Abstract
Background
Traditional approaches to assessing sleep quality in clinical nurses often overlook population heterogeneity and the complex interplay of influencing factors. This study employs Latent Profile Analysis (LPA) and Association Rule Mining (ARM) to identify distinct sleep quality subgroups and uncover key factor combinations, thereby informing targeted intervention strategies.
Methods
A total of 1,686 nurses from 123 hospitals in Shandong Province were recruited through multistage stratified sampling. LPA was used to classify participants based on seven sleep dimensions from the Pittsburgh Sleep Quality Index (PSQI), while ARM was applied to identify frequent itemsets of sleep disorder triggers. Key influencing factors were further examined using univariate analysis and multivariate logistic regression.
Results
Three latent sleep profiles were identified: high (63.11%), moderate (34.10%), and low (2.79%) sleep quality. The low-sleep subgroup was characterized by higher proportions of being unmarried/divorced (42.55%), low monthly income (≤ 3,000 CNY, 42.55%), non-permanent employment (76.60%), and severe psychological distress (44.68%). In contrast, the high-sleep subgroup featured higher rates of being married (85.62%), moderate income (3,001–7,000 CNY, 73.03%), and low psychological distress (51.32%). Key determinants included marital status (OR = 2.153/2.252), income (OR = 9.098), employment type (OR = 1.475), and psychological state (OR = 0.060–0.555). ARM revealed distinct risk combinations: “low income + non-permanent employment” (lift = 3.895) for the low-sleep group; “married + moderate income + non-permanent employment + patient conflict” for the moderate group; and “high income + low psychological distress” buffering night-shift effects in the high-sleep group.
Conclusion
By integrating LPA and ARM, this study reveals the multidimensional heterogeneity and interactive mechanisms underlying clinical nurses’ sleep quality. The findings support a stratified intervention framework combining institutional safeguards with precision strategies to enhance sleep health management in nursing populations.
Keywords: Latent profile analysis, Association rule mining, Clinical nurses, Sleep quality, Heterogeneity, Influencing factors
Background
In China’s current healthcare system, clinical nurses persistently face sleep health challenges due to rotating shift schedules. Under predominant three-shift (8-hour shifts) and two-shift (12-hour shifts) systems, nurses undertake 6–10 night shifts monthly [1]. This high-intensity, irregular work pattern renders nurses a high-risk population for sleep quality deterioration [2, 3]. A nationwide large-scale survey encompassing 274,123 nurses from 136 public hospitals across 31 provinces revealed that 46.87% of nurses averaged less than the WHO-recommended 7 h of daily sleep, with 27.63% reporting low subjective sleep satisfaction [4].
Sleep quality is not only critical to nurses’ individual health but also intrinsically linked to occupational risks. Evidence-based medical research demonstrates that chronic poor sleep quality significantly elevates cardiovascular disease risks [5], while correlating strongly with gastrointestinal dysfunction and cognitive impairment [6, 7]. Furthermore, fatigue induced by sleep deprivation increases the likelihood of medication errors, documentation lapses, and other decision-making failures, directly jeopardizing patient safety and care quality [8]. Thus, systematically investigating sleep quality determinants and enabling early identification of sleep disorder risks are pivotal for safeguarding nurses’ well-being and healthcare safety.
However, current methodologies in nurse sleep research exhibit notable limitations. Traditional scale-based assessments oversimplify multidimensional sleep issues into unidimensional continuous variables, obscuring population heterogeneity and masking critical subtype characteristics of sleep disorders. Conventional regression analyses fail to capture nonlinear interactions among multifactorial variables, thereby neglecting clinically meaningful attribute combination rules [9]. To address these gaps, this study establishes a dual-phase analytical framework: Latent Profile Analysis (LPA) identifies latent subgroups based on multidimensional sleep indicators, overcoming dimensional constraints of conventional classifications [10]; Association Rule Mining (ARM) employs the Apriori algorithm to decode frequent itemsets among physiological parameters, job characteristics, and environmental factors, revealing key pathogenic combinations of sleep disorders [11].
This research aims to achieve precise classification of clinical nurses’ sleep status, elucidate differential influencing factors across subgroups, and provide scientific foundations for developing targeted sleep intervention strategies.
Study participants and methods
Study participants
This study adopted a multistage stratified sampling method to select 1,686 nurses from general hospitals in Shandong Province between November 2021 and January 2022. The sampling process involved stratifying all prefecture-level cities in Shandong Province into three categories (high, medium, and low) based on their 2020 per capita GDP. From each category, one city was randomly selected: Qingdao (high GDP tier), Zaozhuang (medium GDP tier), and Dezhou (low GDP tier). In cities with only one municipal general hospital, that hospital was directly included in the survey. For cities with multiple hospitals, one general hospital was randomly chosen. Subsequently, three counties (or districts/county-level cities) were randomly selected from each sampled prefecture-level city, and one general hospital was randomly selected from each county, resulting in a total of 12 hospitals (3 municipal and 9 county-level). Within each hospital, three departments were randomly selected, and all night-shift nurses in those departments participated in the survey. The study protocol was approved by the Ethics Committee of Shandong University School of Public Health (Approval No. 20181219). It should be noted that the data collection period (November 2021 to January 2022) coincided with the COVID-19 pandemic in China. The heightened workload and psychological stress experienced by clinical nurses during this period may have elevated the overall levels of sleep disturbances and psychological distress observed in this study, thereby influencing the results [12, 13].
Research instruments
Demographic questionnaire
A self-designed demographic questionnaire was developed through systematic literature review and adaptation of prior research tools. The questionnaire collected comprehensive information across 21 items, including gender, age, religious affiliation, highest educational attainment, marital status, average monthly income, number of children, years of nursing experience, years employed at the current institution, department affiliation, and average monthly night shifts.
Pittsburgh Sleep Quality Index (PSQI)
The PSQI, originally developed by Buysse et al. and adapted into Chinese by Liu Xianchen et al. [14], is widely used in healthcare populations. The scale comprises 18 self-rated items assessing seven dimensions: subjective sleep quality, sleep latency, sleep duration, sleep efficiency, sleep disturbances, use of hypnotic medications, and daytime dysfunction. Each dimension is scored from 0 to 3, with a total score ranging from 0 to 21. Higher scores indicate poorer sleep quality. In this study, the Cronbach’s α coefficient for the PSQI was 0.879, demonstrating good reliability.
Kessler psychological distress Scale-10 (K10)
The K10, developed by Kessler and Mroczek in 1994 and validated in Chinese by Zhou Chengchao et al. [15], measures psychological distress. The scale includes 10 items rated on a 5-point Likert scale (“1 = almost never” to “5 = always”), with total scores ranging from 10 to 50. Higher scores reflect greater psychological distress. The Cronbach’s α coefficient for the K10 in this study was 0.930, indicating high internal consistency.
Work-family conflict scale (WAFCS)
The WAFCS, developed by Grzywacz et al. [16], evaluates conflicts between work and family responsibilities. The 11-item scale consists of three subscales: work-to-family interference (4 items), family-to-work interference (4 items), and family-to-work facilitation (3 items). Responses are recorded on a 5-point Likert scale (“1 = strongly disagree” to “5 = strongly agree”). The Cronbach’s α coefficient for the WAFCS in this study was 0.947, confirming its reliability.
