The Effect of Sitting Duration on Peripheral Blood Pressure Responses to Prolonged Sitting, With and Without Interruption: A Systematic Review and Meta-Analysis

Abstract

Background

A previous meta-analysis reported that: (i) an acute bout of prolonged uninterrupted sitting induces a significant increase in peripheral blood pressure (BP) and (ii) the increase in BP can be offset by interrupting the sitting bout with light aerobic activities such as walking. However, the temporal association between prolonged uninterrupted sitting and BP was not determined. A better understanding of temporality, for example, how long it takes BP to increase, will assist in prescribing sitting interruption strategies.

Objectives

We aimed to determine: (1) the temporal association between the duration of uninterrupted sitting and BP and (2) whether regular sitting interruptions moderate the association between uninterrupted sitting and BP.

Data Sources

Electronic databases (PubMed, Web of Science, SPORTDiscus) were searched from inception to July 2022. Reference lists of eligible studies and relevant reviews were also screened.

Study Selection

Inclusion criteria for objective (1) were: (i) participants aged ≥ 18 years; (ii) a prolonged sitting bout ≥ 1 h; and (iii) peripheral BP measurements (systolic BP, diastolic BP, and/or mean arterial pressure) at more than two timepoints during the sitting bout. Additional criteria for objective (2) were: (i) the sitting interruption strategy was implemented during the sitting bout (i.e., not prior to engaging in sitting) and (ii) the study included a control (uninterrupted sitting) condition or group.

Appraisal and Synthesis Methods

There were 1555 articles identified, of which 33 met inclusion criteria for objective (1). Of those articles, 20 met inclusion criteria for objective (2). To investigate the effect of sitting duration on the BP response, unstandardized b coefficients (mmHg/h) and 95% confidence intervals (CIs) were calculated using a three-level mixed-effect meta-regression.

Results

Increased sitting duration was positively associated with systolic BP (b = 0.42 mmHg/h, 95% CI 0.18–0.60), diastolic BP (b = 0.24 mmHg/h, 95% CI 0.06–0.42), and mean arterial pressure (b = 0.66 mmHg/h, 95% CI 0.36–0.90). In trials where sitting was interrupted, there was a significant decrease in systolic BP (b = − 0.24 mmHg/h, 95% CI − 0.42 to − 0.06) and diastolic BP (b = − 0.24 mmHg/h, 95% CI − 0.42 to − 0.12), and a non-significant change in mean arterial pressure (p = 0.69).

Conclusions

Increased uninterrupted sitting duration results in greater increases in BP; however, regularly interrupting sitting may offset negative effects.

1. Introduction

Increased time spent in sedentary behaviors (i.e., waking behaviors in a seated, reclined, or lying posture with a very low energy expenditure) [1] is associated with an increased risk of cardiovascular disease and all-cause mortality [2, 3]. This is particularly concerning as sedentary behaviors, such as prolonged sitting, comprise a large portion of many people’s waking hours worldwide [4–13]. Experimental work and subsequent reviews have shown that acute bouts of sitting negatively impact several cardiometabolic outcomes, including vascular function, triglyceride and glucose metabolism, and peripheral blood pressure (BP) [14–16]. It is hypothesized that repeated and prolonged exposure to this negative phenotype may contribute to an increased cardiovascular burden and, in turn, an increased risk of cardiovascular disease [17].

As a result of emerging evidence related to sedentary behavior and cardiovascular disease, the World Health Organization, amongst others, now recommends reducing sedentary time and replacing that time with light-intensity physical activity [18]. However, in contrast to the well-defined physical activity guidelines, sedentary behavior guidelines remain vague and non-specific, providing little guidance on the total ‘safe’ sitting duration per bout or per day. Previous work has identified the importance of regularly interrupting bouts of prolonged sitting to maintain BP as well as other cardiometabolic markers [14–16]. However, to better support the development of efficacious sitting interruption strategies and specific sedentary behavior guidelines, the temporality of BP changes during sitting needs to be further understood. Further understanding of the temporal BP changes during uninterrupted sitting allows researchers to better target when sitting should be interrupted in future trials and helps clinicians and guideline developers understand the negative time course of a BP change during uninterrupted sitting.

1.1. Objectives

The objectives of this meta-analysis are two-fold: (1) to identify the temporal association between the duration of uninterrupted sitting and BP in adults and (2) to determine whether regular sitting interruptions moderate the association between uninterrupted sitting and BP.

2. Methods

This meta-analysis was carried out in accordance with PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines [19]. Details of the protocol for this systematic review were registered on PROSPERO and can be accessed at https://www.crd.york.ac.uk/prospero/display_record.php?ID=CRD42020206177.

