Abstract
Background and objective
Nurses are essential for safeguarding public health, and their physical condition directly affects care quality and patient safety. Work-related musculoskeletal disorders (WMSDs) are highly prevalent in this workforce and may be amplified by circadian disruption. We therefore integrated individual and environmental risk factors, with special attention to night-shift characteristics, to identify and rank determinants of WMSDs among shift-working nurses. Seven machine-learning algorithms were compared to generate a comprehensive, validated prediction tool that enables managers and nurses to implement targeted, proactive interventions and reduce occupational injury.
Methods
This study is a cross-sectional study. The general information, lifestyle, psychosocial data, working environment and shift characteristics of shift nurses were collected, and the influencing factors of WMSDs were analyzed. The mean square error increase and residual sum of squares are calculated, and the importance of influencing factors is sorted respectively. The independent influencing factors of WMSDs in shift nurses were included. After screening variables again by Lasso regression, seven prediction models of LDA, PLS, RDA, GLM, RF, SVM-Radial and SVM-Linear were established by machine learning. The AUC, accuracy and specificity median were used to evaluate the prediction efficiency, and the best prediction model was obtained and the accuracy of the prediction factors was verified.
Results
Among 1 080 shift-working nurses, the WMSD prevalence at any body site was 85.19%. The top-ranked determinants were perceived control, perceived social support, Pittsburgh Sleep Quality Index (PSQI) score, chronotype, frequent bending over, night-shift nap duration, friend support, work support, family support, prolonged neck flexion and years in nursing. The LASSO-selected predictor set comprised dairy intake frequency, shift pattern, monthly night shifts, number of nurses on night duty, post-night-shift recovery days, post-night-shift catch-up sleep, frequent trunk flexion, prolonged neck flexion, PSQI, chronotype and perceived control. Random forest achieved the highest predictive performance (median AUC = 0.919).
Conclusions
Individual characteristics, lifestyle, physical condition, occupational features, shift schedule, biomechanical load and psychosocial factors collectively influence WMSD occurrence in nurses. Random forest outperformed the other algorithms and should be carried out in conjunction with various factors in the model.
Keywords: Nurse, Shift work, Work-related musculoskeletal disorders, Predictive model
Background
In the process of continuous improvement of the medical and health system, nurses, as an important force in ensuring the safety and physical health of the people, have received increasing attention to their work status and health level. “Shift-working nurses” are defined as registered nurses who rotate between at least two of the following duty periods: day, evening and night [1]. In the healthcare sector, 24-hour shift systems are mandatory to guarantee uninterrupted, high-quality nursing care. However, rotating between day, evening and night duties repeatedly desynchronizes the circadian clock, leading to suppressed nocturnal melatonin and altered reproductive-hormone profiles. This circadian disruption has been consistently associated with higher risks of breast and other cancers, cardiovascular events, metabolic syndrome and type 2 diabetes among shift-working nurses [2–4]. A comprehensive understanding of the determinants underlying these disorders and the implementation of evidence-based preventive interventions are therefore essential to safeguarding nurses’ occupational health and, by extension, maintaining high-quality patient care.
Work-related musculoskeletal disorders (WMSDs) are defined as discomfort, pain, numbness, or other symptoms occurring in one or more major body regions as a result of work or the work environment. These symptoms persist for ≥ 24 h and do not resolve with rest. A recent meta-analysis reported that the annual prevalence of WMSDs among Chinese nurses is 79%, with the lower back being the most frequently affected site, followed by the neck and shoulders [5]. This high prevalence and heavy burden have already precipitated attrition within the nursing workforce and a measurable decline in the quality of care delivered [6]. Over the past decade, the body of evidence on WMSD risk factors among nurses has expanded rapidly. In a cross-sectional survey of 1 803 nurses, Lin et al. [7] identified older age, longer tenure, specialty unit, ≥ 5 work-days per week and shift work as independent predictors. Mekonnen et al. [8]reported that academic degree, shift schedule and absence of health-and-safety training were significantly associated with low-back pain. Yao et al. [9]found that physical inactivity, night-shift rotation and negative affect collectively doubled WMSD risk. A multicentre study covering 20 tertiary ICUs in Hunan province revealed that female sex, unmarried status, high hazard-perception scores and poor ergonomic environments (e.g., lack of patient-lift devices or insufficient lifting space) markedly increased WMSD odds [10]. Taking a psychosocial perspective, Freimann et al. [11]demonstrated that permanent night duty and frequent emergency calls eroded resilience and job satisfaction, thereby amplifying musculoskeletal symptoms. Collectively, these findings indicate that WMSD determinants in nurses cluster into three domains: (i) basic demographics and lifestyle, (ii) work situation and biomechanical load, and (iii) psychosocial conditions.
Importantly, emerging evidence links WMSDs directly to circadian disruption. Skeletal muscle is the largest collection of peripheral circadian rhythms in the human body. Both the central and peripheral circadian rhythms regulate the interaction between the musculoskeletal system and energy metabolism. The skeletal muscle circadian rhythm is crucial in lipid and glucose metabolism, affecting the regulation of circadian rhythms in shift workers. A meta-analysis by Chang et al. [12]showed that shift-working nurses are significantly more likely to develop WMSDs than their fixed-day counterparts. Prolonged shift duration (> 9 h) is an independent predictor of WMSDs [13], and there is a clear dose–response relationship between years of shift work and disorder risk: every additional 5-year increment in shift tenure raises the prevalence of lower-limb WMSDs by 4% [14]. Moreover, a high frequency of night shifts amplifies the likelihood of multi-site pain [15]. Therefore, the present study collected detailed shift-related parameters—whether the nurse rotated shifts on a regular basis, lifetime night-shift tenure, number of night shifts per month, length of each night shift, duration of on-duty naps, and recovery days after night shifts—to quantify how circadian disruption influences WMSD occurrence.
In recent years, machine-learning techniques have gained prominence in artificial-intelligence-assisted diagnosis and prediction because of their speed, high precision, efficiency and security. Compared with conventional statistical approaches, machine-learning algorithms can simultaneously accommodate a large number of predictors, mitigate multicollinearity and automatically capture flexible, data-driven interactions, thereby improving both the accuracy and the efficiency of predictive models [16]. Therefore, our study takes shift-working nurses, a high-risk group characterized by circadian rhythm disorders, as the research subjects to enhance the clinical relevance of the research. Unlike previous studies that primarily focused on biomechanical factors, in addition to considering both individual characteristics and work-related biomechanical features, our study also pays attention to psychosocial factors and the specific circumstances of shift work, and ranks their importance. Furthermore, it employs seven different machine learning algorithms to validate the predictive performance of the model and the accuracy of the predictive variables, providing evidence-based grounds for the development of targeted prevention and intervention strategies for shift-working healthcare professionals.
Subjects and methods
Participants
Using convenience sampling method, nurses engaged in clinical work from 6 hospitals in 3 cities in Guangxi who meet the inclusion and exclusion criteria were selected as the research subjects, and an offline combined with online questionnaire survey was conducted. Inclusion criteria: (1) Holding a valid nursing license and having worked independently for ≥ 1 year; (2) Informed of the study aims and methods and willing to participate. Exclusion criteria: (1) Not currently on hospital duty (e.g., on study leave, maternity or sick leave); (2) Pregnant or breastfeeding within the past 12 months; (3) Pre-existing musculoskeletal conditions attributable to congenital spinal disorders, tumors, gynecological diseases or other non-occupational causes; (4) History of trauma or surgery within the past 12 months; (5) Musculoskeletal complaints clearly resulting from regular exercise, domestic overexertion or similar non-work factors. The study protocol was approved by the hospital ethics committee (Approval No. 2023-KY−0941).