Perceived social support scale (PSS)
The PSS, originally developed by Zimet et al. and adapted into Chinese by Jiang Qianjin [17], assesses perceived social support. The scale contains 12 items rated on a 7-point Likert scale (“1 = strongly disagree” to “7 = strongly agree”), covering three dimensions: family support, friend support, and support from significant others. Total scores range from 12 to 84, with higher scores indicating stronger perceived social support. The Cronbach’s α coefficient for the PSS in this study was 0.956, demonstrating excellent reliability.
Statistical analysis
Latent class analysis (LCA) was performed using Mplus 8.3 to classify participants’ sleep status. The modeling process began with a single-class model (C1), incrementally increasing the number of latent profiles. Model selection relied on goodness-of-fit and comparative tests, evaluated through the following criteria:1.Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), and sample-size-adjusted BIC (SaBIC), where lower values indicate superior model fit. 2.Entropy index to assess classification accuracy, with values ≥ 0.6 and ≥ 0.8 corresponding to classification accuracy rates exceeding 80% and 90%, respectively. 3.Lo-Mendell-Rubin adjusted likelihood ratio test (LMR) and Bootstrap likelihood ratio test (BLRT), where statistically significant P-values (< 0.05) indicated that the k-class model outperformed the (k-1)-class model.
Data entry and preliminary analyses were conducted in SPSS 25.0. The normality of continuous variables was assessed using the Shapiro-Wilk test and visual inspection of Q-Q plots. Given the large sample size (n = 1,686) and the central limit theorem, the use of t-tests and ANOVA was considered robust. Furthermore, due to mandatory fields in the electronic questionnaire, there were no missing data for the core variables used in this analysis. Categorical variables were summarized as frequencies and percentages (%), while continuous variables were reported as mean ± standard deviation. Given the exploratory nature of the numerous univariate group comparisons, the results should be interpreted with caution, and the primary focus is placed on the subsequent multivariate models and effect sizes. Group comparisons utilized t-tests and analysis of variance (ANOVA). Multivariate logistic regression identified factors influencing sleep classifications, with statistical significance set at P < 0.05.
Association rule mining for sleep status was implemented via the Apriori algorithm in SPSS Modeler 18.0. Categorical variables were analyzed using frequency and percentage (%). χ² tests in SPSS 25.0 evaluated antecedent-consequent relationships in the association rules, with a significance threshold of α = 0.05.
Results
Sociodemographic characteristics of participants
The study included 1,686 clinical nurses. The majority were female (95.67%, n = 1,613), with male nurses comprising 4.33% (n = 73). Age distribution showed the largest proportion (44.42%, n = 749) in the 30–39 age group, followed by 20–29-year-olds (36.36%, n = 613). A significant majority (96.74%) reported no religious affiliation. Education levels were predominantly bachelor’s degrees (79.83%, n = 1,346), and most participants were married (81.79%, n = 1,379). Regarding income, 43.71% (n = 737) earned between 3,001 and 5,000 CNY monthly. Over half (52.91%) had one child.
Occupational characteristics revealed that 31.55% had 6–10 years of nursing experience, while 48.10% worked in tertiary Grade A hospitals. Nearly half (46.80%) reported experiencing verbal or physical abuse from patients or their families. Psychologically, 37.96% exhibited low psychological distress, whereas 35.23% reported high work-family conflict. Detailed data are presented in Table 1.
Table 1.
Sociodemographic information of study participants
| Variable | Frequency | Percentage |
|---|---|---|
| Gender | ||
| Male | 73 | 4.33% |
| Female | 1613 | 95.67% |
| Age | ||
| 20–29 | 613 | 36.36% |
| 30–39 | 749 | 44.42% |
| 40+ | 324 | 19.22% |
| Religious Belief | ||
| No | 1631 | 96.74% |
| Yes | 55 | 3.26% |
| Highest Education Level | ||
| Secondary Vocational School | 22 | 1.30% |
| Junior College | 285 | 16.90% |
| Bachelor’s Degree | 1346 | 79.83% |
| Master’s Degree or Above | 33 | 1.96% |
| Marital Status | ||
| Unmarried | 288 | 17.08% |
| Married | 1379 | 81.79% |
| Others | 19 | 1.13% |
| Average Monthly Income | ||
| <3000 RMB | 203 | 12.04% |
| 3001–5000 RMB | 737 | 43.71% |
| 5001–7000 RMB | 511 | 30.31% |
| 7001–9000 RMB | 175 | 10.38% |
| >9000 RMB | 60 | 3.56% |
| Number of Children | ||
| No Children | 486 | 28.83% |
| One | 892 | 52.91% |
| Two or More | 308 | 18.27% |
| Years in Nursing Profession | ||
| 0–5 Years | 452 | 26.81% |
| 6–10 Years | 532 | 31.55% |
| 11–15 Years | 276 | 16.37% |
| 16–20 Years | 173 | 10.26% |
| 20 + Years | 253 | 15.01% |
| Years in Current Institution | ||
| 0–5 Years | 553 | 32.80% |
| 6–10 Years | 507 | 30.07% |
| 11–15 Years | 251 | 14.89% |
| 16–20 Years | 152 | 9.02% |
| 20 + Years | 223 | 13.23% |
| Department | ||
| Internal Medicine | 454 | 26.93% |
| Surgery | 385 | 22.84% |
| Obstetrics and Gynecology | 304 | 18.03% |
| Pediatrics | 223 | 13.23% |
| Others | 320 | 18.98% |
| Professional Title | ||
| No Title | 76 | 4.51% |
| Technician Level | 214 | 12.69% |
| Junior Professional (Initial) | 814 | 48.28% |
| Intermediate Level | 499 | 29.60% |
| Associate Senior or Above | 83 | 4.92% |
| Administrative Position | ||
| No | 1404 | 83.27% |
| Yes | 282 | 16.73% |
| On-the-Job Staff Status | ||
| No | 1056 | 62.63% |
| Yes | 630 | 37.37% |
| Hospital Level | ||
| Second-Class B and Below | 679 | 40.27% |
| Third-Class B | 196 | 11.63% |
| Third-Class A | 811 | 48.10% |
| Average Weekly Work Hours | ||
| ≤ 40 h | 779 | 46.20% |
| 41–50 h | 679 | 40.27% |
| > 50 h | 228 | 13.52% |
| Monthly Night Shifts | ||
| ≤ 7 Days | 1174 | 69.63% |
| > 7 Days | 512 | 30.37% |
| Perceived Respect from Patients | ||
| Disrespected | 96 | 5.69% |
| Neutral | 725 | 43.00% |
| Respected | 865 | 51.30% |
| Experienced Patient/Visitor Violence | ||
| No | 897 | 53.20% |
| Yes | 789 | 46.80% |
| Current Work Environment | ||
| Poor | 355 | 21.06% |
| Moderate | 844 | 50.06% |
| Good | 487 | 28.88% |
| Medical Errors in Past Year | ||
| No | 1641 | 97.33% |
| Yes | 45 | 2.67% |
| Seriously Considered Suicide | ||
| Yes | 155 | 9.19% |
| No | 1531 | 90.81% |
| Psychological Distress | ||
| Low Distress | 640 | 37.96% |
| Moderate Distress | 460 | 27.28% |
| High Distress | 322 | 19.10% |
| Severe Distress | 264 | 15.66% |
| Work-Family Conflict | ||
| Low Conflict | 282 | 16.73% |
| Moderate Conflict | 810 | 48.04% |
| High Conflict | 594 | 35.23% |
| Social Support | ||
| Low Support | 82 | 4.86% |
| Moderate Support | 594 | 35.23% |
| High Support | 1010 | 59.91% |
Scores across sleep status dimensions
The mean scores for sleep quality and sleep efficiency were 0.11 and 0.53, respectively, while the remaining five sleep dimensions (sleep latency, sleep duration, sleep disturbances, use of hypnotic medications, and daytime dysfunction) had mean scores ranging between 1.0 and 1.2. Notably, the standard deviations for all dimensions exceeded half of their corresponding mean values, indicating substantial heterogeneity in participants’ responses across the seven sleep dimensions. Detailed results are presented in Table 2.
