The associations of sedentary behaviour and physical activity with musculoskeletal symptoms in hybrid office workers: a compositional data analysis Authors: Iris Parés-Salomón, Eva Cirera-Viñolas, Alan Coffey, Bette Loef, Kieran P Dowd, Anna M Señé-Mir, Arnau Gustems-Morral, Izabela Luznik, Maja Pajek, Karin I Proper, Anna Puig-Ribera, Judit Bort-Roig Published: 2026 ABSTRACT AbstractBackgroundHybrid work, combining home-office work and working in the office, is the new norm for desk-based jobs. This work modality further increases sedentary behaviour and may subsequently increase the behaviours related effects on health, such as musculoskeletal symptoms. However, little is known about accumulated physical activity (PA) patterns and sedentary behaviours (SB) (and their associations with health) since the adoption of hybrid work after the COVID-19 pandemic. Therefore, the aims of the study are to (i) examine patterns of SB and PA among hybrid office workers, (ii) assess their associations with musculoskeletal symptoms, and (iii) assess the impact of replacing time spent in SB with PA on musculoskeletal symptoms.MethodsIn this cross-sectional study, hybrid office employees from Ireland, Slovenia, Spain and the Netherlands were enrolled. Self-reported measures assessed sociodemographics. Musculoskeletal symptoms were measured using the Standardized Nordic Questionnaire as the presence of any discomfort in the past 7 days. Total, occupational and non-occupational SB, standing, light-intensity PA and moderate-to-vigorous PA (MVPA) were assessed by the activPAL™. A compositional data analysis was employed to investigate the associations between SB and PA patterns and musculoskeletal symptoms. For significantly associated patterns, isotemporal substitution analyses were conducted.ResultsA total of 163 hybrid office workers (71.8% female; median age 39 years) participated in the study. Overall, 51.5% of participants reported musculoskeletal symptoms in at least one body region. On average, participants sat for 11.08 h/day, accumulating 20 min more sitting time while working during home-office work than in the traditional office location (p < 0.001), with greater proportion of prolonged sitting bouts (p = 0.041). The proportion of sitting time relative to the other 24-hour day behaviours on days of home-office work was associated with musculoskeletal symptoms (OR: 2.75, 95% CI: 1.24, 6.55). Reallocating sitting time with MVPA was associated with a lower predicted probability of experiencing musculoskeletal symptoms at home, with greater differences observed at higher amounts of substitution.ConclusionsSitting time during home-office working days is associated with musculoskeletal symptoms, while replacing it with MVPA is associated with lower odds of such symptoms. These findings may inform future strategies to reduce SB and promote PA in hybrid work contexts across a 24-hour framework.Trial registrationClinicalTrials.gov NCT06247228. Registered 30 January 2024.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12889-026-28107-4.Keywords: Sitting time, Physical activity, Musculoskeletal health, Hybrid work, Home-office work Background Hybrid office work, which combines home-office work and working in the office, is currently the mode of work for many desk-based jobs, with around 44% of European employees adopting such arrangements [1, 2]. Traditional office work involves high levels of prolonged sitting [3], which may increase further when working from home [4, 5]. To date, most existing evidence on sedentary behaviour (SB) and physical activity (PA) patterns among hybrid workers has been gathered during the COVID-19 pandemic [6]. However, these data may not be representative of the current hybrid work context, as pandemic restrictions could have influenced behaviour patterns [7]. Therefore, further investigation is needed to understand SB and PA patterns in the post-pandemic hybrid work environment [8]. Occupational SB has been associated with harmful health effects, with poor musculoskeletal health being among the most prevalent, particularly among office workers [9]. Musculoskeletal symptoms, defined as discomfort or pain in body structures such as muscles, tendons, joints, ligaments, nerves or bones [10], have also been associated with occupational SB in home-office settings [11]. These symptoms are a leading cause of additional work-related outcomes, such as sickness absence and lost presenteeism or reduced productivity [12, 13]. However, few studies have investigated the relationship of SB and PA patterns with health among hybrid office work [11]. Therefore, it is essential to understand how hybrid work shapes SB and PA patterns in the current work context to better assess their potential contribution to musculoskeletal risks. To mitigate the detrimental effects of prolonged SB (i.e., > 30 min [14]), studies have recommended reducing and breaking up extended sitting time by replacing it with PA [15]. In fact, emerging evidence suggests that different intensities of PA may differentially influence musculoskeletal symptoms, with more consistent beneficial associations observed for higher-intensity activities [16, 17]. Moreover, the associations between replacing sitting time with more active behaviours and musculoskeletal symptoms may also depend on sleep duration [18], as both insufficient and excessive sleep have been associated with adverse musculoskeletal symptoms [19, 20], highlighting the interdependent nature of sleep, SB, and