Sample-size calculation: Using the formula N= deff[µα/22p(1-p)]/δ2, with α=0.05, estimated prevalence p = 82.5% [17], the minimum required sample was 856. Allowing for 20% incomplete data, we set the final target at ≥ 1 027 shift-working nurses.
Survey content
WMSD risk-factor questionnaire
Sociodemographics: age, sex, education, marital and parity status, BMI, waist and hip circumference, blood type.
Occupational exposures: department, professional title, length of service, shift pattern, monthly income, adequacy of workspace, frequent bending over, often carry heavy objects, maintain a lowered posture for a long time during work, work in an uncomfortable position frequently, bend down and turn around frequently during work, use of vibrating tools, physical/chemical hazards, exposure to cold draughts or temperature variations, perceived lighting quality, previous WMSD-prevention training.
Night-shift specifics: regular rotation, night shift working years, night shifts per month, length of each night shift, interval between two night shifts, night shift nap duration, number of nurses on night shift, recovery days after nights, duration of sleep after night shift, satisfaction with night-shift frequency.
Lifestyle and health: frequency of drinking tea/coffee/alcohol, frequency of consuming dairy products, daily water intake, second-hand smoke exposure, weekly exercise sessions, chronic disease history, self-rated health.
Work-related musculoskeletal disorders
The 9-item Nordic Musculoskeletal Questionnaire (NMQ-E) was used to assess pain or discomfort in nine body regions (neck, shoulder, upper back, lower back, elbow, wrist, hips/thighs, knee, ankles/feet) during the preceding 12 months [18]. Each region was coded 0 = no symptoms, 1 = symptoms present; participants reporting pain in ≥ 1 region were classified as WMSD-positive. Cronbach’s α for the scale in this study was 0.863.
Pittsburgh Sleep Quality Index (PSQI)
The 19-item PSQI evaluates seven components of sleep over the past month: subjective sleep quality, sleep latency, sleep duration, habitual sleep efficiency, sleep disturbances, use of sleep medication and daytime dysfunction [19]. Global scores range 0–21; >5 indicates poor sleep quality. Cronbach’s α= 0.763.
Morningness–eveningness questionnaire−5 (MEQ−5)
MEQ−5 is a simplified morning and night type scale, consisting of 5 questions with a score range of 4–25 [20]. The higher the score, the more inclined an individual’s circadian rhythm is towards the morning type, while the lower the score, the more inclined it is towards the night type. The Cronbach’s α of MEQ−5 in this study is 0.728.
Perceived social support scale (PSSS)
This scale consists of 12 items, including 3 dimensions: family support, friend support, and work support [21]. It uses a Likert 7-point scoring system, ranging from 1 (strongly disagree) to 7 (strongly agree), with higher scores indicating better social support. This study conducted reliability tests on the dimensions of job support, family support, friend support, and the overall scale of PSSS, which were 0.973, 0.944, 0.977, and 0.960, respectively.
Sense of agency scale (SoAS)
The 9-item Sense of Agency Scale developed by Tapal et al. [22], previously translated and validated in Chinese, was administered. Items are rated 1–7; items 2, 3, 6 and 7 are reverse-scored. Total scores range 9–63, with higher values reflecting stronger perceived control. Cronbach’s α= 0.822.
Quality control
Before data collection, all research staff received standardized training on inclusion/exclusion criteria and operational definitions of every variable. On site, participants were given a uniform explanation of the study purpose, questionnaire completion instructions and estimated completion time; clarifications were provided when items were unclear. For online recruitment, the survey link and QR code were distributed through WeChat groups. The electronic questionnaire was programmed to allow only one response per device and to prevent submission until every item was answered. Each evening, newly received records were checked for completeness and logical consistency. After retrieval, data were double-entered and cross-verified. Records exhibiting patterned responses (e.g., identical answers across an entire page) or implausibly short completion times (< 5 min) were excluded.
Statistical analysis
Descriptive statistics were generated with Excel and SPSS 23.0. Continuous variables with a normal distribution are presented as mean ± SD and were compared between groups using Student’s t-test or one-way ANOVA; non-normally distributed variables are expressed as median (inter-quartile range) and were compared with the Mann–Whitney U or Kruskal–Wallis test. Categorical data are summarized as frequencies and percentages, and group differences were tested with the χ² or Fisher’s exact test. Multivariable logistic regression was used to identify independent factors associated with WMSDs among shift-working nurses. All tests were two-tailed; P < 0.05 was considered statistically significant.
Variable importance was quantified with the randomForest package in R Studio (v4.3.3) using the increase in mean-squared error and residual sum of squares. Variables that were significant in univariate analyzes were entered as candidate predictors. Subsequent variable selection was performed with LASSO regression (glmnet package). Seven machine-learning algorithms—linear discriminant analysis (LDA), partial least squares (PLS), regularized discriminant analysis (RDA), generalized linear model (GLM), random forest (RF), radial-kernel support-vector machine (SVM-Radial) and linear-kernel SVM (SVM-Linear)—were implemented with the caret package. To obtain robust generalisability estimates and reduce evaluation bias introduced by different data-splitting schemes, we used repeated 10 × 10-fold cross-validation. The dataset was randomly divided into 10 approximately equal folds; in each of the 10 repeats, nine folds (90%) were used for training and the remaining fold (10%) for validation. The grid search with a tuning length of 5 was used for tuning parameters. Identical training and test folds were employed for all seven models. Model performance was assessed with the median area under the receiver-operating-characteristic curve (AUC), accuracy and specificity; the algorithm with the highest AUC was selected as the optimal predictor.
Results
Prevalence and single-factor analysis of WMSDs among shift nurses
A total of 1 080 shift-working nurses were included in our study, with a mean age of 30.29 ± 6.03 years. The prevalence of WMSDs in at least one body region was 85.19%. The prevalence by specific body region was as follows: neck 66.85%, shoulders 57.31%, upper back 39.44%, lower back 63.15%, elbows 33.33%, wrists 35.19%, hips/thighs 27.22%, knees 37.22%, and ankles/feet 35.83%. Univariate comparisons between nurses with and without WMSDs are summarized below.
Demographics, lifestyle and general health
Significant between-group differences (P < 0.05) were observed for age, blood type, second-hand smoke exposure, frequency of drinking tea/coffee, frequency of consuming dairy products, daily water intake, body weight, waist circumference, hip circumference, self-reported hyperlipidaemia, chronic bronchitis or other respiratory diseases, gastro-intestinal disorders, and self-rated health status (Table 1).
Table 1.