Table 2.
Scores for dimensions of sleep quality
| Dimension | Mean | Median | Std. Deviation | Minimum | Maximum |
|---|---|---|---|---|---|
| Sleep Quality | 0.11 | 0 | 0.46 | 0 | 3 |
| Sleep Latency | 1.20 | 1 | 0.89 | 0 | 3 |
| Sleep Duration | 1.05 | 1 | 0.82 | 0 | 3 |
| Sleep Efficiency | 0.53 | 0 | 0.86 | 0 | 3 |
| Sleep Disturbances | 1.04 | 1 | 0.62 | 0 | 3 |
| Use of Sleep Medications | 1.09 | 1 | 1.08 | 0 | 3 |
| Daytime Dysfunction | 1.20 | 1 | 0.91 | 0 | 3 |
Latent profile analysis of sleep status
Starting with a single-profile model, the number of latent profiles was incrementally increased to identify the optimal model. As shown in the model fit indices (Table 3), the AIC, BIC, and SaBIC values decreased progressively with additional profiles. The three-profile model demonstrated an entropy value of 0.889, with both the Lo-Mendell-Rubin (LMR) and Bootstrap Likelihood Ratio Test (BLRT) yielding statistically significant P-values (P < 0.001). Although the four-profile model achieved a slightly higher entropy (0.893), its LMR P-value (0.084) lacked statistical significance. The five-profile model, despite having the highest entropy (0.926) and significant LMR/BLRT results (P < 0.001), included one profile with zero participants, rendering it impractical. Based on parsimony and interpretability, the three-profile model was selected as optimal. The three profiles comprised: Class 1 (n = 1,064, 63.11%*): Characterized by the highest overall sleep quality. Class 2 (n = 575, 34.10%*): Exhibiting moderate sleep quality. Class 3 (n = 47, 2.79%*): Demonstrating the poorest sleep quality. These groups were labeled as the High Sleep Status Group, Moderate Sleep Status Group, and Low Sleep Status Group, respectively. Mean scores for each class across the seven sleep dimensions are detailed in Table 4 and visualized in Fig. 1.
Table 3.
Latent profile analysis (LPA) indicators of sleep quality
| Classes | AIC | BIC | SaBIC | Entropy | LMR | BLRT | Class Counts/Proportions |
|---|---|---|---|---|---|---|---|
| 1 | 27610.99 | 27687.01 | 27642.53 | ||||
| 2 | 25452.36 | 25571.82 | 25501.93 | 0.828 | < 0.001 | < 0.001 | 1087 (64.47%)/599 (35.53%) |
| 3 | 23289.68 | 23452.58 | 23357.28 | 0.889 | < 0.001 | < 0.001 | 1064 (63.11%)/575 (34.10%)/47 (2.79%) |
| 4 | 22512.82 | 22719.16 | 22598.44 | 0.893 | 0.084 | < 0.001 | 949 (56.29%)/432 (25.62%)/258 (15.30%)/47 (2.79%) |
| 5 | 20744.58 | 20994.36 | 20848.23 | 0.926 | < 0.001 | < 0.001 | 0 (0.00%)/1036 (61.45%)/47 (2.79%)/68 (4.03%)/535 (31.73%) |
Table 4.
Univariate analysis of sleep quality latent profiles
| Variable | High Sleep Status Group (n = 1064) |
Moderate Sleep Status Group(n = 575) | Low Sleep Status Group(n = 47) | Statistic | P-Value |
|---|---|---|---|---|---|
| Gender | 4.868 | 0.088 | |||
| Male | 46(4.32%) | 22(3.83%) | 5(10.64%) | ||
| Female | 1018(95.68%) | 553(96.17%) | 42(89.36%) | ||
| Age | 9.640b | 0.047 | |||
| 20–29 | 382(35.9%) | 205(35.65%) | 26(55.32%) | ||
| 30–39 | 468(43.98%) | 268(46.61%) | 13(27.66%) | ||
| 40+ | 214(20.11%) | 102(17.74%) | 8(17.02%) | ||
| Religious Belief | 4.306 | 0.116 | |||
| No | 1036(97.37%) | 551(95.83%) | 44(93.62%) | ||
| Yes | 28(2.63%) | 24(4.17%) | 3(6.38%) | ||
| Highest Education Level | 3.471 | 0.748 | |||
| Secondary Vocational School | 13(1.22%) | 8(1.39%) | 1(2.13%) | ||
| Junior College | 187(17.58%) | 87(15.13%) | 11(23.4%) | ||
| Bachelor’s Degree | 843(79.23%) | 469(81.57%) | 34(72.34%) | ||
| Master’s Degree or Above | 21(1.97%) | 11(1.91%) | 1(2.13%) | ||
| Marital Status | 38.624 | < 0.001 | |||
|
Unmarried/Divorced/ Widowed |
153(14.38%) | 133(23.13%) | 20(42.55%) | ||
| Married | 911(85.62%) | 442(76.87%) | 27(57.45%) | ||
| Average Monthly Income | 121.928 | < 0.001 | |||
| <3000 RMB | 97(9.12%) | 86(14.96%) | 20(42.55%) | ||
| 3001–5000 RMB | 409(38.44%) | 314(54.61%) | 14(29.79%) | ||
| 5001–7000 RMB | 368(34.59%) | 133(23.13%) | 10(21.28%) | ||
| 7001–9000 RMB | 143(13.44%) | 30(5.22%) | 2(4.26%) | ||
| >9000 RMB | 47(4.42%) | 12(2.09%) | 1(2.13%) | ||
| Number of Children | 25.148b | < 0.001 | |||
| No Children | 320(30.08%) | 140(24.35%) | 26(55.32%) | ||
| One | 563(52.91%) | 312(54.26%) | 17(36.17%) | ||
| Two or More | 181(17.01%) | 123(21.39%) | 4(8.51%) | ||
| Years in Nursing Profession | 13.025 | 0.111 | |||
| 0–5 Years | 286(26.88%) | 146(25.39%) | 20(42.55%) | ||
| 6–10 Years | 328(30.83%) | 193(33.57%) | 11(23.4%) | ||
| 11–15 Years | 172(16.17%) | 96(16.7%) | 8(17.02%) | ||
| 16–20 Years | 111(10.43%) | 62(10.78%) | 0(0.0%) | ||
| 20 + Years | 167(15.7%) | 78(13.57%) | 8(17.02%) | ||
| Years in Current Institution | 10.595 | 0.226 | |||
| 0–5 Years | 351(32.99%) | 181(31.48%) | 21(44.68%) | ||
| 6–10 Years | 311(29.23%) | 186(32.35%) | 10(21.28%) | ||
| 11–15 Years | 160(15.04%) | 83(14.43%) | 8(17.02%) | ||
| 16–20 Years | 97(9.12%) | 55(9.57%) | 0(0.0%) | ||
| 20 + Years | 145(13.63%) | 70(12.17%) | 8(17.02%) | ||
| Department | 3.184b | 0.922 | |||
| Internal Medicine | 285(26.79%) | 155(26.96%) | 14(29.79%) | ||
| Surgery | 249(23.4%) | 123(21.39%) | 13(27.66%) | ||
| Obstetrics and Gynecology | 186(17.48%) | 110(19.13%) | 8(17.02%) | ||
| Pediatrics | 137(12.88%) | 81(14.09%) | 5(10.64%) | ||
| Others | 207(19.45%) | 106(18.43%) | 7(14.89%) | ||
| Professional Title | 29.853 | < 0.001 | |||
| No Title | 47(4.42%) | 23(4.0%) | 6(12.77%) | ||
