Hypertension is a strong risk factor for cardiovascular diseases such as stroke (1) and myocardial infarction (1). On a global scale, hypertension is responsible for 14% of total deaths and 143 million disability-associated life-years (2). The worldwide prevalence of diagnosed hypertension has been projected to escalate from 21% in 2015 (2) to 60% in 2025 (3).
Hypertension is strongly linked with lifestyle (4). More time spent in moderate-to-vigorous physical activity (MVPA) (5), light physical activity (LPA) (6), and sleep (7) and less time in sedentary behavior (8) has been reported to be associated with lower blood pressure (BP), but findings have been mixed (7,8).
Movement behaviors (e.g., LPA, MVPA, sedentary behaviors and sleep) constitute mutually exclusive and exhaustive components of the complete day (i.e., 24 h). Therefore, it is not possible to increase time spent in one movement behavior without a compensatory decrease in the remaining movement behaviors. Consequently, it does not make sense to consider the health impact of changes in one behavior isolated from the remaining behaviors. For example, the health effect of increasing MVPA may vary depending on whether it is sedentary time or sleep that is being displaced. Daily movement behavior data are codependent, or compositional, with important implications for statistical analysis and interpretation. However, this has been overlooked in previous research. Thus, to develop and refine daily movement behaviors recommendations, studies investigating the association between the daily composition of movement behaviors and health outcomes including BP are needed (9,10).
To date, two studies (9,11) have explored the association between movement behavior composition and BP among adults. Both studies found a significant relationship between the movement behavior composition and BP. Specifically, more time spent asleep at the expense of the remaining movement behaviors was favorably associated with diastolic BP (DBP) in a general population (9). The other study found beneficial effects on BP when sitting time was replaced by the remaining behaviors, especially stepping, among desk-based workers (11). However, it is unknown whether the same relationships exist among different populations, such as white and blue collar workers.
The purpose of this study was to investigate the association between the daily composition of movement behaviors (sedentary, LPA, MVPA, and time in bed) and BP among workers. It was hypothesized that the daily composition of movement behaviors would be associated with BP.
Study design and population
This study is part of the DPHACTO cohort (12), comprising workers from 15 companies engaged in cleaning, manufacturing, and transport sectors recruited between December 2011 and March 2013. Workers were excluded if they were pregnant, had a fever, or band aid allergy. The details of the data collection are reported elsewhere (12). Our study used baseline data collected via questionnaires, onsite visits, and accelerometry between spring 2012 and spring 2013.
The participants provided written informed consent before participation. The present study was conducted according to the Helsinki declaration and approved by the Danish data protection agency and local ethics committee (The Capital Region of Denmark, H-2-2012-011).
Measurement of movement behaviors
Workers wore an Actigraph accelerometer (GT3X+; Actigraph LLC, USA) directly attached to the skin on the right thigh (13) for four consecutive days (4 × 24 h), including at least two working days. Workers were asked to complete a short paper-based diary, registering their working hours, time in bed (going to and getting out of bed), nonwear time, and time of reference measurement (i.e., standing in an upright position for 15 s) on the measurement days. Further details are provided elsewhere (14).
A customized MATLAB program, Acti4, was used to identify periods spent sedentary and physical activities (i.e., standing, walking slow and walking fast, running, cycling, and stair climbing). A figure depicting how Acti4 recognizes time distribution of different activities subsequently classified into different levels of activities is shown in the Supplemental Digital Content (see Figure, Supplemental Digital Content 1, Diagrammatical representation of Acti4 software recognizing various movement behaviors and showing the time-usage distribution for a worker during a working interval from 6:30 AM to 3:30 PM, http://links.lww.com/MSS/B298). The Acti4 program has a high sensitivity and specificity in detecting these physical activities during semistandardized conditions (13). Time spent sedentary and in physical activities was determined based on definitions detailed elsewhere (13,14). Time spent standing and slow walking was merged to calculate LPA, whereas time spent fast walking, running, cycling and stair climbing was merged to calculate MVPA. Time in bed was based on information from self-report diary and visual inspection of the accelerometer data using Acti4.