Analysis of WMSDs of demographic data, lifestyle and physical health of shift nurses
| Variables | WMSDs in the past year | t/χ2/Z | P | ||
|---|---|---|---|---|---|
| Yes(n = 920) | No(n = 160) | ||||
| Age | 30.29 ± 6.03 | 30.47 ± 6.07 | 29.26 ± 5.76 | 2.345 | 0.019 |
| Sex | 1.700 | 0.192 | |||
| Male | 89(8.24) | 80(89.89) | 9(10.11) | ||
| Female | 991(91.76) | 840(84.76) | 151(15.24) | ||
| Education | 1.261 | 0.532 | |||
| Junior college or below | 295(27.31) | 247(83.73) | 48(16.27) | ||
| Bachelor | 771(71.39) | 662(85.86) | 109(14.14) | ||
| Master or above | 14(1.30) | 11(78.57) | 3(21.43) | ||
| Marital status | 0.702 | 0.402 | |||
| Married | 534(49.44) | 450(84.27) | 84(15.73) | ||
| Unmarried / divorced / widowed | 546(50.56) | 470(86.08) | 76(13.92) | ||
| Number of children | 2.452 | 0.484 | |||
| 0 | 607(56.20) | 522(86.00) | 85(14.00) | ||
| 1 | 214(19.81) | 182(85.05) | 32(14.95) | ||
| 2 | 245(22.69) | 202(82.45) | 43(17.55) | ||
| >2 | 14(1.30) | 14(100.00) | 0(0.00) | ||
| Blood type | 12.896 | 0.005 | |||
| A | 252(23.33) | 220(87.30) | 32(12.70) | ||
| B | 325(30.09) | 279(85.85) | 46(14.15) | ||
| AB | 102(9.44) | 96(94.12) | 6(5.88) | ||
| O | 401(37.13) | 325(81.05) | 76(18.95) | ||
| Weekly exercise frequency | 1.987 | 0.370 | |||
| 0 | 535(49.54) | 457(85.42) | 78(14.58) | ||
| 1–3 | 454(42.04) | 390(85.90) | 64(14.10) | ||
| >3 | 91(8.43) | 73(80.22) | 18(19.78) | ||
| Second-hand smoke exposure | 20.538 | <0.001 | |||
| Yes | 357(33.06) | 329(92.16) | 28(7.84) | ||
| No | 723(66.94) | 591(81.74) | 132(18.26) | ||
| frequency of drinking tea/coffee | 9.773 | 0.021 | |||
| ≥1 time per day | 201(18.61) | 177(88.06) | 24(11.94) | ||
| 1–6 times per week | 375(34.72) | 329(87.73) | 46(12.27) | ||
| 1–3 times per month | 134(12.41) | 116(86.57) | 18(13.43) | ||
| Rarely | 370(34.26) | 298(80.54) | 72(19.46) | ||
| Dairy-product intake | 21.875 | <0.001 | |||
| ≥1 time per day | 320(29.63) | 248(77.50) | 72(22.50) | ||
| 1–6 times per week | 504(46.67) | 444(88.10) | 60(11.90) | ||
| 1–3 times per month | 147(13.61) | 129(87.76) | 18(12.24) | ||
| Rarely | 109(10.09) | 99(90.83) | 10(9.17) | ||
| Daily water intake | 42.646 | <0.001 | |||
| <500 mL | 171(15.83) | 150(87.72) | 21(12.28) | ||
| 500–999 mL | 399(36.94) | 335(83.96) | 64(16.04) | ||
| 1000–1499 mL | 285(26.39) | 255(89.47) | 30(10.53) | ||
| 1500–2000 mL | 159(14.72) | 141(88.68) | 18(11.32) | ||
| >2000 mL | 66(6.11) | 39(59.09) | 27(40.91) | ||
| Body weight | 54.54 ± 8.69 | 54.83 ± 8.96 | 52.93± 6.75 | 11.892 | 0.002 |
| Waist circumference | 71.33 ± 8.07 | 71.66 ± 8.27 | 69.49± 6.59 | 5.630 | <0.001 |
| Hip circumference | 90.44 ± 9.99 | 90.83 ± 9.97 | 88.24± 9.81 | 0.107 | 0.002 |
| Hyperlipidaemia | 6.422 | 0.011 | |||
| Yes | 1001(92.69) | 845(84.42) | 156(15.58) | ||
| No | 79(7.31) | 75(94.94) | 4(5.06) | ||
| Chronic bronchitis or other respiratory diseases | 8.566 | 0.003 | |||
| Yes | 999(92.50) | 842(84.28) | 157(15.72) | ||
| No | 81(7.50) | 78(96.30) | 3(3.70) | ||
| Gastroenteritis or other digestive disorders | 3.538 | 0.048 | |||
| Yes | 959(88.80) | 810(84.46) | 149(15.54) | ||
| No | 121(11.20) | 110(90.91) | 11(9.09) | ||
| Self-rated health status | 96.444 | <0.001 | |||
| Excellent | 145(13.43) | 90(62.07) | 55(37.93) | ||
| Good | 349(32.31) | 284(81.38) | 65(18.62) | ||
| Fair | 486(45.00) | 449(92.39) | 37(7.61) | ||
| Poor | 100(9.26) | 97(97.00) | 3(3.00) | ||
| PSQI | 6.88± 3.67 | 7.34 ± 3.56 | 4.24 ± 3.11 | 10.439 | <0.001 |
| Subjective sleep quality | 1.34± 0.74 | 1.41 ± 0.72 | 0.95 ± 0.76 | −7.466 | <0.001 |
| Sleep latency | 1.67± 0.96 | 1.74 ± 0.94 | 1.26 ± 0.98 | −6.00 | <0.001 |
| Sleep duration | 1.33± 0.84 | 1.43 ± 0.83 | 0.76 ± 0.69 | −9.588 | <0.001 |
| Sleep efficiency | 0.61± 0.86 | 0.65 ± 0.88 | 0.39 ± 0.66 | −3.462 | <0.001 |
| Sleep disturbances | 1.13± 0.66 | 1.19 ± 0.63 | 0.83 ± 0.73 | −6.593 | <0.001 |
| Hypnotic medication use | 0(0, 0) | 0(0, 0) | 0(0, 0) | −3.208 | <0.001 |
| Daytime dysfunction | 1.58± 0.99 | 1.69 ± 0.96 | 0.94 ± 0.91 | −9.153 | <0.001 |
| Chronotype | 13.46± 3.72 | 13.06 ± 3.53 | 15.80 ± 3.92 | −8.927 | <0.001 |
| Social support | 60.63± 16.97 | 59.51 ± 16.69 | 66.66 ± 15.66 | −5.045 | <0.001 |
| Work support | 18.87± 7.05 | 18.45 ± 6.94 | 21.24 ± 7.26 | 4.650 | <0.001 |
| Family support | 21.44± 5.78 | 21.22 ± 5.81 | 22.73 ± 5.41 | 3.052 | 0.002 |
| Friend support | 20.32± 6.53 | 19.73 ± 6.61 | 23.69 ± 4.83 | 7.248 | <0.001 |
| Sense of agency | 46.66± 7.73 | 45.58 ± 8.07 | 51.06 ± 5.985 | −8.196 | <0.001 |
Occupational and shift characteristics
Univariate analysis revealed statistically significant differences (P < 0.05) between nurses with and without WMSDs in the following variables: nursing tenure (years), intention to choose nursing profession, clinical department, shift-rotation pattern, lifetime night-shift tenure, regularity of shift rotation, number of night shifts per month, duration of each night shift, interval between consecutive night duties, on-duty nap length during nights, number of nurses on night duty, patient load on day shifts, post-night-shift recovery days, catch-up sleep after night blocks, and satisfaction with night-shift frequency (Table 2).
Table 2.