| Technician Level | 138(12.97%) | 62(10.78%) | 14(29.79%) | ||
| Junior Professional (Initial) | 500(46.99%) | 301(52.35%) | 13(27.66%) | ||
| Intermediate Level | 319(29.98%) | 168(29.22%) | 12(25.53%) | ||
| Associate Senior or Above | 60(5.64%) | 21(3.65%) | 2(4.26%) | ||
| Administrative Position | 35.823 | < 0.005 | |||
| No | 842(79.14%) | 521(90.61%) | 41(87.23%) | ||
| Yes | 222(20.86%) | 54(9.39%) | 6(12.77%) | ||
| On-the-Job Staff Status | 41.426 | < 0.001 | |||
| No | 605(56.86%) | 415(72.17%) | 36(76.60%) | ||
| Yes | 459(43.14%) | 160(27.83%) | 11(23.40%) | ||
| Hospital Level | 5.314b | 0.257 | |||
| Second-Class B and Below | 447(42.01%) | 213(37.04%) | 19(40.43%) | ||
| Third-Class B | 120(11.28%) | 73(12.7%) | 3(6.38%) | ||
| Third-Class A | 497(46.71%) | 289(50.26%) | 25(53.19%) | ||
| Average Weekly Work Hours | 8.767b | 0.067 | |||
| ≤ 40 h | 519(48.78%) | 240(41.74%) | 20(42.55%) | ||
| 41–50 h | 410(38.53%) | 251(43.65%) | 18(38.3%) | ||
| > 50 h | 135(12.69%) | 84(14.61%) | 9(19.15%) | ||
| Monthly Night Shifts | 59.943 | < 0.001 | |||
| ≤ 7 Days | 811(76.22%) | 333(57.91%) | 30(63.83%) | ||
| > 7 Days | 253(23.78%) | 242(42.09%) | 17(36.17%) | ||
| Perceived Respect from Patients | 6.717 | 0.152 | |||
| Disrespected | 56(5.26%) | 38(6.61%) | 2(4.26%) | ||
| Neutral | 453(42.58%) | 244(42.43%) | 28(59.57%) | ||
| Respected | 555(52.16%) | 293(50.96%) | 17(36.17%) | ||
| Experienced Patient/Visitor Violence | 45.908 | < 0.001 | |||
| No | 633(59.49%) | 245(42.61%) | 19(40.43%) | ||
| Yes | 431(40.51%) | 330(57.39%) | 28(59.57%) | ||
| Current Work Environment | 2.123 | 0.713 | |||
| Poor | 216(20.30%) | 127(22.09%) | 12(25.53%) | ||
| Moderate | 542(50.94%) | 278(48.35%) | 24(51.06%) | ||
| Good | 306(28.76%) | 170(29.57%) | 11(23.40%) | ||
| Medical Errors in Past Year | 0.732 | 0.693 | |||
| No | 1038(97.56%) | 557(96.87%) | 46(97.87%) | ||
| Yes | 26(2.44%) | 18(3.13%) | 1(2.13%) | ||
| Psychological Distress | 323.654b | < 0.001 | |||
| Low Distress | 546(51.32%) | 88(15.3%) | 6(12.77%) | ||
| Moderate Distress | 295(27.73%) | 151(26.26%) | 14(29.79%) | ||
| High Distress | 146(13.72%) | 170(29.57%) | 6(12.77%) | ||
| Severe Distress | 77(7.24%) | 166(28.87%) | 21(44.68%) | ||
| Work-Family Conflict | 108.674b | < 0.001 | |||
| Low Conflict | 231(21.71%) | 43(7.48%) | 8(17.02%) | ||
| Moderate Conflict | 546(51.32%) | 249(43.3%) | 15(31.91%) | ||
| High Conflict | 287(26.97%) | 283(49.22%) | 24(51.06%) | ||
| Social Support | 31.597 | < 0.001 | |||
| Low Support | 49(4.61%) | 29(5.04%) | 4(8.51%) | ||
| Moderate Support | 325(30.55%) | 251(43.65%) | 18(38.3%) | ||
| High Support | 690(64.85%) | 295(51.3%) | 25(53.19%) |
Fig. 1.
Classification of sleep status types based on latent profile
It is noteworthy that, for contextual comparison, the traditional Pittsburgh Sleep Quality Index (PSQI) global score cutoff of > 7 identified 27.63% (n = 466) of our sample as having poor sleep quality. This prevalence provides a benchmark against which the more nuanced, data-driven latent profile analysis (LPA) can be contrasted. The LPA approach, while identifying a smaller proportion of nurses in the ‘Low Sleep Status’ group (2.79%), further stratified the majority of nurses previously lumped into a ‘good sleeper’ category (PSQI ≤ 7) into distinct ‘High’ (63.11%) and ‘Moderate’ (34.10%) sleep quality subgroups, thereby revealing substantial heterogeneity that is obscured by the unitary PSQI cutoff.
Univariate analysis of latent sleep status profiles
Significant differences (P < 0.05) were observed among the three sleep status groups across multiple sociodemographic and occupational variables, including age, marital status, average monthly income, number of children, professional rank, administrative roles, employment status (permanent vs. contract), night shift frequency, experiences of verbal/physical abuse from patients or families, psychological distress, work-family conflict, and perceived social support. Full comparative data are provided in Table 4.
Multivariate logistic regression analysis of factors influencing sleep status groups
Using the high sleep status group as the reference, multivariate logistic regression models were constructed for the moderate and low sleep status groups. Significant predictors included marital status, average monthly income, number of children, administrative roles, employment status (permanent vs. contract), monthly night shift frequency, experiences of verbal/physical abuse from patients or families, psychological distress, and work-family conflict. In the moderate sleep status group, significant risk factors included being unmarried/divorced/widowed (OR = 2.153, 95% CI: 0.019–0.415), absence of administrative roles (OR = 1.833, 95% CI: 1.262–2.662), and contract employment (OR = 1.475, 95% CI: 1.127–1.929). Protective factors were having no children (OR = 0.339, 95% CI: 0.225–0.512), night shifts ≤ 7 days/month (OR = 0.604, 95% CI: 0.467–0.782), absence of abuse from patients/families (OR = 0.668, 95% CI: 0.524–0.852), and lower psychological distress and work-family conflict. For the low sleep status group, key risk factors were being unmarried/divorced/widowed (OR = 2.252, 95% CI: 1.015–4.996) and monthly income < 3,000 CNY (OR = 9.098, 95% CI: 1.034–80.036), while lower psychological distress and work-family conflict remained protective. Full regression results are detailed in Table 5.