All nonworking days and nonwear periods of accelerometers were excluded according to previously defined criteria (14). One working day consisted of 24 h starting at midnight and was considered valid if it comprised at least 10 h of wear time during waking hours (14). Time in bed as a proxy of sleep time was considered valid if it was at least of 4-h duration. Workers with measurements on at least 1 d including one valid working day and one valid time in bed period were included in the analysis.
The mean of sedentary time, LPA and MVPA, and the median of all valid time-in-bed periods on all valid days were calculated.
Measurement of systolic BP and DBP
Workers were asked to rest in a seated position for 5 min with their back supported, legs uncrossed, and arm supported. Using the Omron M6 Comfort on the right arm, three measurements of resting BP were taken interspaced with regular interval of 1 to 2 min. The average of the last two recordings was used in the analysis.
Measurements of confounders
Potential confounders were chosen a priori based on previous knowledge (15,16). Age and sex of the workers were determined using unique Danish civil registration number and workers were categorized into two groups; ≤45 yr and >45 yr. Using height (Seca, model 213) and weight (Tanita model BC418 MA) measurements, body mass index (BMI) was calculated and categorized into two groups, <30 kg·m−2 and ≥30 kg·m−2. Time spent lifting/carrying at work was determined using a single item with six response categories ranging from “almost all the time” to “never” (14). Prescribed medication intake due to hypertension, depression, and heart disease was determined using four items “Have you in the last three months been taking prescription medication? If yes, is it antihypertensive, heart medication, or anti depressives? With yes and no as response options.” Information about the job sector of the workers was obtained from the payroll.
Analyses were conducted in Rstudio (version 0.99.893-©2009-2016) using the Compositions (17) and robCompositions (18) packages. The significance level was set at P < 0.05. The explanatory variable, the daily composition of movement behaviors is compositional. The sample space for compositional data is the Simplex, defined by Aitchison geometry (19). Standard statistical methods, which rely on Euclidean operations such as addition and multiplication, are incompatible with the Simplex. However, the expression of compositional data as isometric log-ratio (ilr) coordinate systems imposes a Euclidean space structure to the Simplex, and allows standard statistical methods to be applied (19). Therefore, the movement behavior composition was expressed as a set of ilr coordinates before analysis.
The daily composition of time spent sedentary and in LPA, MVPA, and bed was described by the compositional mean, which is the geometric mean of each respective movement behavior, linearly adjusted so that total daily time spent in movement behaviors summed up to 1440 (i.e., 24 h·d−1) allowing interpretation in minutes per day.
Multiple linear regression analysis was used with systolic BP (SBP) or DBP regressed against the daily composition of time spent in movement behaviors (expressed as ilr coordinates). Worker’s age, sex, BMI, lift/carry duration, medication intake, and job sector were included as confounders in the analyses. All confounders were treated as categorical variables except the lift/carry duration which was treated as continuous variable in the model. The significance of each ilr coordinate system was assessed using Wald χ2 ANOVA tables with type II tests.
The R function “PivotCoord” was used to express the daily time composition as ilr coordinates. This function creates ilr coordinate systems where the first log-ratio has the first compositional part as its numerator, and the geometric mean of all other compositional parts as its denominator. By rotating the sequence of the compositional parts so each part was iteratively considered as the first compositional part, four ilr coordinate systems were constructed, each with different behavior (sedentary, LPA, MVPA, or time in bed) as the numerator of the first ilr coordinate. Four different multiple linear models (one for each ilr coordinate system) were executed for each outcome (SBP and DBP). The beta coefficients, standard error of beta, t-statistic and P values were reported for the first ilr coordinate of each coordinate system. The values reported must be interpreted with direct reference to the corresponding ilr coordinate. However, the numerical value associated with the beta coefficients cannot be interpreted in absolute terms, as it corresponds to a relative construct (20). Consequently, the beta coefficients must be interpreted from a reference composition. For this analysis, the sample mean composition was used as the reference composition.
