This study employed a cross-sectional correlational design. Quantitative data were collected at a single time point from each participant, comprising HRV assessment, self-reported perceived stress, a musculoskeletal pain questionnaire, postural assessment, and a sedentary behavior inventory. Data collection was conducted across five corporate locations in Hong Kong between December 8, 2025 and February 27, 2026, spanning 12 weeks of active data collection. Data collection was paused for two weeks over the Christmas and New Year period (December 22, 2025–January 4, 2026) to accommodate reduced corporate staffing, with active recruitment resuming on January 5, 2026. All data collection was completed prior to data analysis, which commenced on March 3, 2026.

All statistical analyses were performed using SPSS Version 27.0 (IBM Corp., Armonk, NY, USA) [39]. Statistical significance was set at α = 0.05 for all primary analyses (two-tailed), and α = 0.10 for exploratory moderation analyses. Effect sizes are reported as r for correlations and f² for regression models, with the following interpretation thresholds: r = 0.10 (small), 0.30 (medium), 0.50 (large); f² = 0.02 (small), 0.15 (medium), 0.35 (large) [40]. p-value reporting follows the convention: p < 0.05, p < 0.01, p < 0.001, with exact values reported where 0.001 < p < 0.10.

The analytical sequence was as follows:

1.

Descriptive statistics for all continuous variables, with normality tested via Shapiro-Wilk.

Frequency-domain HRV variables were natural-log transformed prior to analysis to correct for positive skew, consistent with standard HRV analytic practice [16].

This study was conducted in accordance with the Declaration of Helsinki. Ethics approval was obtained from the Ethics Committee of the Hong Kong Association of Psychology prior to any participant recruitment or data collection (Approval No. HKAP-20251015-001, approved October 15, 2025). Participant recruitment commenced on November 17, 2025, following a two-week pilot protocol testing phase (November 3–14, 2025). All participants provided written informed consent prior to participation, including consent for publication of de-identified aggregated data. Participant confidentiality was maintained throughout, with all identifying information replaced by anonymized participant codes. No animal subjects were involved in this research.

Descriptive statistics for all primary study variables are presented in Table 3. Shapiro-Wilk tests confirmed approximate normality for all continuous variables of interest after log-transformation of frequency-domain HRV measures (all W > 0.96, p > 0.05).

The mean PSS-14 score of 22.6 exceeded the high-stress threshold of 20, consistent with the near-high stress baseline documented in prior Hong Kong organizational research [12]. Mean SDNN (72.4 ms) and normalized coherence (58.4%) fell within but toward the lower end of reference ranges, suggesting subclinical autonomic dysregulation consistent with chronic occupational stress. The mean FHP index of 4.2 cm substantially exceeded the normal threshold of 2.5 cm, indicating widespread postural dysfunction in this sample.

Pearson correlations between HRV parameters, PSS-14, and pain outcomes are presented in Table 4. All reported correlations remained statistically significant after Benjamini-Hochberg FDR correction (q < 0.05).

Three of the four HRV parameters—SDNN, RMSSD, and normalized coherence—demonstrated significant negative correlations with pain intensity, multi-site pain burden, and pain impact on lifestyle, largely supporting H1. MHRR was significantly correlated with pain intensity and multi-site pain burden, but its correlation with pain impact did not reach significance (r = –0.16, p ≈ 0.10). Normalized coherence showed the strongest associations across pain outcomes (r = –0.29 to –0.38), followed by RMSSD (r = –0.26 to –0.33). PSS-14 showed stronger correlations with pain outcomes (r = 0.41 to 0.46) than individual HRV parameters, although the two predictor sets were only moderately correlated with each other (r = –0.19 to –0.34), suggesting they capture partially independent dimensions of WMSD risk.

Site-specific correlations between normalized coherence and pain are presented in Table 5. After Benjamini-Hochberg FDR correction, correlations for neck, upper back, shoulder, lower back, head, and wrist remained significant (q < 0.05); knee, hip, and chest/abdomen correlations did not survive correction.

HRV coherence showed the strongest and most consistent associations with upper-body pain sites—neck, shoulder, and upper back—which are the regions most directly implicated in sedentary postural dysfunction. This pattern is consistent with prior findings in Hong Kong workers [11] and supports the mechanistic interpretation that autonomic dysregulation and postural loading converge on upper-body musculoskeletal structures in sedentary office workers.

Hierarchical regression results are presented in Table 6, with the final model standardized coefficients in Table 7.