Single factor analysis of WMSDs annual occurrence of work and shift characteristics of shift nurses
| Variables | WMSDs in the past year | t/χ2/Z | P | |||
|---|---|---|---|---|---|---|
| Yes(n = 920) | No(n = 160) | |||||
| Length of service | 8.63 ± 6.24 | 8.81 ± 6.29 | 7.58 ± 5.871 | 2.300 | 0.022 | |
| Intention to choose nursing profession | 16.904 | <0.001 | ||||
| Self-selected | 609(56.39) | 495(81.28) | 114(18.72) | |||
| Recommended by others | 415(38.46) | 374(90.12) | 41(9.88) | |||
| Reassigned | 56(5.19) | 51(91.07) | 5(8.93) | |||
| Department | 106.583 | <0.001 | ||||
| Internal medicine | 341(31.57) | 306(89.74) | 35(10.26) | |||
| Surgery | 298(27.59) | 270(90.60) | 28(9.40) | |||
| ICU | 120(11.11) | 104(86.67) | 16(13.33) | |||
| Emergency | 79(7.31) | 54(68.35) | 25(31.65) | |||
| Pediatrics | 92(8.52) | 50(54.35) | 42(45.65) | |||
| Obstetrics & gynecology | 89(8.24) | 78(87.64) | 11(12.36) | |||
| Operating room | 61(5.65) | 58(95.08) | 3(4.92) | |||
| Professional title | 2.723 | 0.436 | ||||
| Staff nurse | 281(26.02) | 237(84.34) | 44(15.66) | |||
| Senior nurse | 464(42.96) | 402(86.64) | 62(13.36) | |||
| Supervisor nurse | 307(28.43) | 255(83.06) | 52(16.94) | |||
| Associate chief nurse or above | 28(2.59) | 26(92.86) | 2(7.14) | |||
| Average monthly income | 7.018 | 0.071 | ||||
| < 4 000 | 106(9.81) | 91(85.85) | 15(14.15) | |||
| 4 000–6 000 | 467(43.24) | 410(87.79) | 57(12.21) | |||
| 6 001–9 000 | 376(34.81) | 306(81.38) | 70(18.62) | |||
| > 9 000 | 131(12.13) | 113(86.26) | 18(13.74) | |||
| Shift pattern | 6.103 | 0.013 | ||||
| Day-Night (DN) | 517(47.87) | 426(82.40) | 91(17.60) | |||
| Morning-Evening-Night (APN) | 563(52.13) | 494(87.74) | 69(12.26) | |||
| History of work error | 3.819 | 0.051 | ||||
| Yes | 536(49.63) | 468(87.31) | 68(12.69) | |||
| No | 544(50.37 | 452(83.09) | 92(16.91) | |||
| Years of shift work | 8.34± 6.14 | 8.52± 6.20 | 7.28± 0.13 | 2.358 | 0.019 | |
| Regular shift rotation | 8.400 | 0.004 | ||||
| Yes | 805(74.54) | 671(83.35) | 134(16.65) | |||
| No | 275(25.46) | 249(90.55) | 26(9.45) | |||
| Night shifts per month | 44.480 | <0.001 | ||||
| ≤ 3 | 159(14.72) | 110(69.18) | 49(30.82) | |||
| 4–5 | 595(55.09) | 510(85.71) | 85(14.29) | |||
| ≥ 6 | 326(30.19) | 300(92.02) | 26(7.98) | |||
| Length of each night shift | 4.753 | 0.029 | ||||
| < 9 h | 253(32.69) | 207(81.82) | 46(18.18) | |||
| 9–12 h | 503(46.57) | 427(84.89) | 76(15.11) | |||
| > 12 h | 324(30.00) | 286(88.27) | 38(11.73) | |||
| Interval between night shifts | 17.638 | <0.001 | ||||
| < 2 days | 142(13.15) | 123(86.62) | 19(13.38) | |||
| 2–3 days | 262(24.26) | 241(91.98) | 21(8.02) | |||
| 4–7 days | 554(51.30) | 462(83.39) | 92(16.61) | |||
| > 7 days | 122(11.30) | 94(77.05) | 28(22.95) | |||
| Night shift nap duration | 56.553 | <0.001 | ||||
| 0 h | 332(30.74) | 285(85.84) | 47(14.16) | |||
| 1 h | 114(10.56) | 111(97.37) | 3(2.63) | |||
| 2 h | 368(34.07) | 332(90.22) | 36(9.78) | |||
| ≥ 3 h | 266(24.63) | 192(72.18) | 74(27.82) | |||
| Average number of nurses on duty per night shift | 22.516 | <0.001 | ||||
| 1 | 244(22.59) | 220(90.16) | 24(9.84) | |||
| 2 | 540(50.00) | 472(87.41) | 68(12.59) | |||
| ≥ 3 | 296(27.41) | 228(77.03) | 68(22.97) | |||
| Average number of patients assigned per day shift | 39.927 | <0.001 | ||||
| ≤ 2 | 41(3.80) | 21(51.22) | 20(48.78) | |||
| 3–5 | 133(12.31) | 115(86.47) | 18(13.53) | |||
| 6–8 | 212(19.63) | 179(84.43) | 33(15.57) | |||
| > 8 | 694(64.26) | 609(87.75) | 89(12.82) | |||
| Post-night-shift recovery days | 25.960 | <0.001 | ||||
| 1 day | 155(14.35) | 142(91.61) | 13(8.39) | |||
| 2 days | 783(72.50) | 676(86.33) | 107(13.67) | |||
| > 2 days | 142(13.15) | 102(71.83) | 40(28.17) | |||
| Sleep duration after night shift | 60.426 | <0.001 | ||||
| < 4 h | 307(28.43) | 289(94.14) | 18(5.86) | |||
| 4–6 h | 567(52.50) | 488(86.07) | 79(13.93) | |||
| > 6 h | 206(19.07) | 143(69.42) | 63(30.58) | |||
| Satisfaction with night-shift frequency | 46.777 | <0.001 | ||||
| Satisfied | 263(24.35) | 197(74.90) | 66(25.10) | |||
| Neutral | 613(56.76) | 524(85.48) | 89(14.52) | |||
| Dissatisfied | 204(18.89) | 199(97.55) | 5(2.45) | |||
Occupational biomechanical factors
Univariate comparisons showed significant between-group differences (P < 0.05) for the following biomechanical variables: frequent bending over, regular heavy lifting, prolonged neck flexion, working in awkward postures, simultaneous trunk flexion and rotation, use of vibrating tools, exposure to physical hazards, exposure to chemical hazards, working in cold draughts or variable ambient temperatures, perceived lighting quality, and attendance at WMSD-prevention training (Table 3).
Table 3.
Single factor analysis of WMSDs annual occurrence of work biomechanics of shift nurses
| Variables | WMSDs in the past year | t/χ2/Z | P | |||
|---|---|---|---|---|---|---|
| Yes(n = 920) | No(n = 160) | |||||
| Adequacy of workspace | 2.438 | 0.118 | ||||
| Yes | 782(72.41) | 658(84.14) | 124(15.86) | |||
| No | 298(27.59) | 262(87.92) | 36(12.08) | |||
| frequent bending over | 131.167 | <0.001 | ||||
| Yes | 866(80.19) | 791(91.34) | 75(8.66) | |||
| No | 214(19.81) | 129(60.28) | 85(39.72) | |||
| Regular heavy lifting | 29.057 | <0.001 | ||||
| Yes | 596(55.19) | 539(90.44) | 57(9.56) | |||
| No | 484(44.81) | 381(78.72) | 103(21.28) | |||
| Prolonged neck flexion during work | 102.568 | <0.001 | ||||
| Yes | 731(67.69) | 678(92.75) | 53(7.25) | |||
| No | 349(32.31) | 242(69.34) | 107(30.66) | |||
| Frequent awkward postures | 103.497 | <0.001 | ||||
| Yes | 607(56.20) | 576(94.89) | 31(5.11) | |||
| No | 473(43.80) | 344(72.73) | 129(27.27) | |||
| Combined trunk flexion and rotation | 75.515 | <0.001 | ||||
| Yes | 664(61.48) | 615(92.62) | 49(7.38) | |||
| No | 416(38.52) | 305(73.32) | 111(26.68) | |||
| Use of vibrating tools | 21.746 | <0.001 | ||||
| Yes | 223(20.65) | 212(95.07) | 11(4.93) | |||
| No | 857(79.35) | 708(82.61) | 149(17.39) | |||
| Exposure to physical hazards | 49.221 | <0.001 | ||||
| Rarely | 429(39.72) | 345(80.42) | 84(19.58) | |||
| 1–4 times per month | 324(30.00) | 259(79.94) | 65(20.06) | |||
| 2–6 times per week | 174(16.11) | 166(95.40) | 8(4.60) | |||
| 1–3 times per day | 92(8.52) | 90(97.83) | 2(2.17) | |||
| > 3 times per day | 61(5.65) | 60(98.36) | 1(1.64) | |||
| Exposure to chemical hazards | 18.793 | <0.001 | ||||
| Rarely | 587(54.35) | 483(82.28) | 104(17.72) | |||
| 1–4 times per month | 244(22.59) | 204(83.61) | 40(16.39) | |||
| 2–6 times per week | 108(10.00) | 102(94.44) | 6(5.56) | |||
| 1–3 times per day | 75(6.94) | 71(94.67) | 4(5.33) | |||
| > 3 times per day | 66(6.11) | 60(90.91) | 6(9.09) | |||
| Cold draughts or temperature variations in work area | 42.785 | <0.001 | ||||
| Yes | 493(45.65) | 458(92.90) | 35(7.10) | |||
| No | 587(54.35) | 462(78.71) | 125(21.29) | |||
| Perceived lighting quality at workstation | 8.997 | 0.011 | ||||
| Too bright | 197(18.24) | 156(79.19) | 41(20.81) | |||
| Comfortable | 733(67.87) | 640(87.31) | 93(12.69) | |||
| Too dim | 150(13.89) | 124(82.67) | 26(17.33) | |||
| Attendance at WMSD-prevention training | 9.413 | 0.002 | ||||
| Yes | 261(24.17) | 207(79.31) | 54(20.69) | |||
| No | 819(75.83) | 713(87.06) | 106(12.94) | |||
Multivariable analysis of annual WMSD occurrence
Using the presence or absence of any WMSD during the preceding 12 months as the dependent variable, we entered the 53 statistically significant factors identified in univariate analysis as candidate predictors. Multivariable logistic regression retained 18 independent determinants (P < 0.05): dairy-product intake frequency, daily water intake, intention to choose nursing profession, clinical department, shift pattern, regular shift rotation, number of night shifts per month, number of nurses on night duty, post-night-shift recovery days, catch-up sleep after nights, frequent bending over, prolonged neck flexion, use of vibrating tools, waist circumference, sleep-onset time, PSQI score, chronotype and sense of agency. Likelihood-ratio χ² = 373.384, P < 0.001, indicating adequate model fit (Table 4).