To interpret the beta coefficients, the ilr multiple linear regression model was used to predict the change in BP when the compositional part in the numerator of the first ilr was increased from its mean value by a fixed duration of time, and the geometric mean of the remaining parts was decreased accordingly, to maintain a daily total of 1440 min (20; one-to-remaining reallocations). Durations reallocated to and from the movement behavior in the numerator were incrementally increased, whereas the remaining parts in the denominator were decreased accordingly. The changes in predicted BP associated with the reallocated compositions were plotted to display the relationship between BP and movement behavior composition. To supplement the findings, similar procedures were performed based on pair-wise reallocations where time was reallocated from one behavior to another, keeping the remaining behaviors constant (9).
The flow of workers is shown in Figure 1. Of the 2107 invited workers, 1119 consented to participate. Of them, 827 attached the accelerometer and had at least one valid workday measured. On average, workers wore the accelerometer for 22.95 ± 1.39 h·d−1.
Descriptive statistics of SBP, DBP, movement behavior composition, and confounders are detailed in Table 1.
The composition of daily movement behaviors was a significant predictor of SBP (f = 2.84, P = 0.04), but not DBP (f = 0.48, P = 0.69).
The analysis of the association between BP and each single movement behavior “at the expense of the remaining movement behaviors within a day” showed that SBP was positively associated with sedentary time (P = 0.01) and negatively associated with time in bed (P = 0.047; see Figure 2 and Table in Supplemental Digital Content 2, Results of multiple linear regression investigating the cross-sectional association between daily composition of time spent in movement behaviors (expressed as ilr coordinates) and SBP among 827 workers, http://links.lww.com/MSS/B299). Similar associations were observed for DBP, but without reaching statistical significance.
Reallocating 120 min of sedentary time equally to other movement behaviors was associated with lower SBP by ≈2 mm Hg (Fig. 2). Higher SBP (+ ≈ 2 mm Hg) was associated with the reallocation of 120 min from time in bed to other movement behaviors. Nonsignificant changes in SBP were observed when LPA and MVPA were reallocated to other movement behaviors. For example, lower SBP (by 0.3 mm Hg) was associated with the reallocation of 15 min to MVPA from the remaining behaviors.
Workers’ movement behavior composition was associated with their SBP but not DBP. Specifically, SBP was deleteriously associated with the reallocation of time to sedentary behavior and LPA and beneficially associated with reallocation of time to MVPA and time in bed. However, these associations were only statistically significant for time spent sedentary and in bed.
We found that higher sedentary time (at the expense of remaining movement behaviors) was deleteriously associated with SBP. Previous reviews on the associations of sedentary time and BP have indicated mixed findings (8,21,22). Discrepancies between previous research and our findings may be explained by a number of factors: (a) previous findings are largely based on self-reported sedentary time (8) rather than objective measures as in this study; (b) previous studies have used counts per minute–based thresholds to determine sedentary postures and physical activities (6), which have shown a poor ability to differentiate between various sedentary behaviors and physical activities (23); and (c) previous studies have not included all daily movement behaviors in the analysis, as in this study. Of the two previous Compositional Data Analysis (CoDA) studies investigating BP, one concurs with our findings (11) but the other does not (9). The differences may be explained by the type of study population and the use of counts per minute–based thresholds to determine movement behaviors.
We observed that spending longer time in bed (at the expense of the remaining movement behaviors) was beneficially associated with BP, consistent with previous research (9). Several studies have found a “U-shaped” relationship between sleep duration and BP (24), suggesting that both very short and very long sleep durations are deleteriously associated. However, the statistical models fit indices in the present study did not indicate any nonlinear relationship between time in bed and BP.
More LPA (at the expense of the remaining movement behaviors) was deleteriously associated with BP, although not reaching statistical significance. Only recently, researchers have started to investigate the supposedly independent effects of LPA on BP. Some studies have found beneficial associations between LPA and indicators of BP (6,25), but not all (9,26). However none of these studies were on working populations largely dominated by blue-collar occupations. Blue-collar workers are highly exposed to LPA at work (particularly standing and slow walking) (27). Notably, previous studies have reported that physical activity at work increases the risk of high ischemic heart diseases, elevated heart rate, and BP (28,29), termed the “health paradox of physical activity.” Thus, our research should be followed up by paying attention on the associations between domain-specific movement behaviors and BP.