PA within the 24-hour movement behaviour framework. However, few studies have considered SB, PA and sleep as co-dependent behaviours (i.e., a change in one behaviour affects the others) to explore their associations with musculoskeletal symptoms, and none has applied a compositional data analysis (CoDA) approach in hybrid office workers or compared movement patterns between home and office contexts [21, 22]. CoDA enables the examination of associations with different outcomes by assessing the effects of reallocating time between the three behaviours within a closed 24-hour compositional framework [21]. Therefore, understanding the 24-hour composition of these behaviours in hybrid workers is important to better characterise their associations with health and to inform interventions and the development of 24-hour guidelines for SB, PA, and sleep [23, 24]. It is also important to consider these contextual differences in relation to the 24-hour composition, as home and office environments may influence movement behaviour patterns in different ways. Therefore, the aims of this study are (i) to examine hybrid office workers SB and PA patterns in the home and office contexts; (ii) to assess the associations between SB and PA patterns with musculoskeletal symptoms; and (iii) to assess the associations of replacing time spent in SB with PA on musculoskeletal symptoms. Materials and methods Study design and participants This study used cross-sectional data from the baseline assessment of the Click2Move program, a two-arm cluster randomized controlled trial evaluating the effectiveness of a multicomponent digital-based intervention in reducing occupational SB and its associated effects on health and work-related outcomes among hybrid office workers in Ireland, Slovenia, Spain, and the Netherlands. The study description has been provided elsewhere [25]. In brief, hybrid workers in an office setting performing at least one day per week of home-office work were eligible to participate. Hybrid office workers were recruited from five companies across the four countries involved in the study (one from Ireland, two from Slovenia, one from Spain, and one from the Netherlands) through company internal communication. The baseline sample included 163 participants (38 from Ireland, 35 from Slovenia, 48 from Spain and 42 from the Netherlands). Based on predefined validity criteria for activPAL data (minimum wear-time and valid working days), 24 participants were excluded from the compositional analyses, resulting in an analytical sample of 139 participants. Of these, 3 were further excluded as they had no home-office working days recorded during the monitoring period and therefore could not be classified as hybrid workers, yielding a final analytical sample of 136 participants. Measures Sociodemographic measures An 11-item survey assessed the sociodemographic outcomes (gender, age, education level, household composition, weight, height, years in the current job, days of home-office per week, distance from home to work, mode, and time of commute). Musculoskeletal symptoms Eight dichotomous questions from the Standardized Nordic Questionnaire (SNQ) evaluated the presence of musculoskeletal discomfort in the last 7 days [26]. To facilitate the description of the musculoskeletal symptoms, the variables on presence have been classified by anatomical regions into four categories [27]: (a) upper body, encompassing the neck and shoulders; (b) back, including the upper and lower back; (c) legs, which consist of the hips, knees, ankles, and feet; and c) arms, covering the elbows and wrists symptoms. A new dichotomous variable was created based on the crude variables for individuals who had symptoms in at least one body part. The SNQ captures musculoskeletal symptoms within the previous 7 days, without distinguishing between acute and chronic conditions. Sedentary behaviour and physical activity patterns The activPAL3TM (PAL Technologies Ltd., Glasgow, UK) device is a valid measure to quantify body posture and SB and PA [28]. The device was attached to the midline of the anterior aspect of the participants’ right thigh using a flexible nitrile sleeve and a transparent film (10 × 10 cm of hypoallergenic Tegaderm™ Foam Adhesive Dressing). Participants wore the activPAL3TM for 7 consecutive days, 24 h per day. Participants were asked to record their daily wake time, bedtime, working hours, working location (office vs. home-office), the mode of commute and any monitor removal time by completing a diary. The activPAL3TM device (PAL Technologies Ltd., Glasgow, UK) was utilized to measure and quantify the sitting time and duration of sitting bouts, standing time, light-intensity PA (LIPA), moderate-to-vigorous PA (MVPA) and bed time hours (estimated as sleep time). Data processing was conducted following a previously described protocol [29]. Briefly, to be included in the analysis, participants had to provide a minimum of 3 working days (including 1 home-office day). Non-wear time was identified as periods of 60 min or more of consecutive zero activity counts in the vertical axis of the activPAL data. If a period of non-wear time was detected, the data for this day was removed from the analysis. Data from the activPAL was first processed and analysed using the PAL Analysis software (version 8.11.8.75). Waking periods were estimated based on participants’ self-reported rise and bed times as previously described [29], and were confirmed by visual inspection of the activPAL