Table 4.
Multivariate analysis of WMSDs in shift nurses
| B | SE | Wald X2 | P | OR | 95% CI | |
|---|---|---|---|---|---|---|
| Frequency of dairy-product intake | 0.364 | 0.132 | 7.591 | 0.006 | 1.439 | 1.111–1.865 |
| Daily water intake | 0.328 | 0.114 | 8.321 | 0.004 | 1.389 | 1.111–1.736 |
| Route into nursing | 0.64 | 0.215 | 8.848 | 0.003 | 1.897 | 1.244–2.892 |
| Clinical department | -0.138 | 0.062 | 5.032 | 0.025 | 0.871 | 0.772–0.983 |
| Shift-rotation pattern | 1.045 | 0.24 | 18.894 | <0.001 | 2.843 | 1.775–4.555 |
| Regular shift rotation | 0.627 | 0.297 | 4.463 | 0.035 | 1.872 | 1.046–3.349 |
| Night shifts per month | 0.724 | 0.189 | 14.733 | <0.001 | 2.063 | 1.425–2.986 |
| Number of nurses on night duty | -0.507 | 0.17 | 8.876 | 0.003 | 0.602 | 0.431–0.841 |
| Post-night-shift recovery days | -0.701 | 0.219 | 10.265 | 0.001 | 0.496 | 0.323–0.762 |
| Catch-up sleep after nights | -0.602 | 0.179 | 11.258 | 0.001 | 0.548 | 0.385–0.779 |
| frequent bending over | 0.672 | 0.292 | 5.292 | 0.021 | 1.959 | 1.105–3.473 |
| Prolonged neck flexion | 0.949 | 0.272 | 12.154 | <0.001 | 2.582 | 1.515–4.401 |
| Use of vibrating tools | 1.168 | 0.408 | 8.205 | 0.004 | 3.217 | 1.446–7.156 |
| Waist circumference | 0.037 | 0.015 | 6.243 | 0.012 | 1.038 | 1.008–1.069 |
| Sleep-onset time | -0.631 | 0.187 | 11.339 | 0.001 | 0.532 | 0.369–0.768 |
| PSQI global score | 0.288 | 0.06 | 22.976 | <0.001 | 1.334 | 1.186–1.501 |
| Chronotype | -0.207 | 0.042 | 24.271 | <0.001 | 0.813 | 0.749–0.883 |
| Sense of agency | -0.087 | 0.019 | 21.816 | <0.001 | 0.917 | 0.884–0.951 |
| Constant | 2.81 | 1.772 | 2.515 | 0.113 | 16.615 |
Importance ranking of factors associated with annual WMSD occurrence
We ranked the 53 candidate predictors by variable importance using both the increase in mean-squared error (MSE) and the residual sum of squares (RSS); the resulting rankings are displayed in Fig. 1. Regardless of which metric was used, the top determinants of WMSDs among shift-working nurses were consistently: sense of agency, perceived social support, PSQI global score, chronotype, frequent bending over, on-duty nap duration during night shifts, friend support, work support, family support, long term bowing and years of work experience.
Fig. 1.
The importance ranking of influencing factors of nurses’ WMSDs based on random forest regression model analysis
Model development and validation
Final predictor selection with LASSO regression
The 18 variables retained in the multivariable logistic model were entered into a LASSO regression to further reduce over-fitting and select the most parsimonious set. Ten-fold cross-validation identified the optimal tuning parameter (λ = 0.03067). Eleven predictors with non-zero coefficients were retained: dairy-product intake frequency, shift-rotation pattern, number of night shifts per month, number of nurses on night duty, post-night-shift recovery days, catch-up sleep after nights, frequent trunk flexion, prolonged neck flexion, PSQI global score, chronotype and sense of agency (Fig. 2). Variable coding is shown in Table 5.
Fig. 2.
Lasso regression model to screen predictive variables. A: Lasso coefficient distribution for screening feature variables; B: Lasso model adaptive method for optimal λ value selection
Table 5.
Variable assignment method
| Variables | Assignment method | |
|---|---|---|
| 1 | Dairy-product intake | ≥ 1 time per day= 1, 1–6 times per week= 2, 1–3 times per month= 3, Rarely= 4 |
| 2 | Shift pattern | DN = 1, APN = 2 |
| 3 | Night shifts per month | ≤ 3 = 1,4–5 = 2, ≥ 6 = 3 |
| 4 | Number of night nurses | 1 person = 1,2 person = 2; ≥ 3 people = 3 |
| 5 | Night shift days off | 1 day = 1, 2 days = 2; > 2 days = 3 |
| 6 | Duration of sleep after night shift | 6 h = 3 |
| 7 | He often bends heavily | Yes = 1, No = 0 |
| 8 | Head down for a long time | Continuous variables |
| 9 | PSQI | Continuous variable |
| 10 | MEQ-5 | Continuous variable |
| 11 | SoAS | Continuous variable |
A B.
Model construction and performance comparison
Seven algorithms (LDA, PLS, RDA, GLM, RF, SVM-Radial and SVM-Linear) were trained on the 11 LASSO-selected features. There is no significant difference in the training time of each model. After 100 repetitions of stratified 10-fold cross-validation on the full dataset, median AUCs are displayed in Fig. 3; corresponding accuracy and specificity are summarized in Table 6. All seven models achieved good discriminative ability, confirming the predictive value of the 11-feature set. Random forest delivered the highest performance, with a median AUC of 0.919, and was therefore retained as the final prediction tool.
Fig. 3.
Comparison of AUC medians of prediction models constructed by 7 machine learning algorithms
Table 6.