The final model explained 44% of variance in pain intensity [R² = 0.44, F(10, 94) = 7.38, p < 0.001, f² = 0.79, large effect]. HRV parameters as a block explained an additional 11% of variance beyond PSS-14 alone (ΔR² = 0.11, p < 0.001), supporting H2. Normalized coherence was the only individually significant HRV predictor in the final model (β = –0.29, 95% CI [–0.45, –0.13], p < 0.001), with PSS-14 (β = 0.31), FHP index (β = 0.24), and kyphosis angle (β = 0.19) also emerging as significant independent predictors.

Diagnostic checks confirmed that all standard regression assumptions were adequately met:

Multicollinearity: All VIF values for the final model were below the conventional threshold of 5.0 (range: 1.18–2.91; tolerance range: 0.34–0.85), indicating that the moderate intercorrelation among HRV parameters did not produce problematic multicollinearity.

These diagnostic results support the validity of the regression-based inferences reported.

A statistical mediation analysis using PROCESS Model 4 [41] with 5,000 bootstrap iterations examined whether FHP index mediated the association between normalized coherence and neck/shoulder pain intensity. The conceptual model is presented in Figure 1.

The analysis yielded the following results:

Direct effect of coherence on neck/shoulder pain (path c’): β = –0.19, p = 0.028.

These results support H3: FHP index partially mediated the association between HRV coherence and neck/shoulder pain, accounting for approximately 39% of the total effect. Workers with lower autonomic coherence showed greater FHP displacement, which in turn predicted greater neck/shoulder pain intensity. A parallel mediation model with thoracic kyphosis angle as mediator yielded a significant indirect effect (β = –0.09, 95% CI [–0.17, –0.02]), suggesting both postural variables independently carry part of the coherence–pain pathway.

We emphasize that, given the cross-sectional design, these results constitute statistical mediation rather than confirmation of causal mechanism. Maxwell and Cole [42] have shown that cross-sectional mediation estimates can substantially diverge from longitudinal estimates of true causal mediation. The present findings are best interpreted as evidence consistent with—but not confirmation of—a coherence → posture → pain causal pathway.

A moderation analysis using PROCESS Model 1 examined whether daily seated hours moderated the association between normalized coherence and pain intensity. The coherence × sedentary hours interaction term was statistically significant (β = –0.22, p = 0.008), supporting H4. Simple slopes are presented in Table 8.

The protective association between higher HRV coherence and lower musculoskeletal pain was substantially stronger among workers with the highest sedentary exposure. Among workers seated more than 8 hours daily—the majority of this sample (51.4%)—each standard deviation increase in normalized coherence was associated with a 0.87-point reduction in pain intensity on the 0–10 scale. This interaction is consistent with the interpretation that HRV coherence may function as a physiological buffer against the musculoskeletal consequences of prolonged sedentary work, although this causal framing requires longitudinal confirmation.

To provide clinically interpretable findings, participants were divided into tertiles based on normalized coherence scores: low coherence (< 47%), moderate coherence (47–68%), and high coherence (> 68%). Group comparisons are presented in Table 9.

Workers in the low coherence tertile reported pain intensity scores averaging 7.1 of 10—consistent with moderate-to-severe pain—compared to 4.4 of 10 in the high coherence group, a difference of 2.7 points exceeding the established minimal clinically important difference of 2.0 points for pain intensity scales [43]. The low coherence group also showed the greatest postural dysfunction (FHP = 5.4 cm vs. 3.1 cm) and longest daily seated exposure (9.4 vs. 7.7 hours), illustrating the convergent risk profile of autonomic dysregulation, postural dysfunction, and sedentary behavior in the highest-risk workers.

The primary finding of this study—that HRV parameters, particularly normalized coherence, are significantly and independently associated with musculoskeletal pain outcomes in Hong Kong office workers—provides empirical evidence consistent with a link between autonomic nervous system function and WMSD-relevant pain profiles in this population. The incremental predictive validity of HRV beyond PSS-14 (ΔR² = 0.11, p < 0.001) is particularly noteworthy, as it suggests that HRV may capture physiological dimensions of WMSD risk that self-report instruments alone do not fully detect.

These findings are consistent with the broader international literature on autonomic dysregulation and chronic pain, including studies in fibromyalgia [17], chronic neck pain, and chronic low back pain populations, where reduced HRV has been documented across diverse clinical samples. The present study extends this literature by demonstrating analogous associations in a non-clinical occupational sample, suggesting that subclinical autonomic dysregulation may be detectable in working populations before progression to chronic pain syndromes. They also extend the foundational occupational HRV work of Low and McCraty [16], who established HRV as a valid objective complement to self-reported stress measures in Hong Kong professionals, into the specific domain of musculoskeletal health.