data. Working hours were identified using participant diaries to extract the self-reported start and end times of work for each day, along with the work location (office vs. home-office). To improve accuracy, the reported start time was aligned with the onset of the first prolonged sedentary bout occurring around that time, assuming participants typically begin working with a period of sitting. These reported time windows were used to segment the activPAL data into working and non-working time segments for each valid day. Working hours were extracted by aligning the time-stamped activPAL data with these diary-reported time windows. The remainder of the waking day was classified as non-working time. Proprietary algorithms classified 15 s epoch data into time spent sitting, standing and stepping, while accelerometer counts were also produced. Although the PAL Analysis software exports data in 15-second epochs for the initial classification of postures (sitting/lying, standing, and stepping), the analysis of SB primarily relied on an event-based approach. This method identifies postural transitions from the accelerometer data recorded by the device, allowing for more precise estimation of continuous periods spent in a sedentary posture. This approach mitigates potential limitations associated with epoch aggregation. Time spent sitting and standing was summed for waking periods for each day. Total waking time spent in LIPA and MVPA was determined according to previously validated count-to activity thresholds [30]. MVPA was determined using a threshold of 8873 counts per 15s, which has been demonstrated to accurately identify MVPA in adult populations [30]. Daily sedentary bout durations were also examined, whereby the specified durations of sedentary bouts (< 30 min, 30–60 min and > 60 min) and the amount of time spent in these sedentary periods during waking hours was quantified using a MATLAB v8.4 (MathWorks®, Natick, MA, USA) program [29]. Statistical analysis Descriptive data analysis was performed using the Statistical Package for the Social Sciences (SPSS®) (Version 29.0; IBM, Armonk, NY, USA), while Compositional Data Analysis (CoDA) and logistic regression analyses were performed in RStudio using Compositions and lme4 packages [31, 32]. A significance level of p < 0.05 was set for all statistical tests. Descriptive statistics were conducted reporting frequencies and proportions for categorical variables, and medians and interquartile ranges (IQRs) for continuous variables. Geometric means were also reported for the five compositional behaviours (sitting, standing, LIPA, MVPA and sleep) to summarise the central tendency of the distributions. Three compositional frameworks were defined to characterize SB and PA patterns: a 24-hour composition (sitting, standing, LIPA, MVPA and sleep), a working-period composition (occupational sitting—further partitioned into bouts of < 30 min, 30–60 min and > 60 min—standing, LIPA and MVPA) and a non-working-period composition (leisure sitting, standing, LIPA, MVPA and sleep). Zero or missing values in these compositions were imputed using the log-ratio expectation-maximization algorithm (IrEM; detection limit = 15 s). To allow meaningful comparisons across contexts, the working-period and non-working-period compositions were rescaled to 8 h and 16 h, respectively. The five compositional behaviours were transformed using isometric log-ratios (ilr) to account for the compositional nature of time-used data. In the standard transformation, ilr₁ contrasted sitting against the remaining behaviours, and subsequent ilrx sequentially contrasted the next behaviour against those not yet referenced. To assess differences in SB and PA patterns across occupational contexts, multivariate analysis of variance (MANOVA) was applied to the ilr-transformed variables. When statistically significant differences were identified, follow-up T-tests were performed on ilr coordinated. Given the multiple comparison involved, p-values were adjusted using the Benjamini–Hochberg procedure to control the false discovery rate [33]. Results from these tests should be interpreted in as relative differences within the composition, rather than absolute time differences in individual behaviours. Multiple logistic regression models were used to examine the associations between ilr-transformed compositional SB and PA patterns for each compositional framework (i.e., 24 h, working-period and non-working-period) and the presence of musculoskeletal symptoms (for each body region and, in at least one of them). For compositions showing statistically significant associations, sensitivity analyses were performed by redefining the ilr coordinates so that each behaviour was contrasted against all remaining behaviours, rather than using the standard sequential ilr transformation. A multivariate analysis adjusting for gender, age and BMI were also performed. To quantify the impact of time reallocations between SB and PA patterns on musculoskeletal outcomes, regression coefficients were interpreted using isotemporal substitution framework. This approach estimates the predicted probability of reporting symptoms when reallocating fixed intervals of time between two behaviours, while holding the total time and all other behaviours constant. Reallocations were applied from sitting time to LIPA and MVPA (in periods of 5 min up to 60 min) for those contexts that were shown to be associated with the dependent variable. Results Descriptive characteristics