across highly variable study conditions rather than precise universal estimates.

Work-related outcomes

Work-related outcomes showed mixed results. Work disability (DASH) demonstrated significant improvement in two of three intervention groups in one study, suggesting a possible benefit for work disability specifically. However, work ability and absenteeism showed no significant change. Possible explanations for the null findings include short intervention duration, low baseline absenteeism, reliance on self-reported measures, and the multifactorial nature of occupational outcomes (stress, job satisfaction, mental health)^42–44. Clinicians and employers should therefore not assume that pain reduction will automatically translate into improved productivity or reduced sick leave. Broader organizational strategies are likely required.

Strength and limitations

Strengths of this review include the exclusive inclusion of RCTs (n = 19), adherence to PRISMA guidance, a comprehensive multi-database search, and evaluation of both clinical and occupational outcomes.

Several limitations should be acknowledged. First, most studies reported only short-term outcomes, leaving long-term effectiveness uncertain. Second, reliance on self-reported measures (pain, disability, quality of life, and work outcomes) increases susceptibility to response bias, compounded by the inability to blind participants or trainers, which may further inflate self-reported effect sizes. Third, work-related outcomes (work ability, absenteeism, and work disability) were each reported by a single study, precluding meta-analysis for these endpoints.

Fourth, substantial heterogeneity was observed across pooled analyses, reflecting variability in intervention characteristics, participant populations, outcome measures, and the broad operational definition of micro-exercises. Formal subgroup analyses or meta-regression were not feasible given the small number of studies per comparison, leaving heterogeneity sources unexplained and limiting the precision and generalizability of pooled estimates. Where studies contributed multiple effect sizes to the same comparison (e.g., left and right shoulder pain from a single cohort), statistical dependence could not be adjusted for, meaning some confidence intervals may be narrower than a properly modelled analysis would yield. Fifth, smaller trials tended to report larger effects, suggesting potential overestimation of the true intervention effect; formal bias-adjustment methods could not be applied due to insufficient study numbers. Together, these constraints resulted in GRADE certainty ratings that were frequently downgraded, with work-related outcomes supported by very low-certainty evidence.

Sixth, eleven of the nineteen included studies were conducted in Denmark, which may limit generalizability to other occupational settings. This predominance reflects Scandinavia’s strong tradition of publicly funded occupational health research, not systematic search bias. However, Danish workplace settings feature strong trade union involvement, employer-supported on-site exercise interventions, legislative frameworks mandating risk assessment, and high organizational support for active breaks, factors that may facilitate recruitment, adherence, and implementation fidelity in ways difficult to replicate elsewhere. Future trials in more diverse occupational and cultural settings are needed.

Implications for practice and research

Micro-exercises offer a low-cost, time-efficient strategy that can be integrated into the workday with minimal disruption. Employers may consider implementing brief (2–10 min), daily, structured resistance or stretching routines targeting the neck/shoulder muscles, ideally with initial supervision to ensure correct technique. However, evidence for work-related benefits remains insufficient to recommend micro-exercises solely for reducing absenteeism or improving work ability. A multidisciplinary approach combining physical exercise with ergonomic adjustments and psychosocial interventions may provide a more comprehensive solution.

Future trials should standardize intervention reporting (duration, intensity, supervision, adherence), use consistent outcome measures, include longer follow-ups (≥ 6 months), and employ objective work outcome metrics (e.g., register-based absenteeism). Multi-country trials are needed to assess generalizability beyond Scandinavian occupational health contexts. Larger, adequately powered trials will enable subgroup analyses or meta-regression to identify sources of heterogeneity.

Conclusion

Workplace micro-exercises may reduce neck/shoulder pain in desk-based employees and may improve neck-related function. However, confidence in these findings is limited by substantial heterogeneity, possible publication bias, and generally low certainty of evidence. Current evidence is insufficient to conclude that micro-exercises improve work ability or reduce absenteeism. Larger, high-quality RCTs with standardized intervention protocols, longer follow-up, and consistent outcome reporting are needed.

Supplementary Information

Supplementary Information 1. ^{(362.8KB, pdf)}

Supplementary Information 2. ^{(382.6KB, pdf)}

Author contributions

All authors [Zohreh Yaghoubitajani], [Mehdi Gheitasi], [Mohammad Bayattork], [Karina Glies Vincents Seeberg], [Lars Louis Andersen] contributed to the study conception and design. [Zohreh Yaghoubitajani], [Mohammad Bayattork] performed material preparation, data collection, and analysis. The first draft of the manuscript was written by [Zohreh Yaghoubitajani], [Karina Glies Vincents Seeberg], and all authors commented on previous versions of the manuscript and revised it critically for important intellectual content. All authors read and approved the version for publication. The corresponding author [Mehdi Gheitasi] agreed to be accountable for all aspects of the work ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Data availability

The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Competing interests

The authors declare no competing interests.