The emergence of normalized coherence as the strongest individual HRV predictor (β = –0.29) is consistent with the neurovisceral integration model [18], which positions coherence-related synchronization as a sensitive index of self-regulatory capacity. Workers with lower coherence may have reduced capacity to modulate both emotional stress responses and nociceptive processing, although the cross-sectional design of the present study cannot directly test this mechanistic interpretation.

The mediation findings provide statistical evidence consistent with a pathway in which autonomic regulation, postural function, and musculoskeletal pain are interrelated in sedentary office workers. The significant indirect association of HRV coherence with neck/shoulder pain via FHP (β = –0.12, 95% CI [–0.21, –0.04]) is consistent with the interpretation that chronic autonomic dysregulation—characterized by sustained sympathetic activation and elevated muscle tension—may contribute to postural dysfunction, which in turn may generate mechanical loading on cervical and upper-thoracic structures. We note that the cross-sectional nature of the data precludes definitive causal inference [42], and alternative directional interpretations (e.g., that postural dysfunction itself contributes to autonomic dysregulation) cannot be ruled out without longitudinal data.

This statistical pattern is theoretically consistent with the trapezius muscle tension model proposed by Lundberg [10], in which psychosocial stress activates low-threshold motor units in the upper trapezius through cortical and subcortical pathways, generating sustained low-level muscle contraction that, over time, may produce tissue damage and pain. The present findings raise the possibility that HRV coherence could serve as an upstream marker of this process—potentially detectable before structural tissue damage occurs—although this hypothesis requires longitudinal investigation.

The moderation finding—that the HRV-pain association was substantially stronger among workers with the highest sedentary exposure (> 8 hours/day)—has potential practical implications. It is consistent with the interpretation that HRV coherence may function as a physiological buffer against the musculoskeletal correlates of sedentary work, with the strongest statistical association observed precisely among the most sedentary workers.

This finding aligns with prior identification of Hong Kong’s sedentary work culture as a significant risk factor for musculoskeletal pain [11], and extends it by suggesting that autonomic regulation may play an important moderating role in this risk landscape. The finding that 51.4% of participants reported more than 8 hours of daily seated work—and that this group showed the strongest HRV-pain associations—underscores the relevance of addressing sedentary behavior as part of comprehensive WMSD prevention in office-based professional populations, a concern that extends beyond Hong Kong to office workers in Western, Asian, and global contexts where sedentary work is increasingly the norm.

The present findings carry particular significance in the cultural context of Hong Kong’s professional workforce. As discussed in Introduction, multiple cultural dynamics—including collectivist values, face-maintenance, endurance norms, somatization, and reduced help-seeking—may systematically bias self-report measures of stress and pain in this population. The mean PSS-14 score of 22.6 in the present sample, while indicating high stress, may itself underestimate true physiological stress burden if cultural suppression effects are operative.

Objective physiological measurement may therefore provide a useful complement to self-report instruments in Asian professional populations. The finding that HRV explained 11% additional variance in pain outcomes beyond PSS-14 is consistent with this interpretation, although it does not directly demonstrate suppression bias and could reflect simply that HRV and PSS-14 capture partially distinct constructs. Future research employing experimental manipulation of disclosure conditions or comparison of objective and self-report measures across cultural samples would be needed to test the cultural suppression hypothesis directly.

The present findings suggest several considerations for WMSD risk assessment and prevention in sedentary professional settings. We emphasize that these are tentative implications, given the cross-sectional design, and should be regarded as hypotheses for future testing rather than evidence-based recommendations for clinical practice.

For occupational health screening. HRV assessment may provide a feasible, non-invasive, and time-efficient (approximately 7 minutes per assessment) objective complement to standard self-report screening tools. Workers identified with consistently low coherence scores may represent a subgroup that could be prioritized for further ergonomic and psychophysiological assessment. We frame this in general terms of HRV assessment rather than in reference to any specific commercial device; methodological details for the present study are reported in Instruments and measures.

For ergonomic intervention design. The mediation findings raise the possibility that interventions targeting postural dysfunction—including workstation ergonomic adjustment, postural awareness training, and movement break protocols—may yield additional benefit for workers with low HRV coherence, by potentially interrupting the autonomic–postural–pain pathway suggested by these data.

For psychophysiological intervention. HRV biofeedback, which has shown effectiveness in improving coherence and reducing perceived stress in prior Hong Kong professional samples [38, 44], represents a logical candidate intervention for workers identified as higher-risk through HRV screening. The present findings are consistent with—but do not establish—the possibility that coherence improvements achievable through HRV biofeedback training may translate into musculoskeletal pain reduction. Randomized controlled intervention trials are required to test this directly.