Characteristics of the participants for sociodemographic data and musculoskeletal symptoms are described in Table 1. The distribution of SB and PA patterns are described throughout the 24 h of workdays, during working hours and non-working hours and differences between home-office work and working in office location (Table 2). Table 1Descriptive characteristics of the included participantsSociodemographic (n = 163)Median (IQR) or n (%)Gender Male46 (28.2%) Female117 (71.8%)Age (years)39 (19)Education level No formal education/Primary education and secondary education14 (8.6%) Higher professional education26 (16%) Bachelor’s degree (University)45 (27.6%) Qualifications higher than bachelor/professional (e.g., Masters)78 (47.9%)Live with Alone17 (10.4%) Partner56 (34.4%) Partner and kids65 (39.9%) Alone with kids7 (4.3%) Other18 (11%)Body mass index (Kg/m2)24.1 (5.17)Job years4 (11)Home-office days < 1 day per week5 (3.1%) 1 day per week19 (11.7%) 2 days per week33 (20.2%) 3 days per week60 (36.8%) 4 days per week38 (23.3%) > 4 days per week8 (4.9%)Distance house – work (km)21 (38)Mode of commute Car or motorbike85 (52.1%) Public transport43 (26.4%) Bicycle19 (11.7%) Walk8 (4.9%) Other8 (4.9%)Time of commute30 (49)Musculoskeletal symptoms Upper body (neck and shoulder)43 (26.4%) Back (upper and lower)41 (25.2%) Legs (hip, knees, ankles, feet)28 (17.2%) Arms (elbows, wrists)14 (8.6%) Any body part84 (51.5)Continuous variables: median (interquartile range). Categorical variables: n (%). BMI body mass index Descriptive characteristics of the included participants Continuous variables: median (interquartile range). Categorical variables: n (%). BMI body mass index Table 2Sedentary behaviour (SB) and physical activity (PA) patterns throughout the whole day, working and non-working timeTotal. working days (h/day)N = 136Home-office days (h/day)Office days (h/day)MANOVApPost hoc p.Working day (24 h) < 0.001 Sitting11.08 (46.2%)10.96 (45.7%)11.19 (46.6%)0.106Standing3.41 (14.2%)3.32 (13.8%)3.62 (15.1%)0.651LIPA0.99 (4.1%)0.98 (4.1%)0.99 (4.1%)0.106MVPA0.33 (1.4%)0.28 (1.2%)0.47 (2%) 0.015 Sleep8.19 (34.1%)8.46 (35.2%)7.72 (32.2%) < 0.001 Sitting bouts 0.015 < 30 min4.5 (40.6%)4.11 (37.5%)4.87 (43.5%)0.10130–60 min2.96 (26.7%)2.97 (27.1%)2.92 (26.1%)0.84> 60 min3.62 (32.7%)3.88 (35.4%)3.41 (30.5%)0.171Working hours (8 h) < 0.001 Sitting6.6 (82.5%)6.71 (83.9%)6.38 (79.7%) < 0.001 Standing1.06 (13.2%)0.97 (12.2%)1.22 (15.3%)0.261LIPA0.27 (3.4%)0.27 (3.4%)0.26 (3.3%) < 0.001 MVPA0.07 (0.8%)0.05 (0.6%)0.14 (1.7%) < 0.001 Sitting bouts 0.003 < 30 min2.21 (33.5%)1.97 (29.4%)2.21 (34.6%)0.57830–60 min1.35 (20.5%)1.27 (19%)1.79 (28.1%)0.399> 60 min3.04 (46%)3.46 (51.6%)2.38 (37.3%) 0.041 Non-working hours (16 h) 0.002 Sitting4.46 (27.9%)4.26 (26.6%)4.76 (29.7%)0.785Standing2.22 (13.9%)2.22 (13.9%)2.27 (14.2%)0.238LIPA0.69 (4.3%)0.68 (4.2%)0.7 (4.4%)0.386MVPA0.19 (1.2%)0.17 (1.1%)0.25 (1.5%)0.185Sleep8.44 (52.8%)8.67 (54.2%)8.02 (50.1%) 0.009 Sitting bouts0.254< 30 min1.57 (35.3%)1.48 (34.7%)1.74 (36.6%)30–60 min1.1 (24.6%)1.03 (24.2%)1.23 (25.8%)> 60 min1.79 (40.1%)1.75 (41.1%)1.79 (37.6%)Values represent geometric expressed as hours per period (% of total period duration). Sitting bouts are expressed as hours (% of total sitting time). The 24-hour composition is standardised to 24 h; working-hour composition to 8 h; non-working-hour composition to 16 h. MANOVA p: overall difference in compositional pattern between occupational contexts (Wilks’ Lambda test on ilr-transformed data). Post-hoc p: adjusted p-values (Benjamini-Hochberg procedure); reflects relative differences within the composition, not necessarily absolute time differences. Bold values indicate p < 0.05. No post-hoc tests were performed for non-working sitting bouts (MANOVA p = 0.254) Sedentary behaviour (SB) and physical activity (PA) patterns throughout the whole day, working and non-working time Values represent geometric expressed as hours per period (% of total period duration). Sitting bouts are expressed as hours (% of total sitting time). The 24-hour composition is standardised to 24 h; working-hour composition to 8 h; non-working-hour composition to 16 h. MANOVA p: overall difference in compositional pattern between occupational contexts (Wilks’ Lambda test on ilr-transformed data). Post-hoc p: adjusted p-values (Benjamini-Hochberg procedure); reflects relative differences within the composition, not necessarily absolute time differences. Bold values indicate p < 0.05. No post-hoc tests were performed for non-working sitting bouts (MANOVA p = 0.254) Sociodemographic data One-hundred and sixty-three hybrid office workers completed the baseline data measurements, with a median age of 39 (IQR: 19) years, 71.8% of which were female. Most participants (85.2%) worked from home at least 2 days per week. Musculoskeletal symptoms Over the last 7 days, 51.5% of participants reported musculoskeletal symptoms in at least one body region. The most affected areas were the upper body (26.4%) and back (25.2%), followed by the legs (17.2%), with the arms the least affected region (8.6%). Sedentary behaviour and physical activity patterns of hybrid office workers and differences between home and office work contexts On average, hybrid employees sat for 11.08 h/day (46.2%), with most SB accumulated in bouts of < 30 min (40.6%; p < 0.001). Regarding PA, participants spent 3.41 h/day standing (14.2%), 0.99 h/day (4.1%) in LIPA and 0.33 h/day (1.4%) performing MVPA. Among PA behaviours, was higher on