2.1. Data Sources and Searches

Electronic databases (PubMed, Web of Science, SPORTDiscus) were searched by two authors (NA, JP) independently. Owing to the wide implementation of peripheral BP as both a primary and secondary outcome measure, the search terms used reflected the need to identify all relevant studies. The reference lists of all identified trials and relevant reviews or editorials were also examined. The search was limited to English language studies published before July 2022. Full details of the search strategy are presented in the Electronic Supplementary Material (ESM).

2.2. Article Selection

For the purpose of this meta-analysis, the term ‘article’ is used synonymously with ‘study’. The term ‘trial’ is the unit included in the meta-analysis; this may include different groups within an individual article that can be easily separated (such as a group of men and a group of women) or may represent different interruption trials within one study. A given article may have resulted in more than one eligible ‘trial’ if the article included more than one distinguishable intervention group, or if the data were separated by sex. Initially, article titles and abstracts were screened for relevance. The full text of potentially eligible articles was obtained to review their eligibility for inclusion. For objective (1), the following criteria were used to select trials for inclusion in the review: (i) participants aged ≥ 18 years, healthy or with clinical conditions, but free from autonomic or neuromuscular diseases or dysfunction (ii) a prolonged sitting bout ≥ 1 h, and (iii) peripheral BP measurements at more than two timepoints during the sitting bout. To address the second objective of this review, additional inclusion criteria were applied, including studies with: (i) any physical activity-related sitting interruption strategy that was implemented during the sitting bout (not prior to engaging in sitting) and (ii) a control (uninterrupted sitting) condition or group.

In trials with multiple treatment arms and a single control group, the sample size of the control group was divided by the number of treatment groups to avoid over-inflation of the sample size in accordance with Cochrane recommendations [20]. Repeated publications for the same studies were excluded. Two researchers (NA, JP) completed the study selection independently. Conflicts between reviewers were resolved by discussion, or when an agreement was not reached, by a third author (CP).

2.3. Data Extraction and Bias Assessment

Data extracted for each eligible trial included bibliographic information (i.e., author, publication year), sample characteristics (i.e., age, sex, body mass index [BMI], health status), details of intervention(s), and systolic BP (SBP), diastolic BP (DBP), and mean arterial pressure (MAP) values (mean and standard deviation) at every timepoint from the start to the end of each individual sitting bout. If data were presented as figures, the study authors were contacted and asked to provide data. A total of 22 authors were contacted for data. In the event of not receiving the requested data, values were extracted using ImageJ image analysis software [21]. As this method has the potential for decreased precision and additional error, members of the research team (NA, JP) completed tests of reliability and validity on existing known data sets in line with the methods described previously [14, 22]. Comparison of extracted values to known values showed both excellent validity and reliability for both researchers (NA, intraclass correlation coefficient = 0.99, r = 0.99, and JP, intraclass correlation coefficient = 0.99, r = 0.99). As such, this approach is likely to have introduced negligible measurement error. Five total records had data extracted using ImageJ [23–27]. If authors did not respond and images were not available, data were excluded from the final analysis. Three total records were not included as authors did not respond to requests for data and images were not available. All data were entered into a spreadsheet, and the aggregation and calculation of final results were conducted by one author (CP).

Risk of bias within studies was assessed using the Cochrane Risk-of-Bias 2 assessment (range; low risk, some concerns, high risk), which includes items related to randomization, blinding, and description of dropout/withdrawals [28]. Because it is difficult (if not impossible) to blind participants to an exercise intervention, we considered the blinding of the operator to the outcome assessment as a quality criterion. The ‘intention-to-treat’ model was assessed as the review aim for domain 2. Data extraction, quality assessment, and scrutiny of the sitting interruptions were independently completed by two members of the research team (NA, JP).

2.4. Data Analysis

Data were analyzed using the ‘metafor’ package (version 3.8.1) in R (version 4.2.1) [29, 30]. Outcome measures were expressed as weighted mean difference (mmHg). As the primary aim of this meta-analysis was to determine the impact of sitting duration on peripheral BP, it was necessary to extract data at multiple timepoints from the same trial. To account for the dependence between data points extracted from the same trial, a three-level random-effects model with restricted maximum likelihood estimation was chosen. A three-level model can account for sampling variance, study-level variance, and between-study variance and thus is able to deal with dependency between related data points. Once data were compiled using the three-level model, a mixed-effect linear meta-regression was used to explore the effect of continuous variables, principally the effect of time. To test non-linearity, restricted cubic splines with 5 knots at equally spaced quantiles were fitted to each model and likelihood ratio tests were used to determine model fit [31].