We observed a beneficial association between higher time in MVPA (at the expense of the remaining movement behaviors) and SBP among workers. Although not reaching statistical significance, this result is in line with the well-documented beneficial effects of MVPA on BP (30). One previous study exploring the effect of changing the movement composition reported improved BP when sitting was replaced with stepping; however, this study used stepping as a measure of activity instead of MVPA and was conducted among desk-based workers only (11).
Overall, we found the movement behavior composition to be associated with SBP, but not DBP, which is similar to one previous study investigating the effects of replacement of sedentary time with standing or walking activities on cardiovascular health (11). Previous studies have indicated that it may require higher volumes and intensity of exercise to show changes in DBP (31,32).
STRENGTHS, LIMITATIONS, AND CLINICAL RELEVANCE OF THE STUDY
This study has a number of strengths. The novel CoDA approach enables all daily movement behaviors to be included in the same analytical model, avoiding the issue of residual confounding due to the omission of intrinsically codependent movement behaviors (10). In addition, unlike traditional multivariate statistical models, the CoDA approach accommodates the relative nature of daily movement behavior data (9,19,20).
Another strength of this study was the technical measurement of movement behaviors, avoiding the potential pitfalls of social desirability bias or recall error that may be associated with self-report. Accelerometer wear-time compliance in this study was close to 24 h·d−1, possibly because the devices were waterproof and therefore did not require removal for water-based activities as has been the case with previous accelerometer protocols (33,34).
The limitations of this study must be considered. First, the cross-sectional design precludes any inference of cause-effect between movement behaviors and BP. Second, we did not distinguish between wakeful time and nonwakeful time in bed. It is possible that the relationship between BP and time in bed may differ, depending on whether an individual is awake or asleep.
The key implication of the study is that modification of movement behaviors may be beneficial to BP, although we cannot be certain of the direction of this relationship due to the cross-sectional nature of the data. Public health campaigns have advocated swapping strategies (e.g., between sitting and standing) to modify movement behaviors (35). We used the CoDA models to estimate the difference in BP associated with the pairwise reallocation of fixed durations of time from one movement behavior to another (See Figure, Supplemental Digital Content 3, Compositional isotemporal substitution models showing association between pairwise change in daily-movement behaviors and SBP, http://links.lww.com/MSS/B300). A meta-analysis (36) suggests that a change of ≈2 mm Hg in SBP is clinically relevant for cardiovascular health. We estimated that this difference in SBP is associated with reallocating 120 min from sedentary or LPA to time in bed, or by reallocating 60 min of sedentary/LPA time to MVPA, keeping the other behaviors constant. A previous meta-analysis on workplace interventions suggests a feasible reduction of sedentary time within 8 h work time to be 40 to 90 min (37). Thus, a 120-min reduction in sedentary behavior seems to be a realistic intervention goal over a whole 24-h day. However, a 15- to 30-min increase in MVPA may be more achievable than a 60-min increase (38,39). In our study, a 15- or 30-min increase in MVPA was associated with −0.3 or −0.6 mm Hg SBP.
Our results can be used to tailor intervention strategies to individual preference. For example, one individual may choose to replace sedentary time (e.g., driving to gym) with MVPA (e.g., running to gym). In contrary, a tired blue-collar worker may choose to replace sedentary time (e.g., watching TV) with time in bed (e.g., going to bed early). However, reallocating even less than 120 min may also be expected to decrease BP by ≈2 mm Hg due to the simultaneous adaptive physiological effects caused by physical activity interventions (40). We recommend future studies to investigate similar strategies for improving health using a prospective design.
Daily composition of time spent in various movement behaviors was associated with SBP among workers. Time spent sedentary (at the expense of other movement behaviors) was deleteriously associated, whereas time spent in bed (at the expense of other movement behaviors) was beneficially associated with SBP. How daily time is distributed among movement behaviors seems to be important for regulating BP. Thus, healthful time reallocations should be further investigated for inclusion in clinical practice and 24-h activity guidelines.