Performance comparison of prediction models constructed by seven machine learning algorithms
| Model | AUC | specificity | accuracy |
|---|---|---|---|
| rf | 0.919(0.899–0.945) | 0.981(0.978–0.989) | 0.923(0.907–0.935) |
| glm | 0.868(0.840–0.904) | 0.973(0.967–0.978) | 0.901(0.889–0.917) |
| pls | 0.864(0.836–0.902) | 0.995(0.989-1.000) | 0.884(0.870–0.889) |
| rda | 0.869(0.845–0.898) | 0.954(0.934–0.967) | 0.885(0.870–0.907) |
| svmLinear | 0.863(0.836–0.899) | 0.978(0.967–0.989) | 0.895(0.887–0.907) |
| lda | 0.861(0.837–0.890) | 0.958(0.945–0.978) | 0.884(0.870–0.898) |
| svmRadial | 0.860(0.833–0.896) | 0.979(0.967–0.989) | 0.905(0.889–0.917) |
Note: Each algorithm was evaluated over 100 cross-validation runs; values are reported as median (inter-quartile range)
Discussion
Importance ranking of determinants of WMSDs in shift-working nurses
The influencing factors of WMSDs related to shift nursing work have multidimensional mechanisms, and their importance is ranked in the following order: (i) sense of agency, (ii) perceived social support, (iii) PSQI score, (iv) chronotype, (v) frequent bending over and (iv) night shift nap duration. Sense of agency emerged as the most important predictor. Nurses who perceive high control over work pace and task allocation can autonomously adjust working postures and redistribute physical loads, thereby lowering musculoskeletal strain [23]. Emotional and instrumental support from colleagues, supervisors and family can significantly alleviate occupational stress, thereby indirectly reducing compensatory muscle recruitment [24]. Previous studies have shown that nurses who perceive low social support report a higher frequency of neck-and-shoulder pain; this is presumably because diminished psychological stress translates into lower resting muscle tension [25]. PSQI is a quantitative indicator of fatigue accumulation, and poor sleep quality directly leads to circadian rhythm disorders and growth hormone secretion inhibition, weakening muscle repair ability. Fragmented sleep caused by frequent night shifts is more likely to trigger compensatory strength in the elbow [26]. Meanwhile, deep sleep is crucial for reducing physical fatigue, but shift workers’ daytime sleep duration conflicts with their body’s sleep wake cycle, often making it difficult to fall asleep, maintain sleep, or sleep calmly. Deep sleep is significantly reduced, making it difficult to feel rested after sleep, thereby exacerbating muscle fatigue [12]. In addition, insufficient sleep can increase pain sensitivity after acute muscle soreness and increase the incidence of WMSDs [27]. The type of circadian rhythm can reflect differences in physiological adaptability, and the morning or evening circadian rhythm affects nurses’ tolerance to shifts [28]。Evidence indicates that distinct sleep-timing phenotypes and divergent sleep-quality scores independently shape both physical exhaustion and psychological burnout. Evening-type nurses, whose endogenous rhythms are phase-delayed, maintain a relatively stable blood-pressure pattern across APN rotations. Nevertheless, chronic shift work still intensifies sleep fragmentation in this group, suppressing nocturnal growth-hormone secretion and blunting overnight muscle repair [29]. 。Frequent bending over is a direct cause of mechanical injury. Employees who frequently bend over to operate the lumbar intervertebral disc will experience an increase in peak pressure, leading to interlaminar shear and fatigue microcracks in the fibrous ring, accelerating intervertebral disc degeneration [30]. Night-shift nap duration represents a critical short-term recovery window. Brief on-duty naps can transiently restore attention [31]. The observed ranking unveils a psycho-physiological-behavioral interplay underlying WMSDs. The top three determinants (sense of agency, social support and sleep quality) are amenable to managerial interventions such as flexible scheduling and team-based collaboration training, whereas the latter three require ergonomic redesign coupled with individual behavioral counseling. Future research can focus on the matching degree between circadian rhythm types and shift patterns to develop precise protective strategies.
Predictor profile of WMSDs among shift-working nurses
Personal-lifestyle domain
The development of WMSDs in shift-working nurses is closely linked to personal lifestyle. First, dairy-product intake frequency serves as a proxy for calcium-homeostasis status: low-frequency consumption reduces intestinal calcium absorption efficiency, thereby impairing skeletal-muscle excitation–contraction coupling and destabilizing bone-mineral density [31]. Calcium imbalance accelerates lumbar-disc degeneration, a risk amplified in night-shift nurses whose irregular meal patterns favor insufficient calcium intake [32]. Daily water intake governs muscle-metabolite clearance. Inadequate hydration slows the removal of catabolites such as lactate, promoting fatigue accumulation and low-grade intramuscular inflammation [33]. Moreover, dehydration diminishes synovial fluid secretion within the carpal tunnel, facilitating the onset of carpal-tunnel syndrome among nurses performing repetitive wrist movements [34]. Waist circumference is closely linked to metabolic inflammation. Adult abdominal obesity indicates visceral fat accumulation; pro-inflammatory cytokines such as IL-6 and TNF-α released from visceral adipose tissue enter the circulation and stimulate skeletal-muscle fibrosis, thereby increasing the incidence of chronic lumbar-strain injuries [35]. Sleep-onset time reflects circadian phase alignment. Nurses who fall asleep after midnight exhibit a delayed nocturnal-melatonin peak, shortening the duration of deep sleep and compressing the critical window for muscle repair [36]. The PSQI quantifies fatigue accumulation. Higher global PSQI scores predict faster declines in muscular endurance, because reduced slow-wave sleep suppresses ultrastructural myofibre restoration [37]. When both difficulty initiating sleep and early-morning awakening are present, the frequency of neck-and-shoulder pain increases 2.3-fold [38].
Occupational and shift-related characteristics
The development of WMSDs in shift-working nurses is intimately linked to job-design and rostering factors. First, career-choice motivation matters: nurses who entered the profession with low intrinsic identification exhibit markedly higher burnout rates [39]. Because their tolerance for repetitive manual tasks is reduced, these individuals appear to amplify musculoskeletal strain through sustained psychophysiological stress pathways that increase resting muscle tension [40]. Second, years of tenure and compensatory capacity play a role. Although long-term nursing workers have improved operational efficiency due to accumulated experience, previous studies have shown that their prevalence of WMSDs is actually higher, suggesting the cumulative effect of chronic injuries [41]. This paradox is likely mediated by age-related declines in muscle-repair capacity and the persistence of previously unresolved micro-injuries that reduce the effectiveness of compensatory movement patterns [42]. Finally, roster design and night-shift workload are critical. High night-shift density, insufficient catch-up sleep after nights, and fewer recovery days each independently increase WMSD risk, largely via circadian disruption and progressive muscle-fatigue accumulation [43]. Inadequate night staffing forces individual nurses to perform multiple, physically demanding tasks without assistance, heightening both psychological strain and biomechanical loading [44]. Moreover, compressed rosters that shorten inter-shift intervals prevent full muscular recovery and further amplify musculoskeletal injury [45]. Therefore, when implementing APN continuous rosters, hospitals should strictly enforce adequate rest intervals between night shifts, guarantee sufficient post-night recovery days and catch-up sleep duration to attenuate muscle fatigue, and provide intensive biomechanics training during orientation to establish protective movement patterns that minimize cumulative musculoskeletal strain [46].
Occupational biomechanical exposures
Prolonged neck flexion is a ubiquitous precipitant of WMSDs in nurses. Tasks such as drug preparation, intravenous insertion and electronic charting oblige staff to maintain a head-down posture for extended periods, continuously loading the cervical and periscapular muscles and precipitating myofascial fatigue and micro-injury [47]. Chronic compression of the posterior disc compartments also accelerates cervical-spondylotic change, further compromising musculoskeletal health [48]. Exposure to vibrating tools, though less frequent, constitutes an additional hazard. In departments such as neurology or rehabilitation, nurses may operate or assist with vibration-based therapeutic devices; transmitted oscillations increase forearm and hand muscular loading and have been associated with carpal-tunnel and elbow-tendon pathologies [49]. Suggest implementing posture management and health intervention programs: nurses should be prompted to adjust working posture at regular intervals, with provision of brief, scheduled neck-relief breaks, while mandatory, progressive exercise sessions emphasizing core-muscle strengthening are introduced to enhance segmental stability and mitigate cervical load.