Table 5.
Multivariate logistic regression models for sleep quality latent profiles
| Variable | Class 2 | Class3 | ||||
|---|---|---|---|---|---|---|
| OR | 95% CI | p | OR | 95% CI | p | |
|
Marital Status (ref: Married) |
||||||
|
Unmarried/Divorced /Widowed |
2.153 | 1.472–3.148 | < 0.001 | 2.252 | 1.015–4.996 | 0.046 |
|
Average Monthly Income (ref: >9000RMB/month) |
||||||
| <3000 RMB | 2.034 | 0.914–4.525 | 0.082 | 9.098 | 1.034–80.036 | 0.047 |
| 3001–5000 RMB | 1.826 | 0.875–3.810 | 0.109 | 1.695 | 0.196–14.622 | 0.631 |
| 5001–7000 RMB | 0.956 | 0.454–2.015 | 0.906 | 1.393 | 0.160-12.145 | 0.764 |
| 7001–9000 RMB | 0.520 | 0.227–1.190 | 0.122 | 0.663 | 0.055–7.982 | 0.746 |
|
Number of Children (ref: Two or More) |
||||||
| No Children | 0.339 | 0.225–0.512 | < 0.001 | 2.001 | 0.602–6.650 | 0.258 |
| One | 0.847 | 0.616–1.165 | 0.307 | 1.537 | 0.494–4.780 | 0.458 |
| Administrative Position(No) | 1.833 | 1.262–2.662 | 0.001 | 0.871 | 0.330–2.302 | 0.780 |
| On-the-Job Staff Status (No) | 1.475 | 1.127–1.929 | 0.005 | 1.291 | 0.607–2.746 | 0.508 |
|
Monthly Night Shifts (≤ 7 Days) |
0.604 | 0.467–0.782 | < 0.001 | 0.847 | 0.434–1.652 | 0.625 |
| Experienced Patient/Visitor Violence (No) | 0.668 | 0.524–0.852 | 0.001 | 0.609 | 0.317–1.171 | 0.137 |
| Psychological Distress(ref: Severe Distress) | ||||||
| Low Distress | 0.099 | 0.067–0.145 | < 0.001 | 0.060 | 0.022–0.165 | < 0.001 |
| Moderate Distress | 0.269 | 0.186–0.387 | < 0.001 | 0.216 | 0.099–0.469 | < 0.001 |
| High Distress | 0.555 | 0.379–0.813 | 0.002 | 0.170 | 0.064–0.456 | < 0.001 |
| Work-Family Conflict(ref: High Conflict) | ||||||
| Low Conflict | 0.329 | 0.217–0.497 | < 0.001 | 0.657 | 0.266–1.623 | 0.362 |
| Moderate Conflict | 0.662 | 0.511–0.856 | 0.002 | 0.439 | 0.216–0.892 | 0.023 |
Association rule analysis of clinical nurses’ sleep status
Association rule mining revealed distinct characteristic combinations for each sleep status group. The high sleep status group exhibited the highest-confidence rules, including: {monthly income = 7,001–9,000 CNY, night shifts > 7 days/month, low psychological distress}, {monthly income < 3,000 CNY, no children, low work-family conflict}, and {monthly income < 3,000 CNY, low psychological distress, low work-family conflict}. The moderate sleep status group was predominantly associated with combinations such as: {married, monthly income = 3,001–5,000 CNY, contract employment, experienced abuse, high psychological distress, moderate work-family conflict}; {married, monthly income = 3,001–5,000 CNY, no administrative role, contract employment, experienced abuse, high psychological distress, moderate work-family conflict}; and {no administrative role, night shifts > 7 days/month, no abuse, very high psychological distress, high work-family conflict}. For the low sleep status group, the most frequent rules centered on low income: {monthly income < 3,000 CNY, contract employment}, {monthly income < 3,000 CNY, no administrative role, contract employment}, and {monthly income < 3,000 CNY}. These findings highlight socioeconomic vulnerability (e.g., low income, contract employment) and occupational stressors (e.g., lack of administrative roles) as critical markers of poor sleep outcomes. Detailed rules are presented in Table 6; Fig. 2.
Table 6.
Association rule analysis of sleep quality in clinical Nurses
| Group | Order | ruler | support | confidence | coverage | lift | count |
|---|---|---|---|---|---|---|---|
| High Sleep Status | 1 | {Average Monthly Income = 7001–9000 RMB, Monthly Night Shifts ≥ 7 days, Psychological Distress Level (K10 Group) = Low Distress} | 0.011 | 1.000 | 0.011 | 1.585 | 19 |
| 2 |
{Average Monthly Income<3000 RMB, Number of Children = No Children, Work-Family Conflict = Low Conflict} |
0.011 | 1.000 | 0.011 | 1.585 | 19 | |
| 3 | {Average Monthly Income<3000 RMB, Psychological Distress Level (K10 Group) = Low Distress, Work-Family Conflict = Low Conflict} | 0.010 | 1.000 | 0.010 | 1.585 | 17 | |
| 4 | {Number of Children = No Children, Administrative Position = Yes, Work-Family Conflict = Moderate Conflict} | 0.011 | 1.000 | 0.011 | 1.585 | 19 | |
| 5 | {Marital Status = Unmarried/Divorced/Widowed, Psychological Distress Level (K10 Group) = Low Distress, Work-Family Conflict = Low Conflict} | 0.011 | 1.000 | 0.011 | 1.585 | 19 | |
| Moderate Sleep Status | 1 | {Marital Status = Married, Average Monthly Income = 3001–5000 RMB, On-the-Job Staff Status = No, Experienced Patient/Visitor Violence = Yes, Psychological Distress Level (K10 Group) = High Distress, Work-Family Conflict = Moderate Conflict} | 0.012 | 0.952 | 0.012 | 2.793 | 20.000 |
| 2 | {Marital Status = Married, Average Monthly Income = 3001–5000 RMB, Administrative Position = No, On-the-Job Staff Status = No, Experienced Patient/Visitor Violence = Yes, Psychological Distress Level (K10 Group) = High Distress, Work-Family Conflict = Moderate Conflict} | 0.011 | 0.950 | 0.012 | 2.786 | 19.000 | |
| 3 | { Administrative Position = No, Monthly Night Shifts ≥ 7 days, Experienced Patient/Visitor Violence = No, Psychological Distress Level (K10 Group) = Severe Distress, Work-Family Conflict = High Conflict} | 0.011 | 0.947 | 0.011 | 2.778 | 18.000 | |
| 4 | {Average Monthly Income = 5001–7000 RMB, Psychological Distress Level (K10 Group) = Severe Distress, Work-Family Conflict = High Conflict} | 0.015 | 0.929 | 0.017 | 2.723 | 26.000 | |
| 5 | {Average Monthly Income = 5001–7000 RMB, Administrative Position = No, Psychological Distress Level (K10 Group) = Severe Distress, Work-Family Conflict = High Conflict} | 0.014 | 0.923 | 0.015 | 2.707 | 24.000 | |
| Low Sleep Status | 1 | {Average Monthly Income<3000 RMB, On-the-Job Staff Status = No} | 0.011 | 0.109 | 0.104 | 3.895 | 19 |
| 2 | {Average Monthly Income<3000 RMB, Administrative Position = No, On-the-Job Staff Status = No} | 0.011 | 0.108 | 0.098 | 3.890 | 18 | |
| 3 | {Average Monthly Income<3000 RMB} | 0.012 | 0.099 | 0.120 | 3.534 | 20 | |
| 4 | {Average Monthly Income<3000 RMB, Administrative Position = No} | 0.011 | 0.098 | 0.114 | 3.531 | 19 | |
| 5 | {Psychological Distress Level (K10 Group) = Severe Distress} | 0.012 | 0.080 | 0.157 | 2.853 | 21 |
Fig. 2.
