Associations between shift work, sleep quality and menstrual disorders among nurses: a cross-sectional analysis study.

This is a cross-sectional study of female nursing staff in Anhui Province, China. Data were obtained from the baseline survey of the Nurses’ Environmental & Occupational exposure and health Nexus cohort (NEON), conducted between July 2022 and December 2023. A cluster sampling method was used to recruit participants. In the first stage, six hospitals of Secondary Grade A or above were selected from different geographic regions of Anhui Province. In the second stage, all female nurses who met the inclusion criteria were invited to participate. A self-administered questionnaire was utilized to collect data on demographic characteristics, lifestyle behaviors, history of shift work, sleep quality, and menstruation. The study adhered strictly to informed consent protocols. Ethical approval for the study was granted by the Ethics Committee of Anhui Medical University (Approval No. 83220409). The inclusion criteria for study participants were as follows: (1) female nurses aged ≥ 18 years; (2) nurses with a professional nursing certificate; (3) informed consent for voluntary participation in the survey. Exclusion criteria included: (1) pregnancy at the time of the survey; (2) menopause during the survey period; (3) use of contraceptives, intrauterine device, or other sex hormones; (4) a diagnosed history of menstrual dysfunction (e.g., polycystic ovary syndrome, endometriosis, premature ovarian failure, ovarian dysfunction, or thyroid disease). A total of 5,433 questionnaires were distributed to eligible nurses across the six hospitals, and 4,520 valid questionnaires were returned, yielding a response rate of 83.0%. Among these, 976 participants were excluded based on the exclusion criteria: one participant was pregnant during the survey period; 326 nurses had menopause; 368 nurses were using contraceptives, intrauterine devices or sex hormone medications; 265 nurses reported a diagnosis of menstrual dysfunction; and 16 nurses had missing data on sleep quality. Consequently, a total of 3,544 participants were included in the final analysis. The flowchart for the inclusion of study participants is shown in Fig.  1 . Fig. 1 Flow chart for inclusion and exclusion of study participants Flow chart for inclusion and exclusion of study participants Information on nurses’ shift schedules and work histories was investigated using a structured questionnaire. In our study, rotating shift work was defined as any work schedule other than a regular day shift (7/8 a.m. to 5/6 p.m.), including alternate day-noon, day-evening, or day-night shift, based on prior definitions [ 3 ]. Participants were asked “Have you had any rotating shift work experience since you started working?” to assess their rotating shift work history. The history of rotation direction was assessed by inquiring about the temporal direction of participants’ shift schedules, with the following options: (1) forward rotation (clockwise: day→evening→night); (2) backward rotation (counterclockwise: day→night→evening); and (3) irregular (variable shifts). Night shifts were defined as work schedules beginning in the afternoon and lasting until 8:00 the following day, encompassing both evening shift (16:00–24:00) and night shift (24:00–8:00) [ 1 ]. To assess night shift work history, participants were asked, “Have you ever had any experience with night shift work since you started working?” This question captured fixed evening or night shift, rotating night shift, and other related night work arrangements. In reviewing participants’ past work schedules, we observed that some had experience with rotating night shift, whereas others had worked permanent fixed night work. Therefore, both types were combined to evaluate the overall history of night work/rotating night shift experience. Data on the number of years of rotating shift work experience and night work/rotating night shift experience were obtained by asking participants to report the specific durations (in years). Based on these reported durations, participants were categorized into the following groups: never, 0–4, 4–8, 8–12, and ≥ 12 years. In addition, the monthly frequency of evening shift (16:00–24:00) and night shifts (24:00–08:00) was defined as the average number of such shifts performed per month over the past year, categorized into never, ≤ 5 and > 5 times. Weekly working hours over the past year were calculated by multiplying the average daily working hours (hours/day) by the number of working days per week (days/week). Working intensity was assessed via self-report (relatively easy, tolerable, or exhausting). Due to the small sample size in the “relatively easy” group ( n  = 38), it was merged with the “tolerable” category to serve as a consolidated reference group for logistic regression. Nurses’ sleep quality was evaluated using the Pittsburgh Sleep Quality Index (PSQI), which assesses sleep over the past month across seven dimensions: subjective sleep quality, sleep latency, sleep duration, habitual sleep efficiency, sleep disturbances, use of sleep medications, and daytime dysfunction. For each component, the score ranges from 0 (good sleep/no problem) to 3 (poor sleep/severe problem), which are conventionally aggregated to yield a total PSQI score ranging from 0 to 21, with higher scores indicating poorer sleep quality. A score > 5 was used as the threshold to classify poor sleep quality [ 30 ]. Cronbach’s alpha coefficient of the scale in this study was 0.73, indicating acceptable internal consistency. Chronotype was evaluated using a single-item self-report question based on the item 19 of the Morningness-Eveningness Questionnaire (MEQ) [ 31 ]: “People can be expressed as ‘morning type’ or ‘evening type’. Which type do you think you belong to? (Note: ‘morning type’ tends to go to bed early and wake up early, and feels relatively alert in the morning; ‘evening type’ tends to go to bed late and wake up late, and feels relatively alert in the evening).” Five response options were provided: (1) definitely evening type; (2) evening type more than morning type; (3) morning type more than evening type; (4) definitely morning type; (5) neither. For analysis, responses (1) and (2) were combined into Evening type, (3) and (4) into Morning type, and response (5) was classified as Intermediate type. This single-item chronotype measure correlates well with the overall MEQ score and has been used in previous epidemiological studies [ 32 ]. Considering the epidemiological design of this study, we adopted a pragmatic and operational definition of menstrual disorders by focusing on three commonly used and clinically relevant indicators: irregular menstrual cycles, dysmenorrhea, and abnormal menstrual bleeding (hypomenorrhea and menorrhagia). This approach enhances feasibility in large-scale population-based research while maintaining consistency with established frameworks and prior epidemiological studies. Menstrual cycle length was defined as the interval from the first day of menstruation to the first day of the following period [ 33 ]. Participants were asked, “In the past year, did you have a regular menstrual cycle under normal circumstances, with a regular cycle defined as of 21–35 days and a menstrual period lasting 2–8 days?” Specific information on menstrual period length and cycle duration was collected. Referring to the guidelines for the treatment of menstrual disorders, an irregular menstrual cycle was defined as an interval of 35 days between two consecutive menstrual periods, or a variation in menstrual cycle length exceeding 7 days [ 34 ]. Dysmenorrhea severity was evaluated using the Wong-Baker Faces Pain Scale–Revised (FPS-R), adapted for Chinese populations [ 35 ], Participants rated their pain on a scale of 6 pain levels (0, 2, 4, 6, 8 and 10) from 0 (no pain) to 10 (worst pain), with six facial expressions serving as visual anchors. Dysmenorrhea was defined as a pain score of ≥ 2. Menstrual blood loss was classified using a self-reported, questionnaire-based approach according to the criteria adapted from a standard obstetrics and gynecology textbook (7th edition) [ 36 ]. This study used an operational classification based on sanitary product use patterns. The categories were: (1) hypomenorrhea (perceived as low volume, e.g., few changes of sanitary napkins per day, with little blood on the sanitary napkins); (2) normal menstrual flow (perceived as moderate volume, e.g., 4–5 changes during the day, with about half of the sanitary napkin covered in blood); (3) menorrhagia (perceived as heavy volume, e.g., frequent changes, using more than 15–20 sanitary napkins per menstrual cycle). This operational approach focuses on observable usage behaviors to reflect perceived blood loss in an epidemiological context [ 37 ]. Menstrual disorder was defined as the presence of at least one of the following four self-reported menstrual outcomes: irregular menstrual cycles, dysmenorrhea, hypomenorrhea, or menorrhagia. Based on prior literature, the following covariates were adjusted for in the analysis: age ( 35 years), BMI (kg/m²), personal annual income ( 200,000 yuan ), marital status (unmarried, married, other), and education level (college degree or below, undergraduate, master’s degree or above). Active smoking was defined as a lifetime consumption of ≥ 100 cigarettes (approximately five packs), while passive smoking was defined as an average daily exposure to second-hand smoke of ≥ 15 min. Given that only six participants in this study were active smokers, active and passive smoking were combined into a single binary variable—tobacco exposure (yes/no) to ensure stable estimation. Drinking history (yes/no) was defined as consuming alcohol at least once per week. Physical activity (yes/no), and bedtime video screen time ( 60 min/day) were also recorded. Bedtime video screen time referred to the average duration spent using electronic devices (e.g., cell phones, tablets, computers) before sleep (min/day). Pregnancy history was defined as whether the participant had ever been pregnant (yes/no). Childbirth history was defined as a history of childbirth (i.e., having delivered at least one child; yes/no). PHQ-9 was used to assess depression (0–4 = no, 5–27 = yes), and GAD-7 was used to assess anxiety (0–4 = no, 5–21 = yes). Covariates with statistical significance ( P  < 0.05) in the univariate analyses were included in the final models, adjusted for specific outcomes. Qualitative data were expressed as frequencies and percentages (%), with between-group differences assessed using the chi-square test (χ²). Quantitative data were described as mean ± standard deviation (SD), and between-group comparisons were analyzed using one-way analysis of variance (ANOVA). Binary logistic regression analysis was performed to evaluate the associations of shift work and sleep characteristics with menstrual cycle irregularity and dysmenorrhea. For the three-category menstrual blood volume outcome (normal volume, hypomenorrhea, and menorrhagia), multinomial logistic regression was applied, with the “Normal Volume” group as the reference category. We assessed the quality of the model fit by using Spearman correlation coefficients and variance inflation factors (VIFs) to examine potential multicollinearity effects. Latent class analysis (LCA) was employed to identify distinct sleep quality profiles among the nursing staff, utilizing the seven components of the PSQI as manifest indicators. To analyze the data in this manner, we binarized the scores on each component into “good” (0 or 1) or “poor” (2 or 3) according to literature [ 38 ]. The model fit indicators include the Akaike information criterion (AIC), the Bayesian information criterion (BIC), and the adjusted BIC (aBIC). The optimal number of latent classes was determined using the Lo Mendell-Rubin test (LMR) and parametric bootstrapped likelihood ratio test (BLRT). Entropy > 0.6 indicated that the accuracy of case categorization is more than 80.0% [ 39 ]. The LCA was performed using Mplus 8.0 software. The mediation analysis was executed using the mediation package (v4.5.0) in R. We estimated the Average Causal Mediation Effect (ACME), the Average Direct Effect (ADE), and the Total Effect. Although the underlying algorithm outputs the term “Causal”, these ACME were interpreted strictly as statistical indirect associations rather than true causal relationships, given the cross-sectional nature of our data. To account for the combination of linear and logistic models, the statistical significance of the indirect effect was determined using the non-parametric bootstrapping method with 1,000 iterations. All analysis models were adjusted for potential confounders. Estimates of parameters were expressed as odds ratios (ORs) with 95% confidence intervals (95% CIs). A two-sided P  < 0.05 was considered statistically significant. In addition, given that dysmenorrhea is a common outcome (prevalence 54.6% in our sample), ORs from logistic regression may overestimate the true association compared with prevalence ratios (PRs). Therefore, we conducted a sensitivity analysis using robust Poisson regression to directly estimate PRs and their 95% CIs for the same associations [ 40 ]. Robust standard errors were used to account for potential overdispersion. The results from the robust Poisson regression were compared with those from the main logistic regression analysis to assess the robustness of our findings.