Comparison of machine-learning models for predicting WMSDs in shift-working nurses
Using the median AUC across repeated cross-validation folds attenuates over-fitting and provides a more conservative estimate of generalisability [50]. Therefore, we adopted median AUC as the primary metric for model comparison. After 100 repetitions of stratified 10-fold cross-validation, the random-forest (RF) algorithm achieved the highest median AUC (0.919) among the seven candidate models, corroborating the validity of the 11 LASSO-selected predictors and offering a practical decision-support tool for mitigating shift-work-related musculoskeletal morbidity. RF aggregates an ensemble of classification trees by bootstrapping both observations and features, enabling robust handling of high-dimensional, heterogeneous data [51]. In the present context—where WMSD risk is jointly shaped by lifestyle habits, shift schedules, workload and ergonomic exposures—RF efficiently captures complex, non-linear interactions among these domains while remaining resilient to outliers, multicollinearity and class imbalance [52]. Consequently, RF provides a reliable and readily deployable predictive framework for occupational health teams seeking to target high-risk shift-working nurses with tailored interventions.
However, several limitations should be acknowledged. First, the cross-sectional design, while effective in revealing the current status of WMSDs and associated factors among shift-working nurses, precludes the establishment of causal relationships between variables. Second, defining WMSDs as the presence of symptoms in at least one body region may have led to an inflated prevalence estimate. Finally, this study only conducted internal validation and lacked independent external samples to further test the performance of the model. The generalization ability of the model in heterogeneous populations is still unclear. Future research should consider employing a prospective cohort design to elucidate the temporal sequence and dynamic evolution of the relationships between various factors and WMSD onset among shift-working nurses. Additionally, subsequent studies could refine the ascertainment of WMSDs by incorporating multidimensional indicators, including symptom severity, functional impairment, and clinical diagnosis, to enhance the precision of outcome definition.
Conclusion
Previous research has focused predominantly on ergonomic stressors, largely overlooking how shift schedules per se influence WMSD risk among nurses. By simultaneously evaluating lifestyle, shift-related and biomechanical exposures, and by benchmarking seven machine-learning algorithms, the present study delivers an evidence-based, multi-factorial framework for safeguarding nursing workforce health. Finally, the current investigation is cross-sectional, although it delineates the prevalence of WMSDs and their association with shift patterns, causal inference cannot be drawn. Future research may consider adopting a longitudinal design to dynamically evaluate the impact of shift cycles and speeds on the occurrence and development of WMSDs, and conduct comprehensive interdisciplinary applied research.
Acknowledgements
We are grateful to the Second Affiliated Hospital of Guangxi Medical University for its guidance and support throughout this study. We also acknowledge the funding agency for its financial assistance and sincerely thank all participants for their valuable cooperation in the survey.
Abbreviations
- WMSDs
Work-related musculoskeletal disorders
- PSQI
Pittsburgh Sleep Quality Index
- NMQ-E
Nordic Musculoskeletal Questionnaire
- MEQ-5
Morningness–Eveningness Questionnaire-5
- PSSS
Perceived Social Support Scale
- SoAS
Sense of Agency Scale
- LDA
Linear discriminant analysis
- PLS
Partial least squares
- RDA
Regularized discriminant analysis
- GLM
Generalized linear model
- RF
Random forest
- SVM-Radial
Radial-kernel support-vector machine
- SVM-Linear
Linear-kernel SVM
- MSE
Mean-squared error
- RSS
Residual sum of squares
- RF
Random-forest
Author contributions
L.L., H.W. and X.C. designed the study and performed the machine-learning analyses.Y.L., P.T., X.P., S.L., Q.P., D.H., C.L., and P.D. collected the data.L.L. drafted the initial manuscript.D.Z. and H.H. supervised the project, interpreted the results, and critically revised the manuscript.
Funding
Guangxi Science and Technology Program: Key Research and Development Project (2025FNFN96726), Guangxi Clinical Key Specialty Construction Project (2025070).
Data availability
The datasets used and/or analyzed in the present study are available from the corresponding author on reasonable request.
Declarations
Ethics approval and consent to participate
The study was approved by the Ethics Committee of The Second Affiliated Hospital of Guangxi Medical University (Approval No. 2023-KY-0941) and conducted in accordance with the ethical standards set forth in the appropriate version of the Declaration of Helsinki. Informed consent was obtained from all participants prior to their enrollment in this study.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Li-Chong Lai, Hai-Chen Wu and Xiao-Ying Cao co-first author.
Contributor Information
Dong-Na Zhou, Email: 183444472@qq.com.
Hui-Qiao Huang, Email: hhq@sr.gxmu.edu.cn.
References
- 1.Jørgensen JT, et al. Shift work and incidence of dementia: A Danish Nurse Cohort study. Alzheimer’s Dement J Alzheimer’s Assoc. 2020;16(9):1268–79. [DOI] [PubMed] [Google Scholar]
- 2.Zhou Z, et al. Circadian disruption in cancer hallmarks: Novel insight into the molecular mechanisms of tumorigenesis and cancer treatment. Cancer Lett. 2024;604:217273–217273. [DOI] [PubMed] [Google Scholar]
- 3.Cao Y, et al. Evaluating the impact of shift work on the risk of cardiometabolic disease: A Mendelian randomization study. Nutrition, metabolism, and cardiovascular diseases. NMCD. 2024;34(12):2720–8. [DOI] [PubMed] [Google Scholar]
- 4.Ljevak I, et al. The Effect of Shift Work on the Overall Health Status of Hospital-Employed Nursing Staff in Bosnia and Herzegovina: A Cross-Sectional Study. Psychiatria Danubina. 2021;33(Suppl 4):771–7. [PubMed] [Google Scholar]
- 5.Wang K, et al. The prevalence and risk factors of work-related musculoskeletal disorders among nurses in China: A systematic review and meta-analysis. Int J Nurs Stud. 2024;157:104826–104826. [DOI] [PubMed] [Google Scholar]
- 6.Chen H, et al. Examining the Predictors of Turnover Behavior in Newly Employed Certified Nurse Aides: A Prospective Cohort Study. Saf health work. 2023;14(2):185–92. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Lin SC, et al. Exploring the factors affecting musculoskeletal disorders risk among hospital nurses. PLoS ONE. 2020;15(4):e0231319–0231319. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Mekonnen TH. Work-Related Factors Associated with Low Back Pain Among Nurse Professionals in East and West Wollega Zones, Western Ethiopia, 2017: A Cross-Sectional Study. Pain therapy. 2019;8(2):239–47. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Yao Y, et al. The associations of work style and physical exercise with the risk of work-related musculoskeletal disorders in nurses. Int J Occup Med Environ Health. 2019;32(1):15–24. [DOI] [PubMed] [Google Scholar]
- 10.Yang S, et al. Prevalence and Risk Factors of Work-Related Musculoskeletal Disorders Among Intensive Care Unit Nurses in China. Volume 67. Workplace health & safety; 2019. pp. 275–87. 6. [DOI] [PubMed]
- 11.Freimann T, Pääsuke M, Merisalu E. Work-Related Psychosocial Factors and Mental Health Problems Associated with Musculoskeletal Pain in Nurses: A Cross-Sectional Study. Pain Res Manage. 2016;2016:9361016–9361016. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Chang W, Peng Y. Differences between fixed day shift nurses and rotating and irregular shift nurses in work-related musculoskeletal disorders: A literature review and meta-analysis. J Occup Health. 2021;63(1):e12208–12208. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Okello A, et al. Prevalence and predictors of work-related musculoskeletal disorders among workers of a gold mine in south Kivu, Democratic Republic of Congo. BMC Musculoskelet Disord. 2020;21(1):797–797. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.H Z. Association between shift work and risk of lower extremity osteoarthritis[D]. Wuhan: Huazhong University of Science and Technology; 2021. [Google Scholar]
- 15.Katsifaraki M, et al. Sleep duration mediates abdominal and lower-extremity pain after night work in nurses. Int Arch Occup Environ Health. 2019;92(3):415–22. [DOI] [PubMed] [Google Scholar]
- 16.Ninan J, et al. Prediction of intradialytic hypotension by machine learning: A systematic review. J Nephrol. 2025;38(8):2077–94. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.A-mei H. Study on the present situation and influencing factors of occupational musculoskeletal disorders among nurses in 9 general hospitals in Xinjiang. Xinjiang Medical University; 2017.