office-location days compared with home-office days (0.47 h/day, 2% versus 0.28 h/day, 1.2%, respectively; p = 0.015). Overall, daily sleep duration among hybrid office employees was 8.19 h/day (34.1%) and differed between home-office and office location workdays, with a significant increase in sleep duration on days when participants performed home-office work (8.46 h/day; 34.1%; p < 0.001). Working time showed significant variation across both contexts, with participants working 8.08 h/day on home-office days versus 8.39 h/day those days of office work (p = 0.012). During working hours, employees sat on average 6.6 h per 8-hour (h/8 h; 82.5%), with a higher proportion when working from home (6.71; 83.9%; p < 0.001). Sitting time was distributed with long bouts (> 60 min; 3.04 h/8 h, 46%; p = 0.003), a pattern that was more predominant during home-office work compared to office location (3.46 h/8 h, 51.6% versus 2.38 h/8 h, 37.3%; p = 0.041). When working in the office, participants engaged in more MVPA (0.14 h/8 h, 1.7% versus 0.05 h/8 h, 0.6%; p < 0.001) at the expense of time spent sitting and in LIPA (p < 0.001 both), whereas in home-office working, they spent more time in SBs and less in MVPA. During non-working hours, employees sat for an average of 4.46 h/16 h (27.9%), with 1.79 h/16 h (40.1%) accumulated in long bouts. As no differences were observed in sitting bout composition, no further stratified analysis by work context were performed. Although no differences were observed in PA and SB, the composition of non-working hours differed significantly, driven by longer sleep duration on home-office days (8.67 h/16 h, 54.2%) compared with office days (8.02 h/16 h, 50.1%; p = 0.014). Associations of sedentary behaviour and physical activity patterns with musculoskeletal symptoms All compositional logistic regression models were estimated as specified in the analytical framework. Only models referring to the presence of musculoskeletal symptoms in at least one body region are reported in the main text, while full results are provided in the Supplementary Material 1. Table 3 shows the ORs from compositional logistic regression models, where each ilr coordinate represents one behaviour relative to the remining components of the 24-hour composition. ORs reflect associations between relative differences in the 24-hour time-use composition and the odds of reporting symptoms. As ilr coordinates are based on log-ratios, these estimates should be interpreted as relative reallocations within the composition rather than direct time in minutes. These associations are complemented by isotemporal substitution models, which provide an alternative framework estimating changes in predicted probability of symptoms associated with fixed time reallocations between behaviours. Table 3Multiple logistic regression analyses of the association between isometric log-ratios (ilr) coordinates and musculoskeletal symptoms for the whole dayMusculoskeletal symptomsSB and PA patternsSitting(OR, 95% CI)Standing(OR, 95% CI)LIPA(OR, 95% CI)MVPA(OR, 95% CI)Sleeping(OR, 95% CI)Unadjusted modelsTotal day (home-office and office location)3.15 (1.28, 8,33)*0.70 (0.35, 1.36)0.72 (0.35, 1.47)0.66 (0.44, 0.95)*0.19 (0.04, 0.79)*Total day (home-office)2.74 (1.25, 6.46)*0.65 (0.33, 1.20)0.77 (0.39, 1.51)0.74 (0.53, 1.00)0.29 (0.09, 0.86)*Total day (office location)1.47 (0.69, 3.29)0.93 (0.48, 1.77)1.06 (0.56, 2.01)0.82 (0.58, 1.14)0.42 (0.12, 1.32)Adjusted modelsTotal day (home-office and office location)3.28 (1.30, 8.96)0.62 (0.30, 1.24)0.76 (0.36, 1.55)0.68 (0.45, 1.00)0.21 (0.04, 0.90)Total day (home-office)2.68 (1.21, 6.42)0.61 (0.31, 1.15)0.79 (0.39, 1.56)0.76 (0.54, 1.05)0.32 (0.09, 1.00)Total day (office location)2.68 (1.21, 6.42)0.81 (0.40, 1.63)1.13 (0.59, 2.19)0.82 (0.58, 1.16)0.41 (0.11, 1.33)OR odds ratio, CI confidence interval, LIPA light-intensity physical activity, MVPA moderate-to-vigorous physical activityDaily time (sitting, standing, LIPA, MVPA, sleep) was analysed using Compositional Data Analysis (CoDA). Behaviours were transformed into isometric log-ratio (ilr) coordinates, where each coordinate contrasts one behaviour against the geometric mean of all remaining behaviours. Each OR reflects a relative shift in one behaviour within the overall daily time budget, independent of absolute duration. Unadjusted models include ilr coordinates only; adjusted models additionally include sex, age and body mass index. Results are presented for three contexts: total working days, home-office days, and office-location days* p < 0.05 Multiple logistic regression analyses of the association between isometric log-ratios (ilr) coordinates and musculoskeletal symptoms for the whole day OR odds ratio, CI confidence interval, LIPA light-intensity physical activity, MVPA moderate-to-vigorous physical activity Daily time (sitting, standing, LIPA, MVPA, sleep) was analysed using Compositional Data Analysis (CoDA). Behaviours were transformed into isometric log-ratio (ilr) coordinates, where each coordinate contrasts one behaviour against the geometric mean of all remaining behaviours. Each OR reflects a relative shift in one behaviour within the overall daily time budget, independent of absolute duration. Unadjusted models include ilr coordinates only; adjusted models additionally include sex, age and body mass index. Results are presented for three contexts: total working days, home-office days, and office-location days * p < 0.05 Impact of sedentary behaviour and