For organizational policy. The finding that workers exceeding 8 hours of daily seated work showed the strongest HRV-pain associations is consistent with the rationale for organizational policies promoting regular movement breaks, height-adjustable workstations, and integration of physical activity into the workday as structural WMSD prevention measures. These policies are supported by independent ergonomic evidence and need not depend on the present findings for their justification.

Several limitations warrant acknowledgement.

Cross-sectional design. The cross-sectional design precludes causal inference. The mediation and moderation analyses presented should be interpreted as describing patterns of statistical association rather than confirmed causal pathways [42]. Longitudinal or experimental designs are required to establish directionality.

Selection bias and sampling. Participants were recruited via purposive convenience sampling across five corporate sites, drawn from the financial, legal, healthcare, technology sectors, and other professionals. This approach was feasible within Hong Kong’s corporate access constraints but introduces potential selection bias: participants who self-referred to the study may differ systematically from non-volunteers in stress level, pain experience, or health-consciousness. The sample does not represent Hong Kong’s full occupational landscape, including manual labor, service, manufacturing, and small-enterprise sectors. Generalizability beyond white-collar sedentary occupations is therefore limited.

HRV measurement device. HRV was measured via PPG (emWave Pro Plus), which is a non-gold-standard method relative to clinical-grade ECG. While PPG-derived time-domain HRV measures show strong agreement with ECG under controlled resting conditions [32], PPG cannot fully substitute for ECG, particularly for fine-grained spectral analyses. Future research should validate the present findings using ECG-based HRV measurement.

Self-report bias in pain and sedentary behavior measures. Pain intensity, multi-site pain burden, pain impact, and daily sedentary hours were all assessed via self-report, and are therefore subject to recall bias, social desirability, and the cultural reporting biases discussed in Introduction. Objective measurement of sedentary behavior (e.g., accelerometry) and of pain-related disability (e.g., functional task performance) would strengthen future studies.

Pain instrument selection. The Low and Ho musculoskeletal pain questionnaire was selected for direct comparability with prior Hong Kong research [11] and for its inclusion of pain intensity and functional impact measures not available in the standard NMQ. However, the NMQ remains the internationally established standard for WMSD prevalence assessment, and its use in future research would enhance comparability with the broader international WMSD literature. We recommend that future studies in this area employ the NMQ in parallel with intensity-based measures.

Postural assessment precision. While inter-rater reliability for the FHP and kyphosis measures was high (ICC > 0.85), flexicurve and clinical anthropometric methods lack the precision of three-dimensional motion capture systems. Future research employing motion capture or inertial measurement units could provide more refined postural data.

Cultural framing. While we discuss cultural dynamics that may shape stress and pain reporting in Hong Kong, the present study did not directly measure these cultural variables. Cultural interpretations should be regarded as theoretically motivated context rather than empirically tested mechanisms within this study.

This study provides cross-sectional empirical evidence that HRV parameters—particularly normalized coherence—are significantly associated with musculoskeletal pain outcomes in Hong Kong office workers, and explain additional variance in pain intensity (ΔR² = 0.11) beyond self-reported perceived stress. The mediation analysis is consistent with a pathway involving postural dysfunction, and the moderation analysis suggests that HRV-pain associations are strongest among the most sedentary workers.

These findings should be interpreted within the constraints of a cross-sectional, single-site (Hong Kong), white-collar sample. The data support HRV as a promising objective complement to self-reported stress in WMSD risk research, but do not establish HRV as a clinically validated screening tool. Definitive claims regarding clinical application require longitudinal validation and intervention trial evidence, which are beyond the scope of this study.

Future research priorities include: longitudinal cohort studies tracking HRV trajectories and WMSD onset over 12–24 months to establish temporal precedence and test causal pathways suggested by the present cross-sectional patterns; randomized intervention trials testing whether HRV biofeedback-mediated improvements in coherence produce measurable reductions in musculoskeletal pain burden, building on the clinical case evidence in Hong Kong professional samples [38, 44]; multi-modal wearable studies combining HRV monitoring with inertial measurement units and surface electromyography to capture real-time autonomic–postural–muscular dynamics during the workday; cross-cultural comparative studies examining whether the HRV-WMSD association documented in Hong Kong generalizes to other Asian professional contexts (e.g., Singapore, Japan, South Korea) and to Western professional populations, to test whether the patterns observed here are culturally specific or reflect more general autonomic–musculoskeletal dynamics in sedentary work; dose-response investigations examining the relationship between magnitude of HRV coherence improvement and magnitude of musculoskeletal pain reduction, to inform optimal HRV biofeedback intervention protocols for WMSD prevention; and replication using the NMQ as the primary musculoskeletal pain instrument, to enhance comparability with the broader international WMSD literature.