The DPHACTO cohort is partly supported by a grant from the Danish government (satspulje). This study was financially supported by the Danish Work Environment Research Fund (journal number 20150017496/4) and by the Australian Government Research Training Program Scholarship. We would like to thank the DPHACTO research group at the National Research Centre for the Working Environment, Copenhagen, Denmark.
Authors declare none conflicts of interest. Authors also declare that the results of the study are presented clearly, honestly, and without fabrication, falsification, or inappropriate data manipulation. The results of the present study do not constitute endorsement by ACSM.
1. Gaziano TA, Bitton A, Anand S, Weinstein MC. The global cost of nonoptimal blood pressure. J Hypertens
2. Forouzanfar MH, Liu P, Roth GA, et al. Global burden of hypertension and systolic blood pressure
of at least 110 to 115 mm Hg, 1990-2015. JAMA
3. Kearney PM, Whelton M, Reynolds K, Muntner P, Whelton PK, He J. Global burden of hypertension: analysis of worldwide data. Lancet
4. Forman JP, Stampfer MJ, Curhan GC. Diet and lifestyle risk factors associated with incident hypertension in women. JAMA
5. Warburton DE, Charlesworth S, Ivey A, Nettlefold L, Bredin SS. A systematic review of the evidence for Canada’s Physical Activity Guidelines for Adults. Int J Behav Nutr Phys Act
6. Hamer M, Bruwer EJ, de Ridder JH, et al. The association between seven-day objectively measured habitual physical activity and 24 h ambulatory blood pressure: the SABPA study. J Hum Hypertens
7. Guo X, Zheng L, Wang J, et al. Epidemiological evidence for the link between sleep duration and high blood pressure: a systematic review and meta-analysis. Sleep Med
8. Lee PH, Wong FK. The association between time spent in sedentary behaviors and blood pressure: a systematic review and meta-analysis. Sports Med
9. Chastin SF, Palarea-Albaladejo J, Dontje ML, Skelton DA. Combined effects of time spent in physical activity, sedentary behaviors and sleep on obesity and cardio-metabolic health markers: a novel compositional data analysis approach. PLoS One
10. Pedišić Ž, Dumuid D, Olds T. Integrating sleep, sedentary behaviour, and physical activity research in the emerging field of time-use epidemiology: definitions, concepts, statistical methods, theoretical framework, and future directions. Kinesiology
11. Winkler EA, Chastin S, Eakin EG, et al. Cardiometabolic impact of changing sitting, standing, and stepping in the workplace. Med Sci Sports Exerc
12. Jørgensen MB, Korshøj M, Lagersted-Olsen J, et al. Physical activities at work and risk of musculoskeletal pain and its consequences: protocol for a study with objective field measures among blue-collar workers. BMC Musculoskelet Disord
. 2013;14:213. 2474-14-213.
13. Skotte J, Korshøj M, Kristiansen J, Hanisch C, Holtermann A. Detection of physical activity types using triaxial accelerometers. J Phys Act Health
14. Gupta N, Stordal CC, Hallman D, Korshøj M, Gomes CI, Holtermann A. Is objectively measured sitting time associated with low back pain? A cross-sectional investigation in the NOMAD study. PLoS One
15. Clays E, De Bacquer D, Janssens H, et al. The association between leisure time physical activity and coronary heart disease among men with different physical work demands: a prospective cohort study. Eur J Epidemiol
16. Law MR, Morris JK, Wald NJ. Use of blood pressure lowering drugs in the prevention of cardiovascular disease: meta-analysis of 147 randomised trials in the context of expectations from prospective epidemiological studies. BMJ
17. Boogaart KG, Tolosana R, Bren M. Compositions: compositional Data Analysis. R (version ≥ 2.2.0)
18. Templ M, Hron K, Filzmoser P. robCompositions: an R-package for robust statistical analysis of compositional data. In: Pawlowsky-Glahn V, Buccianti A editors. Compositional Data Analysis. Theory and Applications
. Chichester (UK) John Wiley & Sons; 2011. pp. 341–55.