The basic characteristics of the 3,544 included nurses are shown in Table  1 . The mean (± SD) age was 28.87 ± 4.15 years and mean body mass index (BMI) was 21.95 ± 3.28 kg/m 2 . Most of the nurses (70.8%) fell within the normal BMI range (18.5–23.9 kg/m²). A majority held a bachelor’s degree (62.4%) and were married (65.9%). The annual income distribution showed that 49.2% of them earned between 50,000 and 110,000 yuan . Lifestyle factors indicated that 19.8% of the nurses were exposed to tobacco smoke (including active and passive inhalation), 27.0% reported alcohol consumption, and 21.3% performed regular physical exercise. Notably, 67.2% of the nurses spent ≥ 45 min nightly using electronic devices before sleep. Pregnancy history showed that 60.6% of the nurses had been pregnant and 58.8% had given birth. Individually, the prevalence of irregular menstrual cycles, dysmenorrhea, hypomenorrhea, and menorrhagia was 20.3%, 54.6%, 12.6%, and 10.0%, respectively. Overall, the prevalence of general menstrual disorders (defined as having at least one of the four symptoms) was 70.7% ( n  = 2,507), while the remaining 29.3% of the participants ( n  = 1,037) reported none of these conditions and were categorized as completely asymptomatic. Table 1 Demographic characteristics of the study population by menstrual disorder status ( n  = 3544) Basic characteristic N (%) Irregular menstrual cycles Dysmenorrhea Menstrual blood loss Menstrual disorder a Case (%) Non–Case (%) Case (%) Non–Case (%) Hypomenorrhea (%) Normal (%) Menorrhagia (%) Case (%) Non–Case (%) Age (years)  < 25 644 (18.2) 154 (21.3) 490 (17.4) 362 (18.7) 282 (17.5) 63 (14.1) 542 (19.8) 39 (11.1) 462 (18.4) 182 (17.6)  25–35 2110 (59.5) 428 (59.3) 1682 (59.6) 1189 (61.5) 921 (57.3) 279 (62.4) 1630 (59.4) 201 (56.9) 1523 (60.8) 587 (56.6)  > 35 790 (22.3) 140 (19.4) 650 (23.0) 384 (19.8) 406 (25.2) 105 (23.5) 572 (20.8) 113 (32.0) 522 (20.8) 268 (25.8) BMI (kg/m 2 )  < 18.5 383 (10.8) 66 (9.1) 317 (11.2) 227 (11.7) 156 (9.7) 44 (9.8) 317 (11.6) 22 (6.2) 272 (10.8) 111 (10.7)  18.5–23.9 2508 (70.8) 495 (68.6) 2013 (71.4) 1336 (69.0) 1172 (72.8) 322 (72.0) 1941 (70.7) 245 (69.4) 1747 (69.7) 761 (73.4)  ≥ 24 653 (18.4) 161 (22.3) 492 (17.4) 372 (19.3) 281 (17.5) 81 (18.2) 486 (17.7) 86 (24.4) 488 (19.5) 165 (15.9) Marriage  Unmarried 1120 (31.6) 251 (34.7) 869 (30.8) 671 (34.7) 449 (27.9) 138 (30.9) 918 (33.5) 64 (18.1) 837 (33.4) 283 (27.3)  Married 2337 (65.9) 451 (62.5) 1886 (66.8) 1218 (62.9) 1119 (69.6) 286 (64.0) 1768 (64.4) 283 (80.2) 1609 (64.2) 728 (70.2)  Other 87 (2.5) 20 (2.8) 67 (2.4) 46 (2.4) 41 (2.5) 23 (5.1) 58 (2.1) 6 (1.7) 61 (2.4) 26 (2.5) Education  Specialty and below 1307 (36.9) 265 (36.7) 1042 (36.9) 678 (35.0) 629 (39.1) 157 (35.1) 1034 (37.7) 116 (32.9) 893 (35.6) 414 (39.9)  Undergraduate 2212 (62.4) 453 (62.7) 1759 (62.4) 1245 (64.3) 967 (60.1) 287 (64.2) 1692 (61.6) 233 (66.0) 1595 (63.6) 617 (59.5)  Master’s degree and above 25 (0.7) 4 (0.6) 21 (0.7) 12 (0.7) 13 (0.8) 3 (0.7) 18 (0.7) 4 (1.1) 19 (0.8) 6 (0.6) Annual income (ten thousand yuan )  < 5 711 (20.1) 162 (22.4) 549 (19.5) 430 (22.2) 281 (17.5) 96 (21.5) 557 (20.3) 58 (16.4) 524 (20.9) 187 (18.0)  5–11 1743 (49.2) 357 (49.5) 1386 (49.1) 994 (51.4) 749 (46.5) 236 (52.8) 1340 (48.8) 167 (47.3) 1278 (51.0) 465 (44.8)  11–20 997 (28.1) 188 (26.0) 809 (28.6) 472 (24.4) 525 (32.6) 107 (23.9) 775 (28.2) 115 (32.6) 647 (25.8) 350 (33.8)  > 20 93 (2.6) 15 (2.1) 78 (2.8) 39 (2.0) 54 (3.4) 8 (1.8) 72 (2.7) 13 (3.7) 58 (2.3) 35 (3.4) Tobacco exposure  No 2844 (80.2) 553 (76.6) 2291 (81.2) 1498 (77.4) 1346 (83.7) 335 (74.9) 2237 (81.5) 272 (77.1) 1959 (78.1) 885 (85.3)  Yes 700 (19.8) 169 (23.4) 531 (18.8) 437 (22.6) 263 (16.3) 112 (25.1) 507 (18.5) 81 (22.9) 548 (21.9) 152 (14.7) Alcohol consumption  No 2587 (73.0) 491 (68.0) 2096 (74.3) 1410 (72.9) 1177 (73.2) 298 (66.7) 2035 (74.2) 254 (72.0) 1806 (72.0) 781 (75.3)  Yes 957 (27.0) 231 (32.0) 726 (25.7) 525 (27.1) 432 (26.8) 149 (33.3) 709 (25.8) 99 (28.0) 701 (28.0) 256 (24.7) Physical activity  No 2790 (78.7) 589 (81.6) 2201 (78.0) 1565 (80.9) 1225 (76.1) 379 (84.8) 2125 (77.4) 286 (81.0) 2020 (80.6) 770 (74.3)  Yes 754 (21.3) 133 (18.4) 621 (22.0) 370 (19.1) 384 (23.9) 68 (15.2) 619 (22.6) 67 (19.0) 487 (19.4) 267 (25.7) Bedtime screen time (min/d)  < 30 906 (25.6) 164 (22.7) 742 (26.3) 440 (22.7) 466 (29.0) 106 (23.7) 699 (25.5) 101 (28.6) 581 (23.2) 325 (31.3)  31–45 254 (7.2) 44 (6.1) 210 (7.4) 179 (9.3) 75 (4.7) 37 (8.3) 187 (6.8) 30 (8.5) 210 (8.4) 44 (4.2)  46–60 1093 (30.8) 198 (27.4) 895 (31.7) 597 (30.8) 496 (30.8) 128 (28.6) 863 (31.5) 102 (28.9) 768 (30.6) 325 (31.3)  > 60 1291 (36.4) 316 (43.8) 975 (34.6) 719 (37.2) 572 (35.5) 176 (39.4) 995 (36.2) 120 (34.0) 948 (37.8) 343 (33.2) Childbirth history  No 1460 (41.2) 327 (45.3) 1133 (40.1) 858 (44.3) 602 (37.4) 182 (40.7) 1189 (43.3) 89 (25.2) 1075 (42.9) 385 (37.1)  Yes 2084 (58.8) 395 (54.7) 1689 (59.9) 1077 (55.7) 1007 (62.6) 265 (59.3) 1555 (56.7) 264 (74.8) 1432 (57.1) 652 (62.9) Pregnancy history  No 1398 (39.4) 310 (42.9) 1088 (38.6) 829 (42.8) 569 (35.4) 165 (36.9) 1148 (41.8) 85 (24.1) 1037 (41.4) 361 (34.8)  Yes 2146 (60.6) 412 (57.1) 1734 (61.4) 1106 (57.2) 1040 (64.6) 282 (63.1) 1596 (58.2) 268 (75.9) 1470 (58.6) 676 (65.2) Depression symptoms  No 1366 (38.5) 214 (29.6) 1152 (40.8) 633 (32.7) 733 (45.6) 112 (25.1) 1148 (41.8) 106 (30.0) 843 (33.6) 523 (50.4)  Yes 2178 (61.5) 508 (70.4) 1670 (59.2) 1302 (67.3) 876 (54.4) 335 (74.9) 1596 (58.2) 247 (70.0) 1664 (66.4) 514 (49.6) Anxiety symptoms  No 1480 (41.8) 234 (32.4) 1246 (44.2) 680 (35.1) 800 (49.7) 138 (30.9) 1230 (44.8) 112 (31.7) 913 (36.4) 567 (54.7)  Yes 2064 (58.2) 488 (67.6) 1576 (55.8) 1255 (64.9) 809 (50.3) 309 (69.1) 1514 (55.2) 241 (68.3) 1594 (63.6) 470 (45.3) a Menstrual disorder was defined as the presence of at least one of the following four self-reported menstrual outcomes: irregular menstrual cycles, dysmenorrhea, hypomenorrhea, or menorrhagia Demographic characteristics of the study population by menstrual disorder status ( n  = 3544) a Menstrual disorder was defined as the presence