physical activity patterns on musculoskeletal symptoms After adjusting for gender, age and BMI, higher proportion of sitting time relative to the other 24-hour cycle behaviours was associated with higher odds of reporting musculoskeletal symptoms of at least a body part (OR: 3.28; 95% CI: 1.30, 8.96). Whereas a higher proportion of sleeping time were associated lower odds of reporting musculoskeletal symptoms (OR: 0.21; 95% CI: 0.04, 0.90). When sitting bouts were considered, no associations were found. Influence of work context on the associations of sedentary behaviour and physical activity patterns with musculoskeletal symptoms Focusing on work context multivariant models, on home-office days, a higher proportion of sitting time relative to the other 24-hour behaviours was associated with higher odds of reporting musculoskeletal symptoms (OR: 2.68; 95% CI: 1.21, 6.42). No significant associations were found for office days. When only working hours were considered, neither home-office nor traditional office work was associated with the risk of musculoskeletal symptoms. Similarly, when only the non-working time was considered, no statistically significant results were presented for any ilr at either home-office or office location. Associations of replacing time spent in sedentary behaviour with physical activity on musculoskeletal symptoms Figure 1 presents the results of the compositional isotemporal substitution analysis, showing the predicted probability of reporting musculoskeletal symptoms associated with reallocating time from sitting to MVPA and LIPA in increments of 5 to 60 min, reflecting changes in predicted probability relative to the centroid composition after time reallocation between behaviours. Substitutions were applied to the full 24-hour period for workdays (combining both work contexts) and home-office days, as no significant behavioural components were observed for office days. Estimated probability of reporting musculoskeletal symptoms when sitting time was reallocated with moderate-to-vigorous physical activity (MVPA) and light-intensity physical activity (LIPA) in 5- to 60-minute increments Considering the centroid participants’ behavioural composition throughout the 24-hour day, the estimated probability of reporting musculoskeletal symptoms was 0.505, and 0.503 when doing home-office work. Replacing 30 min of daily sitting time with MVPA, was associated with a lower predicted probability of musculoskeletal symptoms to 0.480 when considering all workdays (home and office), and to 0.494 when considering only home-office workdays. When 60 min of sitting time was reallocated by the same amount of MVPA, the estimated probability decreased further, reaching 0.454 and 0.479 respectively. When sitting time was replaced with LIPA, the probability of experiencing musculoskeletal symptoms was reduced; however, this reduction was smaller compared to when sitting time was replaced with MVPA. Discussion The present study provides novel insights into hybrid office work research. Participants spent more sitting time when doing home-office work compared with working in the office, and those with higher proportion of total sitting time, especially on days of home-office work, were more likely to experience musculoskeletal symptoms. The greater the amount and intensity of PA replacing daily sitting time, the greater the reduction in the risk of experiencing musculoskeletal symptoms. Hybrid employees spent more time sitting during home-office work than in traditional office locations, consistent with previous studies [5, 22]. However, this difference did not persist when considering the total 24-hour day, suggesting that changes in sitting time during home-office work may be balanced by differences in non-working time. Among existing literature, results on sedentary bouts are mixed. Existing studies have reported that home-office workers switched from a seated to standing posture more frequently than office workers [4, 34], or found no differences between traditional office and home office settings [22, 35]. Our findings indicated significant differences in the sitting bouts composition during working hours, with a higher proportion of prolonged bouts during home-office work. Previous studies were conducted during the COVID-19 pandemic [22, 35] or did not use the gold standard device-based measure (i.e., activPAL) [4, 22, 34, 35]. Therefore, the findings of increased sedentary time during home-office work may have been influenced by travel restrictions during COVID-19, or by an underestimation of sitting time due to measurement errors [36, 37]. These limitations, along with the heterogeneity observed in previous studies, attributable to the use of varied measurement tools [6], highlight the need for further post-pandemic research using precise device-based measurement tools to yield reliable and realistic insights into the current SB and PA pattern reality among hybrid or home office workers. We found that hybrid office workers were less active during home-office work, with 5.4 min (1.1%) less MVPA and slightly higher LIPA compared with traditional office settings, while standing time did not differ between contexts. Reduced MVPA may be explained by isolation during home-office work, which may reduce engagement in activities such as having a coffee, walking to a meeting room or using the stairs [38]. The higher proportion of LIPA on home-office work may result from small housework tasks performed during working hours. While existing