- 18.Dawson AP, et al. Development and test-retest reliability of an extended version of the Nordic Musculoskeletal Questionnaire (NMQ-E): a screening instrument for musculoskeletal pain. J pain. 2009;10(5):517–26. [DOI] [PubMed] [Google Scholar]
- 19.Buysse DJ, et al. The Pittsburgh Sleep Quality Index: a new instrument for psychiatric practice and research. Psychiatry Res. 1989;28(2):193–213. [DOI] [PubMed] [Google Scholar]
- 20.Adan A, Almirall H. Horne & Östberg morningness-eveningness questionnaire: A reduced scale. Pers Indiv Differ. 1991;12(3):241–53. [Google Scholar]
- 21.Zimet GD, et al. The Multidimensional Scale of Perceived Social Support. J Pers Assess. 1988;52(1):30–41. [Google Scholar]
- 22.Tapal A, et al. The Sense of Agency Scale: A Measure of Consciously Perceived Control over One’s Mind, Body, and the Immediate Environment. Front Psychol. 2017;8:1552–1552. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Albrecht SC, et al. Association of work-time control with sickness absence due to musculoskeletal and mental disorders: An occupational cohort study. J Occup Health. 2020;62(1):e12181–12181. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Song J, et al. Effect of changes in sleeping behavior on skeletal muscle and fat mass: a retrospective cohort study. BMC Public Health. 2023;23(1):1879–1879. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Alammari MA, et al. Psychosocial and Occupational Factors Associated With Low Back Pain Among Healthcare Professionals: A Systematic Review. Cureus. 2025;17(1):e77426–77426. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Palsson TS, Rubio-Peirotén A, Doménech-García V. Sleep deprivation increases pain sensitivity following acute muscle soreness. Sleep Med. 2023;109:75–81. [DOI] [PubMed] [Google Scholar]
- 27.Hiestand S, et al. Associations between excessive fatigue and pain, sleep, mental-health and work factors in Norwegian nurses. PLoS ONE. 2023;18(4):e0282734–0282734. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Zhao H, et al. Circadian rhythm types and shift work demands shape sleep quality and depressive symptoms in shift-working nurses. Front public health. 2025;13:1667778–1667778. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Zhang Y, et al. Digital circadian and sleep health in individual hospital shift workers: A cross sectional telemonitoring study. EBioMedicine. 2022;81:104121–104121. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Hulshof CTJ, et al. The prevalence of occupational exposure to ergonomic risk factors: A systematic review and meta-analysis from the WHO/ILO Joint Estimates of the Work-related Burden of Disease and Injury. Environ Int. 2021;146:106157–106157. [DOI] [PubMed] [Google Scholar]
- 31.Alkhatatbeh MJ, Khwaileh HN, Abdul-Razzak KK. High prevalence of low dairy calcium intake and association with insomnia, anxiety, depression and musculoskeletal pain in university students from Jordan. Public Health Nutr. 2021;24(7):1778–86. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Elma Ö, et al. Do Nutritional Factors Interact with Chronic Musculoskeletal Pain? A Systematic Review. J Clin Med. 2020;9(3):702. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Yu L, et al. The value of blood lactate and lactate clearance rate in evaluating the prognosis of athletes with heat illness of varying degrees after high-intensity exercise. BMC sports Sci Med rehabilitation. 2025;17(1):7–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Ulbrichtová R, et al. Association of the role of personal variables and nonoccupational risk factors for work-related carpal tunnel syndrome. Cent Eur J Public Health. 2020;28(4):274–8. [DOI] [PubMed] [Google Scholar]
- 35.Wu H, Ballantyne CM. Skeletal muscle inflammation and insulin resistance in obesity. J Clin Investig. 2017;127(1):43–54. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Melone M, et al. Disruption of the circadian rhythm of melatonin: A biomarker of critical illness severity. Sleep Med. 2023;110:60–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Hidaka T, et al. Associations of endurance, muscle strength, and balanced exercise with subjective sleep quality in sedentary workers: A cross-sectional study. Work (Reading Mass). 2024;78(3):761–70. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Li MT et al. The Influence of Sleep Disturbance on Chronic Pain. Current pain and headache reports, 2022. 26(10): pp. 795–804. [DOI] [PubMed]
- 39.Monti M, et al. Association Among Professional Identity, Burnout, and Mental Health in Medical Students: A Cross-Sectional Study. Acad medicine: J Association Am Med Colleges. 2025;100(11):1298–306. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Deng H, He D, Li F. Factors Influencing Job Burnout and Musculoskeletal Disorders among Coal Miners in the Xinjiang Uygur Autonomous Region. Pain research & management, 2021. 2021: pp. 6629807–6629807. [DOI] [PMC free article] [PubMed]
- 41.Sun W, et al. Prevalence of Work-Related Musculoskeletal Disorders among Nurses: A Meta-Analysis. Iran J public health. 2023;52(3):463–75. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Gorce P, Jacquier-Bret J. Work-Related Musculoskeletal Disorder Prevalence by Body Area Among Nurses in Europe: Systematic Review and Meta-Analysis. J Funct morphology Kinesiol. 2025;10(1):66. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Andreadi A et al. Modified Cortisol Circadian Rhythm: The Hidden Toll of Night-Shift Work. International journal of molecular sciences, 2025. 26(5): p. 2090. [DOI] [PMC free article] [PubMed]
- 44.Naserian E, et al. Relationship between mental workload and musculoskeletal disorders with intention to leave service among nurses working at neonatal and pediatric departments: a cross-sectional study in Iran. BMC Nurs. 2024;23(1):438–438. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Durak Z, Mutlu O. Home health care nurse routing and scheduling problem considering ergonomic risk factors. Heliyon. 2023;10(1):e23896–23896. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Abdollahi T, et al. Effect of an Ergonomics Educational Program on Musculoskeletal Disorders in Nursing Staff Working in the Operating Room: A Quasi-Randomized Controlled Clinical Trial. Int J Environ Res Public Health. 2020;17(19):7333. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Kumagai G, et al. The effect of low back pain and neck-shoulder stiffness on health-related quality of life: a cross-sectional population-based study. BMC Musculoskelet Disord. 2021;22(1):14–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48.Xu F, et al. Magnitude, temporal trend and inequality in burden of neck pain: an analysis of the Global Burden of Disease Study 2019. BMC Musculoskelet Disord. 2025;26(1):202–202. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49.Charles LE, et al. Vibration and Ergonomic Exposures Associated With Musculoskeletal Disorders of the Shoulder and Neck. Saf health work. 2018;9(2):125–32. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50.Li R, et al. Comprehensive Evaluation of Machine Learning Models and Gene Expression Signatures for Prostate Cancer Prognosis Using Large Population Cohorts. Cancer Res. 2022;82(9):1832–43. [DOI] [PubMed] [Google Scholar]
- 51.Hu J, Szymczak S. A review on longitudinal data analysis with random forest. Brief Bioinform. 2023;24(2):bbad002. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 52.Albaradei S, et al. Machine learning and deep learning methods that use omics data for metastasis prediction. Comput Struct Biotechnol J. 2021;19:5008–18. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The datasets used and/or analyzed in the present study are available from the corresponding author on reasonable request.