of at least one of the following four self-reported menstrual outcomes: irregular menstrual cycles, dysmenorrhea, hypomenorrhea, or menorrhagia After adjusting for potential confounders, a history of rotating shift work was significantly positively correlated with menstrual disorders (Table  2 ). Compared with non-shift workers, individuals with a history of rotating shift work had higher odds of irregular menstrual cycles ( OR  = 1.31, 95% CI : 1.03–1.65) and dysmenorrhea ( OR  = 1.71, 95% CI : 1.42–2.06). Regarding the rotation direction, backward rotation ( OR  = 1.45, 95% CI : 1.08–1.95) and irregular rotation ( OR  = 1.46, 95% CI : 1.12–1.89) were significantly tied to irregular menstrual cycles, while all three rotation patterns substantially elevated the odds of dysmenorrhea, with forward rotation presenting the highest likelihood ( OR  = 1.94, 95% CI : 1.58–2.39). Additionally, a history of night work/rotating night shift was specifically associated with dysmenorrhea ( OR  = 1.40, 95% CI : 1.04–1.87). Table 2 Association between shift work patterns and menstrual disorders ( N  = 3544) Shift work characteristic N Irregular menstrual cycles Dysmenorrhea Hypomenorrhea c Menorrhagia c Model 1 Model 2 Model 1 Model 2 Model 1 Model 2 Model 1 Model 2 Rotating shift work history  No 597 Ref Ref Ref Ref Ref Ref Ref Ref  Yes 2947 1.32 (1.05–1.67)* 1.31 (1.03–1.65)* 1.73 (1.44–2.08)* 1.71 (1.42–2.06)* 1.00 (0.77–1.31) 0.99(0.76–1.30) 1.24 (0.89–1.72) 1.23(0.88–1.70) History of rotation direction  Never 597 Ref Ref Ref Ref Ref Ref Ref Ref  Forward rotation 1279 1.13 (0.87–1.45) 1.12 (0.87–1.46) 1.80 (1.48–2.19)* 1.94 (1.58–2.39)* 1.06 (0.79–1.42) 1.03 (0.77–1.39) 1.37 (0.97–1.95) 1.22 (0.86–1.74)  Backward rotation 566 1.46 (1.10–1.95)* 1.45 (1.08–1.95)* 1.42 (1.12–1.78)* 1.52 (1.19–1.94)* 0.93 (0.65–1.32) 0.91 (0.64–1.30) 1.25 (0.83–1.88) 1.20 (0.79–1.82)  Irregular 1102 1.41 (1.09–1.82)* 1.46 (1.12–1.89)* 1.37 (1.12–1.67)* 1.57 (1.28–1.94)* 0.97 (0.72–1.31) 0.98 (0.72–1.34) 1.40 (0.98–2.00) 1.24 (0.87–1.79) Night work/rotating night shift history  No 217 Ref Ref Ref Ref Ref Ref Ref Ref  Yes 3327 1.36 (0.94–1.98) 1.33 (0.92–1.94) 1.42 (1.06–1.90)* 1.40 (1.04–1.87)* 1.07 (0.69–1.66) 1.03 (0.66–1.60) 1.80 (0.96–3.37) 1.75 (0.94–3.29) Years of rotating shift work experience (years) a  Never 597 Ref Ref Ref Ref Ref Ref Ref Ref  0–4 747 1.36 (1.03–1.78)* 1.17 (0.87–1.57) 1.67 (1.34–2.07)* 1.57 (1.23–1.99)* 0.85 (0.61–1.18) 0.97 (0.68–1.39) 0.73 (0.48–1.12) 0.94 (0.60–1.49)  4–8 720 1.38 (1.04–1.81)* 1.42 (1.06–1.90)* 1.70 (1.37–2.12)* 1.81 (1.43–2.29)* 1.00 (0.72–1.38) 0.95 (0.68–1.34) 1.08 (0.73–1.61) 1.07 (0.71–1.62)  8–12 715 1.24 (0.93–1.63) 1.33 (0.99–1.80) 1.78 (1.43–2.22)* 2.11 (1.66–2.68)* 0.90 (0.64–1.26) 0.82 (0.58–1.17) 1.66 (1.14–2.42)* 1.36 (0.92–2.02)  ≥ 12 714 1.19 (0.90–1.58) 1.37 (1.00–1.88)* 1.10 (0.88–1.37) 1.47 (1.14–1.89)* 1.35 (0.98–1.85) 1.32 (0.93–1.88) 2.19 (1.52–3.15)* 1.61 (1.08–2.38)* Years of night work/ rotating night shift experience (years) b  Never 217 Ref Ref Ref Ref Ref Ref Ref Ref  0–4 830 1.32 (0.89–1.95) 1.20 (0.80–1.79) 1.30 (0.96–1.75) 1.27 (0.92–1.74) 1.01 (0.63–1.61) 1.00 (0.62–1.62) 1.28 (0.66–2.50) 1.48 (0.75–2.92)  4–8 804 1.32 (0.90–1.96) 1.49 (0.98–2.25) 1.32 (0.97–1.78) 1.46 (1.05–2.03)* 1.19 (0.75–1.90) 1.01 (0.62–1.65) 1.80 (0.93–3.47) 1.64 (0.84–3.23)  8–12 817 1.25 (0.84–1.85) 1.47 (0.96–2.25) 1.43 (1.06–1.93)* 1.79 (1.28–2.51)* 1.07 (0.67–1.72) 0.85 (0.52–1.41) 2.55 (1.34–4.85)* 1.90 (0.97–3.72)  ≥ 12 807 1.13 (0.76–1.68) 1.41 (0.91–2.19) 0.93 (0.69–1.25) 1.31 (0.92–1.85) 1.54 (0.97–2.44) 1.33 (0.80–2.21) 3.55 (1.88–6.73)* 2.40 (1.22–4.71)* Monthly frequency of evening shift over the past year (times)  Never 290 Ref Ref Ref Ref Ref Ref Ref Ref  ≤ 5 2498 1.30 (0.93–1.80) 1.28 (0.92–1.78) 1.32 (1.03–1.71)* 1.31 (1.01–1.70)* 1.21 (0.80–1.82) 1.18 (0.78–1.78) 1.29 (0.81–2.08) 1.28 (0.80–2.05)  > 5 756 1.58 (1.11–2.26)* 1.55 (1.08–2.22)* 1.45 (1.09–1.93)* 1.43 (1.07–1.90)* 1.51 (0.97–2.35) 1.44 (0.92–2.25) 1.55 (0.93–2.58) 1.52 (0.91–2.54) Monthly frequency of night shift over the past year (times)  Never 411 Ref Ref Ref Ref Ref Ref Ref Ref  ≤ 5 2589 1.19 (0.90–1.57) 1.16 (0.88–1.53) 1.37 (1.10–1.71)* 1.35 (1.08–1.68)* 1.04 (0.75–1.45) 1.00 (0.72–1.40) 1.40 (0.94–2.08) 1.36 (0.91–2.02)  > 5 544 1.55 (1.12–2.15)* 1.50 (1.08–2.07)* 1.59 (1.21–2.08)* 1.54 (1.17–2.02)* 1.32 (0.90–1.96) 1.24 (0.84–1.84) 1.61(1.01–2.56)* 1.55 (0.97–2.48) Weekly working hours over the past year (hours)  ≤ 40 2723 Ref Ref Ref Ref Ref Ref Ref Ref  > 40 821 1.25 (1.03–1.51)* 1.18 (0.98–1.43) 1.20 (1.02–1.45)* 1.14 (0.96–1.34) 0.94 (0.74–1.19) 0.89 (0.70–1.14) 1.12 (0.86–1.45) 1.07 (0.82–1.39) Perceived work intensity over the past year  Relatively easy/Tolerable 2005 Ref Ref Ref Ref Ref Ref Ref Ref  Exhausted 1539 1.23 (1.04–1.46)* 1.10 (0.92–1.31) 1.47 (1.28–1.68)* 1.29 (1.12–1.49)* 1.27 (1.04–1.56)* 1.10 (0.89–1.35) 1.65 (1.32–2.07)* 1.49 (1.18–1.88)* Note: Model 1 was adjusted for outcome–specific covariates. For irregular menstrual cycles, the model included age, BMI, smoking, alcohol consumption, physical activity, bedtime screen time, pregnancy history, and childbirth history; for dysmenorrhea, age, BMI, marital status, education, annual income, smoking, physical activity, bedtime screen time, pregnancy history, and childbirth history were included; for both hypomenorrhea and menorrhagia, the model included age, BMI, marital status, smoking, alcohol consumption, physical activity, pregnancy history, and childbirth history Model 2 was further adjusted for anxiety and depression symptoms based on Model 1 * P  < 0.05, indicating a statistically significant difference a Fifty-one individuals were excluded from the analysis due to missing data on years of rotating shift work experience, resulting in a final analytical sample of 3,493 b Sixty-nine individuals were excluded from the analysis due to missing data on years of night shift work experience, resulting in a final analytical sample of 3,475 c Multinomial logistic regression analysis for the three-category menstrual volume outcome ( N  = 3,544), comprising hypomenorrhea ( n  = 447) and menorrhagia ( n  = 353), with “Normal Volume” ( n  = 2,744) serving as the reference category Association between shift work patterns and menstrual disorders ( N  = 3544) Note: Model 1 was adjusted for outcome–specific covariates. For irregular menstrual cycles, the model included age, BMI, smoking, alcohol consumption, physical activity, bedtime screen time, pregnancy history, and childbirth