evidence has also shown reduced MVPA during home-office work (both pre- and post-pandemic) in comparison to office location [39, 40], our results regarding LIPA and standing are not comparable with others since no study has considered these behaviours in home or hybrid contexts prior to this study. Future research employing validated device-based measures should consider the distinct features of the different work settings to better capture a whole range of SB patterns and PA intensities supporting the development of context-specific strategies. In this study, participants with a higher proportion of total time spent sitting were more likely to experience musculoskeletal symptoms, with stronger associations observed during home-office working days compared with office-location days. The findings align with existing studies carried out during the COVID-19 pandemic [11, 41]. However, we were unable to compare our results with any study performed in the new hybrid working mode since the associated restrictions from COVID-19 have been lifted, and resulting changes in mobility, commuting, and workplace attendance may have influenced current behavioural patterns compared with the pandemic period. No associations between SB and musculoskeletal symptoms were observed when SB was examined separately during working and non-working periods, which may reflect behavioural compensation outside of work [42]. Nevertheless, total sitting time across the day remained associated with musculoskeletal symptoms, suggesting that the cumulative accumulation of sedentary time throughout the 24-hour period may be relevant for musculoskeletal health [43]. Thus, future research should evaluate the associations between SB and PA patterns and musculoskeletal symptoms among hybrid office workers, considering both occupational and non-occupational contexts, to inform interventions targeting reductions in total daily sedentary time and increases in movement across the whole day. In addition to SB, differences in musculoskeletal symptoms between home-office and traditional office settings may also be influenced by ergonomic and psychosocial factors. Variations in workstation setup, such as the availability of ergonomic chairs, external monitors, or adjustable desks in office environments compared with more improvised home setups, may contribute to differences in musculoskeletal load [44]. Similarly, psychosocial aspects such as work-related stress, autonomy, and changes in work organisation during remote work may also play a role in musculoskeletal outcomes [45]. Future research should also examine potential associations or the mediation role between SB and work-related outcomes such as absenteeism or occupational fatigue, given their reported associations in office workers [46]. We found that replacing sitting with PA was associated with a reduction in the likelihood of experiencing musculoskeletal symptoms, with higher PA intensity yielding greater reduction in risk. While no other CoDA studies have examined hybrid or home-office settings, similar effects have been reported in office employees [16]. They found that as the time replaced by higher intensities of PA increased; the risk of total musculoskeletal symptoms reduced. Specifically, reallocating 30 min of sitting time with MVPA was associated with a 3% relative reduction in the probability of musculoskeletal symptoms across the 24-hour period. Other studies have reported reductions of up to 48% [16], but they did not apply CoDA methods. In our study, replacing 30–60 min of sitting with MVPA during home-office work corresponded to a 4–7% relative reduction in risk, while substitution with LIPA also conferred benefits, although to a lesser extent. The magnitude of these associations is comparable to that reported in other studies applying CoDA methods, which reflect the constrained nature of 24-hour time-use data and help avoid overestimation of effects. Although modest at the individual level, such reductions may still be clinically, and public-health relevant given the high prevalence of musculoskeletal symptoms and their cumulative social burden at the population level [47]. These results highlight the importance of considering work environment when designing health interventions. Future studies are needed to confirm these associations and to develop tailored strategies for the evolving hybrid work models. Strengths and limitations This is the first study in the hybrid office work context providing device-based measured activity data and its associations with musculoskeletal symptoms, using CoDA to enhance robustness of the results. The present study is the starting point of hybrid office workers’ studies giving way to future evidence to investigate and evaluate the associations of SB and PA patterns with other health and work-related outcomes. However, the limitations of this study should be acknowledged. First, the cross-sectional design precludes causal inference, and the observed associations between SB, PA, and musculoskeletal symptoms should therefore be interpreted with caution. The sample was limited to a few companies in four high-income European countries, which may affect the generalisability to other regions or sectors, and a predominance of female participants. Cultural and organizational differences could introduce variability in SB and PA patterns. However, organisational-level factors such as workplace policies, management practices and organisational culture were not explicitly measured