19. Aitchison J. The Statistical Analysis of Compositional Data. J R Stat Soc B Methodol
20. Dumuid D, Stanford TE, Martin-Fernández JA, et al. Compositional data analysis for physical activity, sedentary time and sleep research. Stat Methods Med Res
21. Proper KI, Singh AS, van Mechelen W, Chinapaw MJ. Sedentary behaviors and health outcomes among adults: a systematic review of prospective studies. Am J Prev Med
22. de Rezende LF, Rodrigues Lopes M, Rey-López JP, Matsudo VK, Luiz Odo C. Sedentary behavior and health outcomes: an overview of systematic reviews. PLoS One
23. Hart TL, Ainsworth BE, Tudor-Locke C. Objective and subjective measures of sedentary behavior and physical activity. Med Sci Sports Exerc
24. Wang Y, Mei H, Jiang YR, et al. Relationship between duration of sleep and hypertension in adults: a meta-analysis. J Clin Sleep Med
25. Gerage AM, Benedetti TR, Farah BQ, et al. Sedentary behavior and light physical activity are associated with brachial and central blood pressure in hypertensive patients. PLoS One
26. Gomez-Marcos MA, Recio-Rodriguez JI, Patino-Alonso MC, et al. Relationship between objectively measured physical activity and vascular structure and function in adults. Atherosclerosis
27. Hallman DM, Gupta N, Heiden M, et al. Is prolonged sitting at work associated with the time course of neck-shoulder pain? A prospective study in Danish blue-collar workers. BMJ Open
28. Korshøj M, Lidegaard M, Kittel F, et al. The relation of ambulatory heart rate with all-cause mortality among middle-aged men: a prospective cohort study. PLoS One
29. Allesøe K, Holtermann A, Aadahl M, Thomsen JF, Hundrup YA, Søgaard K. High occupational physical activity and risk of ischaemic heart disease in women: the interplay with physical activity during leisure time. Eur J Prev Cardiol
30. Diaz KM, Shimbo D. Physical activity and the prevention of hypertension. Curr Hypertens Rep
31. Fagard RH. Exercise is good for your blood pressure: effects of endurance training and resistance training. Clin Exp Pharmacol Physiol
32. Whelton SP, Chin A, Xin X, He J. Effect of aerobic exercise on blood pressure: a meta-analysis of randomized, controlled trials. Ann Intern Med
33. Vanhelst J, Béghin L, Drumez E, Coopman S, Gottrand F. Awareness of wearing an accelerometer does not affect physical activity in youth. BMC Med Res Methodol
34. Vanhelst J, Mikulovic J, Bui-Xuan G, et al. Comparison of two ActiGraph accelerometer generations in the assessment of physical activity in free living conditions. BMC Res Notes
35. O’Hara BJ, Grunseit A, Phongsavan P, Bellew W, Briggs M, Bauman AE. Impact of the swap it, don’t stop it Australian National Mass Media campaign on promoting small changes to lifestyle behaviors. J Health Commun
36. Lewington S, Clarke R, Qizilbach N, Peto R, Collins R. Age-specific relevance of usual blood pressure to vascular mortality: a meta-analysis of individual data for one million adults in 61 prospective studies. Lancet
37. Chu AH, Ng SH, Tan CS, Win AM, Koh D, Muller-Riemenschneider F. A systematic review and meta-analysis of workplace intervention strategies to reduce sedentary time in white-collar workers. Obes Rev
38. Boekhout JM, Berendsen BAJ, Peels DA, Bolman CAW, Lechner L. Evaluation of a computer-tailored healthy ageing intervention to promote physical activity among single older adults with a chronic disease. Int J Environ Res Public Health
39. Overgaard K, Nannerup K, Lunen MKB, Maindal HT, Larsen RG. Exercise more or sit less? A randomized trial assessing the feasibility of two advice-based interventions in obese inactive adults. J Sci Med Sport
40. Whelton PK, He J, Appel LJ, et al. Primary prevention of hypertension: clinical and public health advisory from The National High Blood Pressure Education Program. JAMA