history; for dysmenorrhea, age, BMI, marital status, education, annual income, smoking, physical activity, bedtime screen time, pregnancy history, and childbirth history were included; for both hypomenorrhea and menorrhagia, the model included age, BMI, marital status, smoking, alcohol consumption, physical activity, pregnancy history, and childbirth history Model 2 was further adjusted for anxiety and depression symptoms based on Model 1 * P  < 0.05, indicating a statistically significant difference a Fifty-one individuals were excluded from the analysis due to missing data on years of rotating shift work experience, resulting in a final analytical sample of 3,493 b Sixty-nine individuals were excluded from the analysis due to missing data on years of night shift work experience, resulting in a final analytical sample of 3,475 c Multinomial logistic regression analysis for the three-category menstrual volume outcome ( N  = 3,544), comprising hypomenorrhea ( n  = 447) and menorrhagia ( n  = 353), with “Normal Volume” ( n  = 2,744) serving as the reference category Using individuals without rotating shift work experience as the reference group, nurses with 4–8 years ( OR  = 1.81, 95% CI : 1.43–2.29) and 8–12 years ( OR  = 2.11, 95% CI : 1.66–2.68) of rotating shift work experience exhibited significantly higher odds of dysmenorrhea. Likewise, relative to nurses without a history of night shift work, nurses with 4–8 years ( OR  = 1.46, 95% CI : 1.05–2.03) and 8–12 years ( OR  = 1.79, 95% CI : 1.28–2.51) was also significantly linked to increased likelihood of dysmenorrhea. Notably, individuals with ≥ 12 years of rotating shift work ( OR  = 1.61, 95% CI : 1.08–2.38) and night shift work ( OR  = 2.40, 95% CI : 1.22–4.71) exhibited a starkly elevated likelihood of menorrhagia. An increasing trend was observed in the odds of dysmenorrhea with higher frequency of evening and night shifts over the past year. High-frequency evening shifts (> 5 times/month) significantly increased the likelihood of irregular menstrual cycles ( OR  = 1.55, 95% CI : 1.08–2.22) and dysmenorrhea (OR  = 1.43, 95% CI : 1.07–1.90). Similarly, high-frequency night shifts (> 5 times/month) were linked to elevated odds of irregular menstrual cycles ( OR  = 1.50, 95% CI : 1.08–2.07) and dysmenorrhea ( OR  = 1.54, 95% CI : 1.17–2.02). In addition, no statistically significant association was found between perceived work intensity and irregular menstrual cycles; however, perceived work intensity was associated with both dysmenorrhea and menorrhagia. Compared with the relatively easy/tolerable group, the exhausted group showed increased odds of dysmenorrhea ( OR  = 1.29, 95% CI : 1.12–1.49) and menorrhagia ( OR  = 1.49, 95% CI : 1.18–1.88). In Table S1 , the model fit results indicated that while AIC, BIC, and aBIC decreased continuously as the number of classes increased, the LMR test became statistically non-significant at the 5-class model ( P  = 0.086). Among the statistically viable 2- to 4-class configurations, the 3-class model achieved the highest entropy value of 0.826, substantially outperforming the 2-class (0.705) and 4-class (0.690) models, thereby demonstrating that the 3-class solution yielded the optimal balance of classification distinctiveness and statistical parsimony. Consequently, three distinct latent classes of sleep quality were identified among the nursing staff based on their conditional probability profiles across seven dimensions (Supplementary Table S2 and Figure S1 ): (1) Class 1 (the Good Sleep Quality Group, accounting for 30.5%) served as the statistical baseline and exhibited the lowest probability of impairment across almost all dimensions; except for a minor elevation in sleep onset latency prolongation (33%), its conditional probabilities for other core components remained remarkably low, with only 9% reporting poor subjective sleep quality and near-zero rates for poor sleep efficiency, sleep disturbances, and medication use. (2) Class 2 (the Severe Sleep Dysfunction Group, accounting for 53.3%) was defined by a multidimensional collapse of sleep functions, where individuals demonstrated extraordinarily high conditional probabilities of poor subjective sleep quality (80%), prolonged sleep onset latency (85%), and severe daytime dysfunction (78%). (3) Class 3 (the Latency‑low Efficiency Group, accounting for 16.2%) was characterized by a distinct pattern of localized sleep deficits, where a striking 100% of the individuals within this class suffered from poor sleep efficiency and 64% experienced prolonged sleep onset latency, yet their reported rates of poor subjective sleep quality (18%), sleep disturbances (4%), and daytime dysfunction (9%) remained at exceptionally low levels. Adjusted models revealed significant associations between sleep characteristics and menstrual disorders (Table  3 ). Higher PSQI scores were associated with higher odds of irregular menstrual cycles ( OR  = 1.04, 95% CI : 1.02–1.07), hypomenorrhea ( OR  = 1.09, 95% CI : 1.06–1.12), and menorrhagia ( OR  = 1.05, 95% CI : 1.02–1.09). Compared with normal sleep duration (6–8 h), sleep duration of less than 6 h was significantly associated with higher odds of irregular menstrual cycles ( OR  = 1.23, 95% CI : 1.03–1.47) and hypomenorrhea ( OR  = 1.31, 95% CI : 1.06–1.62). In addition, Using Class 1 (Good Sleep Quality Group) as the reference category, individuals in Class 2 (Severe Sleep Dysfunction Group) exhibited significantly higher odds of experiencing all examined menstrual disorders in the fully adjusted model (Model 2). Specifically, significant statistical associations were found between Class 2 and irregular menstrual cycles ( OR  = 1.41, 95% CI : 1.16–1.71), dysmenorrhea ( OR  = 1.24, 95% CI : 1.05–1.46), hypomenorrhea ( OR  = 1.68, 95% CI : 1.33–2.12), and menorrhagia ( OR  = 1.34, 95% CI : 1.04–1.74). Table 3 Associations between sleep characteristic and menstrual disorders ( N  = 3544) Sleep characteristic N Irregular menstrual cycles Dysmenorrhea Hypomenorrhea a Menorrhagia a Model 1 Model 2 Model 1 Model 2 Model 1 Model 2 Model 1 Model 2 PSQI score 3544 1.06 (1.02–1.09)* 1.04 (1.02–1.07)* 1.04 (1.02–1.06)* 1.02 (0.99–1.04) 1.11 (1.08–1.14)* 1.09 (1.06–1.12)* 1.07 (1.04–1.11)* 1.05 (1.02–1.09)* Sleep quality  Good 1172 Ref Ref Ref Ref Ref Ref Ref Ref  Poor 2372 1.44(1.20–1.74)* 1.24 (1.02–1.51)* 1.28 (1.11–1.48)* 1.05 (0.90–1.23) 1.68 (1.33–2.12)* 1.36 (1.06–1.75)* 1.39 (1.08–1.78)* 1.16 (0.89–1.51) Sleep onset latency  ≤ 30 min 1963 Ref Ref Ref Ref Ref Ref Ref Ref  > 30 min 