Association rule analysis of sleep status in clinical nurses
Discussion
This study identified three distinct sleep profiles among 1,686 clinical nurses using latent profile analysis (LPA): high sleep status (63.11%), moderate sleep status (34.10%), and low sleep status (2.79%). This classification challenges the traditional homogeneity assumption of sleep assessment based solely on total scale scores, highlighting multidimensional differences in sleep quality.
The proportion of nurses in our study identified as having poor sleep quality via the traditional PSQI cutoff of >7 was 27.63%. This prevalence is consistent with the high burden of sleep problems documented among nurses globally, particularly in high-stress environments like China’s healthcare system [18, 19], and underscores the persistent and critical nature of addressing sleep health in this population. However, our LPA results refine this understanding. The small but severely affected ‘Low Sleep Status’ group (2.79%) captured the most vulnerable individuals, whom the PSQI cutoff would have included within the broader 27.63%. More importantly, LPA succeeded in segmenting the remaining population into meaningful subgroups, revealing that a substantial proportion (34.10%) experience moderate but distinct sleep challenges that warrant targeted attention. This demonstrates that LPA moves beyond a simple ‘good-poor’ dichotomy to uncover the underlying heterogeneity and complex interplay of factors affecting nurses’ sleep.
Rather than seeking a universal “natural law” of sleep determinants, our findings provide a nuanced, data-driven case study that captures the complex interplay of factors within a specific socio-cultural and occupational context. In line with a “coherence” perspective on scientific validity, sleep quality is best understood not as a decontextualized biological absolute, but as a construct shaped by an individual’s specific circumstances, including their work environment, socioeconomic resources, and psychological state. The latent profile analysis employed here is particularly suited to this view, as it does not presume population homogeneity but actively identifies meaningful subgroups that may be obscured by aggregate-level analysis. The three profiles we identified—each with its own configuration of risk and protective factors—serve as a testament to this underlying heterogeneity. Thus, the value of this study lies in its contribution to a growing body of literature; its patterns and associations await confirmation and comparison through future replications in diverse settings to build a more coherent, generalizable understanding of nurses’ sleep health.
The low sleep status group (n = 47) exhibited marked vulnerability: 42.55% were unmarried/divorced/widowed, 42.55% had a monthly income ≤ 3,000 CNY, 76.60% held non-permanent employment status, and 44.68% reported severe psychological distress—all significantly higher than other groups (p < 0.001). Notably, this group showed greater variability in sleep efficiency (mean = 0.53, SD = 0.86) and daytime dysfunction (mean = 1.20, SD = 0.91), indicating pronounced internal heterogeneity. These findings suggest that the convergence of occupational instability (non-permanent status), financial strain (low income), and inadequate social support (unmarried status) can converge to lowers stress tolerance thresholds, potentially explaining why minor stressors may trigger systemic dysregulation [20]. McEwen and Karatsoreos’ allostatic load theory provides a plausible biological framework for these patterns: chronic stress disrupts cortisol rhythms via HPA axis hyperactivity, suppresses melatonin secretion, and induces neuroinflammation, ultimately destabilizing sleep homeostasis [21]. In contrast, the high sleep status group demonstrated protective characteristics: 85.62% were married, 43.14% held permanent employment, and most had moderate incomes. This group exhibited significantly lower psychological distress, which aligns with the social support buffering model—whereby stable marital relationships may reduce sympathetic nervous system activity through emotional and economic support [22], and permanent employment could help to mitigates shiftwork-induced circadian disruptions [23].
Univariate and multivariate logistic regression analyses identified core differentiating variables across sleep subgroups: marital status, economic income, occupational stability, night shift frequency (p < 0.001), and psychological state (p < 0.001) collectively formed a multidimensional predictive network for sleep quality. Specifically, unmarried/divorced/widowed nurses had 2.153-fold (95% CI: 1.472–3.148) and 2.252-fold (95% CI: 1.015–4.996) higher risks of belonging to the moderate and low sleep groups, respectively, compared to married individuals. This aligns with prior research indicating that the absence of spousal emotional support and economic collaboration depletes stress-coping resources, exacerbating sympathetic nervous system activity and impairing sleep maintenance mechanisms [22].The impact of financial strain on sleep exhibited a dose-response relationship: nurses with monthly incomes ≤ 3,000 CNY faced a 9.098-fold (95% CI: 1.034–80.036) increased risk of entering the low sleep group compared to high-income counterparts (≥ 9,000 CNY), corroborating findings by Goh et al. that economic hardship amplifies anxiety and reduces sleep efficiency via hyperactivity in prefrontal-limbic neural circuits [24]. Occupational stressors demonstrated significant interactive effects: non-permanent employment status combined with frequent night shifts (>7 days/month) elevated sleep disorder risk by 2.7-fold. Notably, psychosocial factors revealed clinically meaningful gradients: nurses with low psychological distress had 90.1% (OR = 0.099, 95% CI: 0.067–0.145) and 94.0% (OR = 0.060, 95% CI: 0.022–0.165) lower risks of moderate and low sleep group membership, respectively, compared to those with severe distress. Additionally, the moderating role of work-family conflict in the moderate sleep group highlighted a unique dilemma among married nurses: while partial balancing of family responsibilities and occupational demands may transiently alleviate stress, it fails to fully counteract sleep quality deterioration. Sustained depletion of psychological resources under dual work-family role conflicts weakens stress-buffering capacity, ultimately leading to sleep fragmentation [25]. These findings underscore the need for interventions targeting married nurses to prioritize flexible scheduling and childcare support, fostering dynamic equilibrium of stressors.
The application of the Apriori algorithm (minimum support = 0.01, confidence = 0.90) uncovered distinct characteristic combinations and interaction mechanisms across nurse sleep subgroups. For nurses in the low sleep status group, the core association rule {monthly income ≤ 3,000 CNY + non-permanent employment status} (support = 1.1%, lift = 3.89) highlighted a synergistic mechanism wherein financial precarity and occupational instability jointly accounted for 76.60% of sleep disorder risk. This underscores the compounding effects of socioeconomic vulnerability and systemic employment inequities. In the moderate sleep status group, the rule {married + monthly income 3,001–5,000 CNY + non-permanent employment + experienced verbal/physical abuse + high psychological distress} (support = 1.1%, lift = 2.78) revealed a multidimensional stress burden among married nurses. Despite potential emotional support from marital relationships, the interplay of employment insecurity (non-permanent status) and workplace violence effectively neutralized the protective effects of familial bonds, suggesting that structural stressors dominate over individual resilience factors in this subgroup. Conversely, the high sleep status group exhibited adaptive resilience through rules such as {monthly income ≥ 7,001 CNY + night shifts > 7 days/month + low psychological distress} (support = 1.2%, lift = 2.79). Here, financial security mitigated anxiety-related neurophysiological dysregulation, while low psychological distress buffered the physiological impacts of frequent night shifts, such as melatonin suppression, demonstrating how resource abundance and psychological stability interact to sustain sleep health.