or included in the analyses, despite their potential influence on SB and PA behaviours. Musculoskeletal symptoms were assessed using a dichotomous 7-day recall measure, which does not capture symptom chronicity or severity (e.g., number of affected anatomical sites). The 7-day recall period of the Standardized Nordic Questionnaire captures only recent symptom presence and may not reflect usual musculoskeletal symptom patterns, potentially introducing misclassification and attenuating observed associations. Sleep time was estimated based on inactivity detected by the activPAL, which cannot distinguish sleep from periods of wakefulness, and aligning diary-reported times with device data may have introduced minor variability. Sleep quality was not assessed, which may be relevant given the established association between poor sleep quality and musculoskeletal disorders. Finally, MVPA classification using a count-to-activity threshold is validated but not without limitations, although other methods for estimating MVPA also face similar issues, and no approach is without potential error in free-living assessments [48]. Practical implications Interventions should aim to reduce sedentary behaviour by encouraging PA throughout the day, particularly in hybrid and home-office settings. Practical strategies to achieve a daily substitution of 30 to 60 min from sitting to PA may include a range of intensities, such as taking regular movement breaks during prolonged screen-based work, using activity prompts or scheduling reminders in digital calendars, and integrating light activity during remote meetings (e.g., standing or walking calls). More intense PAs, such as brisk walking during lunch breaks or short bouts of structured exercise before or after work, may provide additional health benefits. These insights may inform future interventions designed for hybrid work arrangements across both occupational and non-occupational contexts. Conclusion This study provides the first device-based estimates of SB and PA patterns in hybrid office workers, including sitting, standing, LIPA, MVPA and sleeping time throughout the 24-hour cycle, during both working and non-working hours. During working hours, hybrid office workers were more sedentary and perform less MVPA when working from home than in the traditional office setting. Furthermore, total sitting time, especially when working from home, was associated with a higher likelihood of experiencing musculoskeletal symptoms, and replacing sitting with PA, especially MVPA, was associated with a lower likelihood of such symptoms. These findings highlight the potential relevance of approaches addressing SB and PA in hybrid work arrangements, considering both home and hybrid office settings, as well as occupational and non-occupational contexts within a 24-hours framework, and their associations with musculoskeletal symptoms. Supplementary Information Supplementary Material 1. Acknowledgements The authors gratefully acknowledge all the companies and participants for being part of the study. Abbreviations Sedentary behaviour Physical activity Standardized Nordic questionnaire Light intensity physical activity Moderate-to-vigorous physical activity Statistical Package for the Social Sciences Compositional data analysis Interquartile ranges Isometric log-ratios Multivariate analysis of variance Odd ratio Square meters World Health Organization Confidence interval Authors’ contributions IPS, ECV, and JBR made substantial contributions to the conception and design of the study, acquisition, analysis, and interpretation of data. IPS, ECV, and JBR drafted the manuscript, while all authors critically revised it for important intellectual content. KPD contributed significantly to the design of the activPAL protocol and data analysis. All authors were involved in company selection and data collection within their respective countries. All authors approved the final version for publication and agree to be accountable for all aspects of the work, ensuring accuracy and integrity. Funding This research was funded by the Cooperation Partnerships program of the European Commission (101050490-ERASMUS-SPORT-2021-SCP). Data availability All data generated and analyzed in this study are available from the corresponding author upon reasonable request. Declarations Ethics approval and consent to participate The protocol was approved by the Research Ethics Committee of the University of Vic – Central University of Catalonia (UVic-UCC) (278/2023), the Committee for Ethics, Faculty of Sport of the University of Ljubljana (033 − 2/2025-3) and the Research Ethics Committee of the Technological University of the Shannon (2024/03/06). The Clinical Expertise Centre of the National Institute for Public Health and the Environment, Ministry of Health, Welfare and Sport, Netherlands, determined that given the nature of the project, approval by an ethical research committee was not necessary. All procedures performed in the present study were in accordance with the ethical standards of the institutional and/or national research committees and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. Eligible participants completed the informed consent digitally prior to baseline measures. 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 Associated Data Supplementary Materials Supplementary Material 1. Data Availability Statement All data generated and analyzed in this study are available from the corresponding author upon reasonable request.