1581 1.26 (1.06–1.50)* 1.19 (1.00–1.42)* 1.08 (0.94–1.24) 1.00 (0.87–1.16) 1.58 (1.29–1.94)* 1.46 (1.18–1.79)* 1.34 (1.06–1.68)* 1.26 (1.00–1.58) Sleep duration  < 6 h 1316 1.34 (1.12–1.59)* 1.23 (1.03–1.47)* 1.35 (1.13–1.60)* 0.98 (0.65–1.46) 1.47 (1.19–1.81)* 1.31 (1.06–1.62)* 1.37 (1.08–1.72)* 1.25 (0.99–1.58)  6–8 h 2057 Ref Ref Ref Ref Ref Ref Ref Ref  > 8 h 171 1.04 (0.70–1.55) 1.05 (0.70–1.57) 1.01 (0.68–1.51) 1.21 (0.80–1.83) 0.78 (0.45–1.36) 0.79 (0.45–1.37) 0.78 (0.41–1.47) 0.79 (0.42–1.49) Sleep efficiency  < 80% 1551 Ref Ref Ref Ref Ref Ref Ref Ref  ≥ 80% 1993 1.01 (0.81–1.27) 0.95 (0.80–1.13) 1.12 (0.97–1.28) 0.94 (0.82–1.08) 0.77 (0.63–0.94)* 0.82 (0.67–1.01) 0.96 (0.76–1.21) 1.02 (0.81–1.28) Chronotype  Morning type 784 Ref Ref Ref Ref Ref Ref Ref Ref  Intermediate type 1304 1.21 (0.96–1.53) 1.16 (0.92–1.47) 1.35 (1.12–1.62)* 1.29 (1.08–1.56)* 1.24 (0.93–1.64) 1.16 (0.87–1.54) 1.09 (0.81–1.48) 1.03 (0.76–1.40)  Evening type 1456 1.12 (0.89–1.41) 1.06 (0.84–1.33) 1.17 (0.97–1.40) 1.09 (0.91–1.32) 1.17 (0.88–1.55) 1.07 (0.81–1.42) 1.15 (0.85–1.55) 1.07 (0.79–1.45) PSQI latent categories  Class 1 1890 Ref Ref Ref Ref Ref Ref Ref Ref  Class 2 1079 1.64 (1.37–1.97)* 1.41 (1.16–1.71)* 1.49 (1.28–1.74)* 1.24 (1.05–1.46)* 1.78 (1.31–2.40)* 1.68 (1.33–2.12)* 1.74 (1.22–2.47)* 1.34 (1.04–1.74)*  Class 3 575 1.01 (0.79–1.29) 0.91 (0.71–1.17) 1.29 (1.07–1.56)* 1.14 (0.94–1.39) 0.84 (0.62–1.13) 1.09 (0.81–1.48) 1.11 (0.79–1.56) 0.90 (0.64–1.28) Note: Model 1 was adjusted for outcome–specific covariates. For irregular menstrual cycles, the model included age, BMI, smoking, alcohol consumption, physical activity, bedtime screen time, pregnancy history, and childbirth history; for dysmenorrhea, age, BMI, marital status, education, annual income, smoking, physical activity, bedtime screen time, pregnancy history, and childbirth history were included; for both hypomenorrhea and menorrhagia, the model included age, BMI, marital status, smoking, alcohol consumption, physical activity, pregnancy history, and childbirth history Model 2 was further adjusted for anxiety and depression based on Model 1 PSQI latent categories: Class1: Good Sleep Quality Group; Class2: Severe Sleep Dysfunction Group; Class3: Latency‑low Efficiency Group * P  < 0.05, indicating a statistically significant difference a Multinomial logistic regression analysis for the three-category menstrual volume outcome ( N  = 3,544), comprising hypomenorrhea ( n  = 447) and menorrhagia ( n  = 353), with “Normal Volume” ( n  = 2,744) serving as the reference category Associations between sleep characteristic and menstrual disorders ( N  = 3544) Note: Model 1 was adjusted for outcome–specific covariates. For irregular menstrual cycles, the model included age, BMI, smoking, alcohol consumption, physical activity, bedtime screen time, pregnancy history, and childbirth history; for dysmenorrhea, age, BMI, marital status, education, annual income, smoking, physical activity, bedtime screen time, pregnancy history, and childbirth history were included; for both hypomenorrhea and menorrhagia, the model included age, BMI, marital status, smoking, alcohol consumption, physical activity, pregnancy history, and childbirth history Model 2 was further adjusted for anxiety and depression based on Model 1 PSQI latent categories: Class1: Good Sleep Quality Group; Class2: Severe Sleep Dysfunction Group; Class3: Latency‑low Efficiency Group * P  < 0.05, indicating a statistically significant difference a Multinomial logistic regression analysis for the three-category menstrual volume outcome ( N  = 3,544), comprising hypomenorrhea ( n  = 447) and menorrhagia ( n  = 353), with “Normal Volume” ( n  = 2,744) serving as the reference category Mediation analysis showed that sleep quality had a significant indirect effect on the association between shift work and irregular menstrual cycles, accounting for approximately 7.36% of the total effect. No significant mediation effect was observed for dysmenorrhea. For hypomenorrhea and menorrhagia, the ACME were statistically significant, whereas the total effects were not. Notably, a suppression effect was observed for hypomenorrhea, indicating that the mediation patterns were not consistent across outcomes. Overall, the mediating role of sleep quality appears to be limited and outcome-specific. Detailed mediation pathways are presented in Fig.  2 . Fig. 2 Mediation analysis of sleep quality on the association between shift work and menstrual disorders. Proportion estimates of the association between shift work mediated by sleep quality and ( a ) Irregular menstrual cycles, ( b ) Dysmenorrhea, ( c ) Hypomenorrhea and ( d ) Menorrhagia. ACME: average causal mediation effects; ADE: average direct effects; Total effect=ACME + ADE; Prop.Mediated: Proportion of mediation effect. a For mediation models of menstrual blood volume (hypomenorrhea and menorrhagia), the reference category is “Normal Volume” ( N  = 2,744). The analytical sample sizes are N  = 3191 for hypomenorrhea and N  = 3097 for menorrhagia Mediation analysis of sleep quality on the association between shift work and menstrual disorders. Proportion estimates of the association between shift work mediated by sleep quality and ( a ) Irregular menstrual cycles, ( b ) Dysmenorrhea, ( c ) Hypomenorrhea and ( d ) Menorrhagia. ACME: average causal mediation effects; ADE: average direct effects; Total effect=ACME + ADE; Prop.Mediated: Proportion of mediation effect. a For mediation models of menstrual blood volume (hypomenorrhea and menorrhagia), the reference category is “Normal Volume” ( N  = 2,744). The analytical sample sizes are N  = 3191 for hypomenorrhea and N  = 3097 for menorrhagia To address the potential overestimation of ORs for dysmenorrhea, we re-estimated the associations using robust Poisson regression to obtain PRs. As shown in Supplementary Table S3 and Table S4, the PRs were consistent with the ORs from the main logistic regression analyses in terms of direction, magnitude, and statistical significance. For example, compared with individuals without a history of shift work, the OR for dysmenorrhea among those with shift-work history was 1.71 (95% CI: 1.42–2.06), with a corresponding PR of 1.27 (95% CI: 1.16–1.40). Likewise, the associations for sleep characteristic exposures (e.g., Chronotype, PSQI latent categories) showed comparable ORs and PRs estimates. These findings suggest that the potential overestimation of ORs for dysmenorrhea does not materially affect the interpretation of our main conclusions.