Building on these findings, which unveil distinct risk configurations through both LPA and association rule mining, a stratified intervention strategy is imperative. Moving beyond one-size-fits-all approaches, our data advocate for precision interventions that target the specific vulnerabilities of each subgroup, with an emphasis on pragmatism within the current healthcare system. For the vulnerable Low-Sleep Subgroup, interventions must address the synergistic threat of socioeconomic precarity. System-level advocacy to improve income stability and contract conditions for non-permanent nurses is fundamental, as financial strain is a paramount risk factor [26, 27]. At the organizational level, guaranteed access to low-threshold psychological first aid [28, 29] and stringent enforcement of rest periods after night shifts [30] are critical, feasible steps to prevent allostatic overload. For the Moderate-Sleep Subgroup, characterized by the interplay of work-family strain and occupational stressors, targeted support is key. Hospital administrators should prioritize the implementation of flexible scheduling where feasible [31], robust violence de-escalation training [32], and the provision of affordable childcare support. These measures directly address the multidimensional stress burden that erodes sleep quality in this large segment of the nursing workforce. For the resilient High-Sleep Subgroup, the focus should shift to preservation. Efforts here can center on reinforcing existing protective factors through continued professional respect, access to mindfulness resources to maintain low psychological distress, and upholding fair shift schedules to sustain their adaptive resilience [33].
Despite revealing heterogeneous sleep characteristics and multidimensional mechanisms among nurses through latent profile analysis and association rule mining, this study has several limitations. This study has several limitations, which should be taken into account when interpreting the study findings. First, regarding the study design, the cross-sectional nature of our data precludes the establishment of a causal relationship between the identified factors and sleep profiles. Second, in terms of measurement, our reliance on self-reported questionnaires may introduce recall bias and social desirability bias. The lack of objective sleep measurements and data on key potential confounders—such as specific clinical conditions and napping behavior—limits the comprehensiveness and clinical specificity of our assessments. Third, regarding sample representativeness, our data were derived from one province in China, which may affect the generalizability of the study findings to nurses in other healthcare systems and cultural contexts. Furthermore, the small size of the subgroup with poor sleep quality (n = 47) may compromise the statistical stability of its associated characteristics. Additionally, due to the exclusion of nurses who had left the profession due to severe sleep issues, this study may be subject to survivorship bias. Finally, from a methodological and epistemological perspective, although our analytical approach is robust in identifying data-driven subgroups, it operates within a framework that seeks a coherent internal structure. This inherently embeds assumptions about the divisibility of the population and fails to incorporate organizational-level influences, potentially overlooking broader contextual effects. Future studies should directly address these limitations. Longitudinal cohorts are essential for validating sleep profiles and establishing causal relationships. Integrating multimodal assessments—including objective sleep measurements and clinical evaluations—will provide a more robust understanding of sleep architecture and underlying mechanisms. Crucially, incorporating qualitative methods can help verify whether the identified statistical profiles resonate with nurses’ subjective lived experiences. Finally, studies need to be conducted to test the feasibility and effectiveness of the proposed targeted interventions in diverse healthcare settings.
Future research should address these gaps by expanding to multiregional, large-scale cohorts to enhance external validity and subtype identification accuracy. Longitudinal designs are needed to track dynamic interactions between sleep quality trajectories and occupational health outcomes, particularly in high-stress healthcare environments. Integrating multimodal assessments—including actigraphy, PSG, and neuroendocrine biomarkers (e.g., cortisol, melatonin)—would strengthen the mechanistic understanding of sleep disturbances and validate intervention efficacy. Implementation studies testing the proposed “institutional safeguards–precision interventions” framework across diverse clinical settings are critical to refining strategies for real-world applicability. Collectively, these advancements will provide robust evidence to establish a comprehensive “prevention-intervention-follow-up” continuum in nurse sleep health management.
Acknowledgements
This study was not specifically funded by any funding organisation.
We sincerely thank all the clinical nurses who participated in this study for their valuable time and data. We thank the nursing administration of Qilu Hospital of Shandong University for facilitating the recruitment of participants and access to data. We would also like to thank the experts at Shandong University for their guidance in applying latent profiling and association rule mining. We obtained The Ethics Committee of Shandong University School of Public Health reviewed and approved the study (approval number: 20181219).
Author contributions
Ning Wei: Conducted all statistical analyses, including latent class analysis, association rule mining (Apriori algorithm), and regression modeling. Drafted the full manuscript, integrated results with theoretical frameworks, and finalized tables/figures. Lulu Hu：Designed and administered the survey instruments, coordinated participant recruitment, and ensured data integrity through dual-entry verification and logic checks. Managed ethical compliance and dataset standardization. Jian Li：Performed interdisciplinary interpretation of findings, linking empirical results to occupational stress theories and neuroendocrine pathways. Developed the “Institutional Safeguards–Precision Interventions” framework.Jianying Chu：Conceptualized the study objectives, formulated hypotheses, and designed the analytical workflow. Integrated literature on sleep disorders, socioeconomic determinants, and psychophysiological interactions into the theoretical model. All authors critically reviewed the manuscript, approved the final version, and ensured adherence to ethical guidelines.
Funding
There was no research funding for this study.
Data availability
Due to the regulatory requirements of the Ethics Committee, the supporting data of this study are not available to the public, but can be obtained upon reasonable request from the corresponding author with the consent of the Ethics Committee of the School of Public Health, Shandong University.
Declarations
Ethical approval
In accordance with the ethical principles of the Declaration of Helsinki, confidentiality and anonymity were guaranteed in this study. All participants signed an informed consent form before starting the questionnaire. The Ethics Committee of Shandong University School of Public Health reviewed and approved the study (approval number: 20181219).
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.