The present study demonstrates that shift work is significantly associated with increased odds of menstrual disorders—specifically irregular menstrual cycles and dysmenorrhea—among nurses in China. Notably, a higher monthly frequency of evening or night shifts was associated with a greater likelihood of dysmenorrhea. Mediation analysis showed that sleep quality accounted for a significant but modest proportion of the total effect of shift work on irregular menstruation via an indirect pathway. Collectively, these findings suggest that shift work may influence menstrual health through both direct and indirect pathways, with sleep disruption playing a limited mediating role. The results align with findings from previous studies. A cross-sectional study of nurses in China reported that night shifts and rotating shift work have significant adverse effects on menstrual disorders [ 41 ]. Furthermore, working more than five night shifts per month or more than three consecutive night shifts has been linked to elevated immune cell counts [ 42 ], which may contribute to immune dysregulation in the reproductive tract and disrupt cyclic ovulation [ 43 ]. Additionally, in the present study, sleep quality was positively associated with the likelihood of menstrual disorders. This finding is supported by a substantial body of literature linking sleep disturbances to menstrual disorders. For example, Jeong et al. reported that components of the PSQI, including subjective sleep quality, sleep latency, sleep disorders, use of sleep medications, and daytime dysfunction, were significantly associated with the frequency and severity of dysmenorrhea [ 44 ]. Similarly, a study of female college students reported a significantly higher prevalence of irregular menstrual cycles, menorrhagia, dysmenorrhea, and premenstrual syndrome among those suffering from sleep disorders [ 45 ]. Another study noted that short sleep duration was associated with a 44% increased likelihood of irregular menstrual cycles, while greater insomnia severity, poor sleep quality, and more daytime sleepiness also predicted cycle irregularity [ 46 ]. In our analyses, we additionally adjusted for depressive and anxiety symptoms as covariates; the significant associations between shift work, sleep quality, and menstrual disorders, as well as the mediating effect of sleep quality, remained essentially unchanged. These findings suggest that depressive and anxiety symptoms do not act as major confounders or indispensable mediators in the pathway linking shift work to menstrual disorders via sleep disturbance. Rather, the robustness of our results after controlling for emotional states indicates that the detrimental effects of shift work on menstrual health are primarily driven by circadian and sleep-related physiological mechanisms that operate independently of psychological distress. The study employed a mediation model to examine the role of sleep in the association between shift work and menstrual disorders. The results suggests that sleep quality may represent one of the potential pathways through which shift work affects menstrual cycle regularity. From a biological perspective, this mediating effect is likely mediated by circadian and endocrine mechanisms. Melatonin, a neurohormone that regulates multiple circadian functions, is strongly influenced by retinal photostimulation [ 47 , 48 ].Shift work, particularly night shifts involving exposure to artificial light at night, may suppress nocturnal melatonin secretion, a key regulator of circadian timing [ 49 ]. Such suppression has been associated with elevated levels of estradiol and gonadotropins, thereby disrupting normal ovulatory function and menstrual cyclicity [ 50 ]. In addition, melatonin may modulate the hypothalamic-pituitary-ovarian (HPO) axis, further affecting the production of ovarian steroid hormones [ 51 ]. The uterine and ovarian cycles are tightly regulated by hormones such as estrogen, progesterone, luteinizing hormone, and follicle-stimulating hormone [ 52 ], and sleep disturbances may alter these hormonal profiles, impair ovulation, and exacerbate dysmenorrhea [ 53 ]. Moreover, night shift work and sleep deprivation may induce oxidative stress, which has been shown to impair oocyte quality and reproductive function [ 54 ]. Furthermore, due to the close interaction between the hypothalamic-pituitary-adrenal (HPA) axis and the HPO axis, persistent circadian misalignment and stress-related cortisol dysregulation may interfere with the pulsatile secretion of gonadotropin-releasing hormone (GnRH), thereby affecting ovulation and menstrual cyclicity. Collectively, melatonin suppression, cortisol dysregulation, and oxidative stress may represent potential biological mechanisms linking shift work and sleep disturbances to irregular menstruation and other menstrual disorders [ 55 , 56 ]. However, the mediating role of sleep varied across different menstrual outcomes. For dysmenorrhea, although shift work was positively associated with its occurrence, the mediating effect of sleep did not reach statistical significance, suggesting that this association may be primarily driven by non-sleep-related pathways, such as inflammatory responses, psychological stress, or autonomic nervous system dysregulation [ 57 ]. Similarly, the role of sleep in menstrual volume abnormalities was inconsistent. A suppression effect was observed in hypomenorrhea, where the indirect and direct effects were in opposite directions, resulting in a negative proportion mediated. This finding suggests that sleep may function as a competing pathway in this association. Taken together, these findings indicate that the role of sleep in the relationship between shift work and menstrual disorders is limited and outcome-specific. While sleep may serve as an important behavioral and biological pathway linking shift work to certain menstrual disturbances, its explanatory power appears modest, and both the direction and magnitude of its effects vary across outcomes. Moreover, the influence of psychological factors, such as anxiety and depression symptoms, should not be overlooked, as they may substantially modify or confound sleep-related associations. Therefore, future studies employing longitudinal designs or structural equation modeling are warranted to further clarify causal relationships and multiple mediation pathways, thereby enhancing the robustness of the findings. A key strength of this study is the comprehensive assessment of multiple shift work characteristics (type, frequency, and duration) and their association with menstrual disorders. In addition, the use of LCA allowed for a nuanced identification of at-risk sleep profiles, and formal mediation testing provided insight into the potential pathway through which shift work impacts menstrual health. Several limitations of this study should be acknowledged. First, the cross-sectional design inherently limits causal inference, as the temporal sequence between shift work and menstrual outcomes cannot be firmly established. Notably, our assessment of shift work relied on lifetime cumulative exposure, which may not adequately reflect the temporal proximity of the exposure to current menstrual outcomes. This potential misalignment could lead to a dilution of the observed effects, potentially biasing the results toward the null hypothesis. Future research should employ prospective designs with repeated, time-specific exposure assessments to more accurately capture the dynamic relationship between shift work and reproductive health. Second, both shift work and menstrual outcomes were self-reported, introducing potential recall or misclassification bias. Specifically, menstrual