References
- 1.Liu Q, Jiang L, Ye J, et al. Association between shift work and sleep disorders among healthcare workers in tertiary hospitals. J Environ Occup Med. 2021;38(1):76–82. 10.13213/j.cnki.jeom.2021.20313. [Google Scholar]
- 2.Caruso CC. Negative impacts of shiftwork and long work hours. Rehabil Nurs. 2014;39(1):16–25. 10.1002/rnj.107. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Yin L, Duan S, Peng X. Effects of sleep deprivation on sleep quality, emotional state, and nursing errors among night-shift nurses. World J Sleep Med. 2023;10(11):2740–2. 10.3969/j.issn.2095-7130.2023.11.073. [Google Scholar]
- 4.Liu D, Wu Y, Jiang F, Liu Y, Tang YL. The prevalence and associated factors of short sleep duration among nurses in tertiary public hospitals in china: findings from a National survey. Risk Manag Healthc Policy. 2021;14:2547–52. 10.2147/RMHP.S315262. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Gu F, Han J, Laden F, et al. Total and cause-specific mortality of U.S. Nurses working rotating night shifts. Am J Prev Med. 2015;48(3):241–52. 10.1016/j.amepre.2014.10.018. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Fadeyi BA, Ayoka AO, Fawale MB, et al. Prevalence, predictors and effects of shift work sleep disorder among nurses in a Nigerian teaching hospital. Sleep Sci Pract. 2018;2:6. 10.1186/s41606-018-0027-x. [Google Scholar]
- 7.Kecklund G, Axelsson J. Health consequences of shift work and insufficient sleep. BMJ. 2016;355:i5210. 10.1136/bmj.i5210. [DOI] [PubMed] [Google Scholar]
- 8.James L, James SM, Wilson M, et al. Sleep health and predicted cognitive effectiveness of nurses working 12-hour shifts: an observational study. Int J Nurs Stud. 2020;112:103667. 10.1016/j.ijnurstu.2020.103667. [DOI] [PubMed] [Google Scholar]
- 9.Feng H, Qi X, Xia C, et al. Association between night shift and sleep quality and health among Chinese nurses: A cross-sectional study. J Nurs Manag. 2021;29(7):2123–31. [DOI] [PubMed] [Google Scholar]
- 10.Zeng X, Xiao L, Zhang Y. Principles and case studies of latent class analysis. Chin J Health Stat. 2013;30(6):815–7. [Google Scholar]
- 11.Zhao M, Su X, Li S, et al. Analysis of fall-related injury characteristics in elderly inpatients based on association rules. J Nurs Sci. 2020;35(18):61–4. [Google Scholar]
- 12.Liu XK, Huang DL, Meng LP, et al. Prevalence and risk factors underlying occupational stress and depression among clinical nurses in secondary and tertiary hospitals of China during COVID-19 setting. Risk Manag Healthc Policy. 2023;16:1377–89. 10.2147/RMHP.S415349. Published 2023 Aug 1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Li H, Dong S, He L, et al. Nurses’ core emergency competencies for COVID-19 in china: A cross-sectional study. Int Nurs Rev. 2021;68(4):524–32. 10.1111/inr.12692. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Liu X, Tang M, Hu L, et al. Reliability and validity of the Pittsburgh sleep quality index. Chin J Psychiatry. 1996;29(2):103–7. [Google Scholar]
- 15.Zhou C, Chu J, Wang T, et al. Reliability and validity of the Chinese version of Kessler 10 psychological distress scale. Chin J Clin Psychol. 2008;16(6):627–9. [Google Scholar]
- 16.Grzywacz JG, Marks NF. Reconceptualizing the work-family interface: an ecological perspective. J Occup Health Psychol. 2000;5(1):111–26. 10.1037/1076-8998.5.1.111. [DOI] [PubMed] [Google Scholar]
- 17.Jiang Q. Perceived social support scale. Chin J Behav Med Sci. 2001;10(10):41–3. [Google Scholar]
- 18.Zare Khormizi H, Salehinejad MA, Nitsche MA, Nejati V. Sleep-deprivation and autobiographical memory: evidence from sleep-deprived nurses. J Sleep Res. 2019;28(1):e12683. 10.1111/jsr.12683. [DOI] [PubMed] [Google Scholar]
- 19.Lee LJ, Wehrlen L, Ding Y, Ross A. Professional quality of life, sleep disturbance and health among nurses: A mediation analysis. Nurs Open. 2022;9(6):2771–80. 10.1002/nop2.978. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Cheval B, Cullati S, Pihl-Thingvad J, et al. Impact of CAre-related regret upon sleep (ICARUS) cohort study. BMJ Open. 2018;8(3):e022172. 10.1136/bmjopen-2018-022172. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.McEwen BS, Karatsoreos IN. Sleep deprivation and circadian disruption. Sleep Med Clin. 2022;17(2):253–62. 10.1016/j.jsmc.2022.03.005. [DOI] [PubMed] [Google Scholar]
- 22.Kent de Grey RG, Uchino BN, Trettevik R, et al. Social support and sleep: a meta-analysis. Health Psychol. 2018;37(8):787–98. 10.1037/hea0000628. [DOI] [PubMed] [Google Scholar]
- 23.Irwin MR, Olmstead R, Carroll JE. Sleep disturbance and inflammation. Biol Psychiatry. 2016;80(1):40–52. 10.1016/j.biopsych.2015.05.014. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Hamilton OS, Steptoe A. Financial stress and sleep duration. Brain Behav Immun. 2025. 10.1016/j.bbi.2025.03.006. [DOI] [PubMed] [Google Scholar]
- 25.Cheng S, Lin P, Chang Y, et al. Sleep quality mediates work-family conflicts and health status in nurses. J Nurs Manag. 2019;27(2):381–7. 10.1111/jonm.123694. [DOI] [PubMed] [Google Scholar]
- 26.Rote S, Cantu P, Angel J, Moon H, Antequera F, Markides K. Financial Strain, mental Health, and sleep disruption among Racially and ethnically diverse family caregivers during the COVID-19 pandemic. J Cross Cult Gerontol. 2025;40(3):289–305. 10.1007/s10823-025-09543-y. [DOI] [PubMed] [Google Scholar]
- 27.Chai L, Lu Z. The association between financial strain and mental health: the mediating and moderating roles of sleep problems in the UK household longitudinal study (UKHLS). J Affect Disord. 2025;377:245–53. 10.1016/j.jad.2025.02.060. [DOI] [PubMed] [Google Scholar]
- 28.Kameno Y, Hanada A, Asai D, et al. Individual psychotherapy using psychological first aid for frontline nurses at high risk of psychological distress during the COVID-19 pandemic. Psychiatry Clin Neurosci. 2021;75(1):25–7. 10.1111/pcn.13170. [DOI] [PubMed] [Google Scholar]
- 29.Caillet A, Conejero I, Allaouchiche B. Job strain and psychological impact of COVID-19 in ICU caregivers during pandemic period. Anaesth Crit Care Pain Med. 2021;40(2):100850. 10.1016/j.accpm.2021.100850. [DOI] [PubMed] [Google Scholar]
- 30.Li Y, Sun J, Sun D, et al. Nurses’ perceptions of night shifts: A qualitative study. Int Emerg Nurs. 2022;64:101202. 10.1016/j.ienj.2022.101202. [DOI] [PubMed] [Google Scholar]
- 31.Zhang Y, Murphy J, Lammers-van der Holst HM, Barger LK, Lai YJ, Duffy JF. Interventions to improve the sleep of nurses: A systematic review. Res Nurs Health. 2023;46(5):462–84. 10.1002/nur.22337. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Matricciani L, Clarke J, Wiley S, et al. Sleep of nurses: A comprehensive scoping review. J Adv Nurs. 2025;81(5):2333–44. 10.1111/jan.16603. [DOI] [PubMed] [Google Scholar]
- 33.Stutting HL. The relationship between rest breaks and professional burnout among nurses. Crit Care Nurse. 2023;43(6):48–56. 10.4037/ccn2023177. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
Due to the regulatory requirements of the Ethics Committee, the supporting data of this study are not available to the public, but can be obtained upon reasonable request from the corresponding author with the consent of the Ethics Committee of the School of Public Health, Shandong University.