blood volume was estimated based on self-reported sanitary pad use and an operational classification definition rather than objective measures, which may introduce misclassification due to variability in product absorbency, individual hygiene practices, and recall bias. While we adjusted for key lifestyle and socioeconomic factors, residual confounding from unmeasured occupational stressors (such as job strain, exposure to endocrine-disrupting chemicals in the hospital environment, or departmental workload differences) cannot be entirely ruled out [ 58 – 60 ]. Third, this study examined associations between multiple shift work characteristics (e.g., shift type, years of shift work, shift frequency) and several menstrual outcomes (e.g., cycle regularity, amount of bleeding, dysmenorrhea). Given the number of comparisons, the observed findings may be subject to an increased risk of Type I error. The exploratory nature of these analyses should be emphasized, and replication in independent cohorts is needed to confirm the robustness of the findings. Furthermore, it should be noted that dysmenorrhea is a common outcome in our study population (prevalence 54.6%). The ORs derived from logistic regression may overestimate the true association relative to PRs. However, sensitivity analysis using robust Poisson regression showed that the PRs were consistent with the ORs in direction, magnitude, and statistical significance, supporting the robustness of our findings. Nevertheless, the ORs magnitude should be interpreted with caution, and readers are encouraged to refer to the PRs from the sensitivity analysis for more direct estimates of the strength of association. Finally, physical activity was assessed using a single item rather than a validated scale, which may limit measurement precision and suggests that more detailed assessments are warranted in future studies. Additionally, as the sample was restricted to six hospitals in Anhui Province, multi-center studies across more diverse regions and healthcare systems are required to enhance the generalizability of our findings.

In summary, shift work and poor sleep quality are associated with menstrual disorders among nurses. Mediation analysis suggests that sleep quality may represent a potential pathway underlying this association, with a statistically significant mediating effect observed in irregular menstruation. These findings highlight the potential importance of incorporating sleep health considerations into occupational strategies aimed at supporting the reproductive well-being of shift-working women. Future longitudinal studies are warranted to further elucidate the temporal relationships, and interventions targeting shift scheduling practices alongside sleep hygiene may hold promise for mitigating these risks.

Shift work, characterized by non-standard working hours, is a common occupational practice across various industries, particularly in healthcare. Shift schedules generally consist of three categories: permanent shifts, continuous shifts, and rotating shifts, which may or may not include night shifts [ 1 ]. These shift patterns can be either fixed or variable. The traditional three-shift system typically comprises 8-hour shifts, including day shift (8:00 to 16:00), evening shift (16:00 to 24:00), and night shift (24:00 to 8:00) [ 2 ]. Alternatively, 12-hour shifts (day shift: 8:00 to 20:00, night shift: 20:00 to 8:00) are also widely adopted [ 3 ]. Shift work is highly prevalent among healthcare workers. In the United States, a large multi-state study of 4,540 registered nurses across 13 states and Washington, D.C. reported that over one-third worked night shifts and approximately 10–14% worked rotating shifts [ 4 ]. In comparison, in China, approximately 61.6% of female medical workers have experienced shift work, and 44.6% of nurses follow evening or night shift schedules monthly [ 5 , 6 ]. Given the high prevalence of shift work among nurses, understanding its impact on reproductive health has become a critical public health issue. In 2007, the International Agency for Research on Cancer (IARC) classified shift work as a probable human carcinogen (Group 2 A) [ 7 ]. Epidemiological studies have consistently linked shift work, particularly night shifts, to adverse reproductive health outcomes, including hormonal imbalances, menstrual disorders, preterm births, miscarriages, and decreased fertility rates [ 8 – 12 ]. Specifically, shift work has been associated with irregular menstrual cycles, dysmenorrhea, premenstrual syndrome [ 13 , 14 ].A meta-analysis revealed that women engaged in shift work face a 1.35-fold higher risk of menstrual disorders compared to those working regular daytime hours, with women aged 30 years and older being particularly susceptible to dysmenorrhea [ 15 ]. This association is further supported by a secondary analysis of the 2010–2012 Korean National Health and Nutrition Examination Survey. Among 2,418 female workers, 22.1% were identified as shift workers. While the weighted prevalence of irregular menstrual cycles was 14.3% in the overall sample, it rose to 29.2% among shift workers, underscoring a markedly higher risk of menstrual irregularities in this subgroup [ 16 ]. Irregular menstruation is not only a symptom of reproductive disorders such as premature ovarian failure, polycystic ovarian syndrome, and infertility [ 17 , 18 ], but is also linked to metabolic diseases, including obesity, hypertension, diabetes, and heart disease [ 19 – 21 ]. Sleep disturbances are another significant concern among shift worker, with 46.9% of nurses reporting sleep disorders [ 22 ]. Nurses working shifts, particularly night shifts, experience poorer sleep quality compared to those on regular day shifts. Poor sleep quality is associated with a range of physical and mental health issues, including appetite loss, depression, impaired concentration, and obesity [ 23 – 26 ]. There is growing evidence suggests that sleep problems may also adversely affect menstrual health. Ecochard et al. analyzed data from 27,000 ovarian cycles and found that ovarian cycles are highly correlated with circadian rhythms, and that sleep disruption may affect hormone secretion, which in turn may interfere with menstrual patterns [ 27 ]. Maher et al. demonstrated that poor sleep quality (e.g., difficulty falling asleep, sleep fragmentation) was an independent predictor for menstrual disorders, with each unit decrease in the PSQI score raising the odds of menstrual cycle change by 12%, by conducting a cross-sectional survey study of 1,034 U.S. women of reproductive age [ 28 ]. Shift work-induced sleep disruptions may disrupt the body’s circadian rhythm, leading to endocrine dysregulation, which in turn affects the physiological cycle of women [ 29 ]. Despite evidence linking shift work and sleep patterns to menstrual disorders, the mediating role of sleep quality in this pathway remains underexplored. We therefore hypothesize that sleep quality serves as a key mediator in the relationship between shift work and menstrual health. This study aims to: (1) analyze the associations between shift work characteristics, sleep quality, and menstrual outcomes—including menstrual irregularity, dysmenorrhea, and abnormal volume (hypomenorrhea and menorrhagia)—while adjusting for potential confounders; and (2) examine whether sleep quality mediates these associations. The findings are expected to inform evidence-based strategies for optimizing nursing shift schedules and improving sleep hygiene, ultimately enhancing the reproductive health and professional well-being of nursing personnel.

Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1