Unstandardized beta (b) values with corresponding 95% confidence intervals (CIs) were the preferred outcome estimate for linear meta-regression. After meta-regression, the robustness of the pooled effect was assessed using Cook’s distance and studentized residuals at the study level to identify potential outliers and/or influential trials [32]. Heterogeneity at different levels was assessed using I² where < 25%, 25–75%, and > 75% represent small, moderate, and considerable, respectively [33]. In the presence of statistical heterogeneity, a stratified analysis of uninterrupted and interrupted sitting data was performed, and the robustness of pooled results was again assessed via Cook’s distance and studentized residuals. In the continued presence of statistically significant heterogeneity, an a priori determined moderator analysis was performed. A previous meta-analysis has shown that the BP assessment method (i.e., continuous vs discrete) may affect the observed effect of sitting on BP outcomes [14]. Consequently, sensitivity analyses were performed whereby trials utilizing continuous measures were excluded from the overall model to test the robustness of the effect. Separate analyses were performed for SBP, DBP, and MAP.

3. Results

3.1. Literature Search and Trial Selection

The literature search strategy and outcomes are outlined in Fig. 1. Initial database searches identified 1555 potentially eligible articles in July of 2022. A total of 351 articles were removed as duplicates, and the remaining 1204 abstracts were subjected to title and abstract screening. A total of 1135 articles were removed as they did not meet all inclusion criteria, and 69 articles were processed through full-text screening. Of these, 36 articles did not meet the inclusion criteria and 33 articles were identified for inclusion. The final analyses for objective (1) included 37 trials (20 articles) for MAP, and 24 trials (21 articles) for SBP and DBP. The final analyses for objective (2) included nine trials (seven articles) for MAP and 28 trials (18 articles) for SBP and DBP.

Fig. 1.

PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) flow diagram of study selection [19]. BP blood pressure

3.2. Description of the Included Trials

3.2.1. Trial Setting and Characteristics of Included Studies

Included trial characteristics are summarized in Table 1. The number of participants in each article ranged from 9 [34, 35] to 67 [25]. The mean age of the participants ranged from 19 [36] to 67 [25] years. Body mass index was not reported in two studies, and in the remaining 28 studies, the mean BMI was between 20 [36] and 33 kg/m² [37]. Of the 30 studies included, 13 were conducted in the USA, five in the UK, four in Australia, two in Canada, two in Japan, and one in New Zealand, Norway, Brazil, and South Korea. Most studies were published within the previous 10 years, with one study published in 2008 [38] and one published in 1994 [39]. Sitting durations ranged from 1.5 [40] to 10 [41] hours with a median sitting duration of 3 h.

Table 1.

Characteristics of the included studies

Study	Year	Sample [n (F); age (y, mean (SD or range)]	BMI [kg/m2, mean (SD)]	Participant disease statusb	Reported BP outcomes	Method of BP assessment	Sitting duration (h)	Interruption strategy
Bailey and Locke [26]	2015	10 (3); 24 (3)	26.5 (4.3)	Healthy	SBP, DBP	Auto oscillometric	5	Aerobic, standing
Ballard et al. [53]	2017	11 (0); 21.2 (1.9)	24.7 (1.0)	Healthy	MAP	Auto oscillometric	3	None
Barone Gibbs et al. [42]	2017	25 (9); 42 (12)	31.9 (5.0)	Pre-to-stage 1 hypertension	SBP, DBP, MAP	Auto oscillometric	7	Standing
Carter et al. [40]	2017	10 (4); 27.3 (8.1)	NR	Healthy	MAP	Auto oscillometric	1.5	SRA
Carter et al. [51]	2019	15 (5); 35.8 (10.2)	25.5 (3.2)	Healthy	MAP	Auto oscillometric	4	Aerobic
Champion et al. [23]	2018	24 (12); 35.8 (14.7)a	25.7 (4.8)a	Healthy, sedentary > 7 h/day	SBP, DBP	Auto oscillometric	6.5	Aerobic
Cho et al. [52]	2020	12 (5); 23.5 (2.9)	23.4 (2.7)	Healthy	SBP, DBP	Auto oscillometric	4	Aerobic
Credeur et al. [67]	2019	20 (7); 26 (7)	30.0 (7.0)	Healthy	MAP	Auto oscillometric	3	None
Dempsey et al. [37]	2016	24 (10); 62 (6)	33.0 (3.4)	T2D, BMI 25–40 kg/m2	SBP, DBP	Auto oscillometric	8	Aerobic, SRA
Dobashi et al. [34]	2021	9 (0); 20.6 (0.7)	23.8 (NR)	Healthy	MAP	Auto oscillometric	2	Standing
Evans et al. [ 46]	2019	20 (14); 21.7 (2.5)	25.5 (6.1)	Healthy, inactive	SBP, DBP, MAP	Auto oscillometric	3	Aerobic
Freire et al. [41]	2019	25 (15); 24.4 (3.8)	26.1 (3.4)	Healthy, inactive	SBP, DBP, MAP	Auto oscillometric	10	Aerobic
Garten et al. [59]	2019	20 (4); 24.5 (3.2)a	23 (6.3)a	Healthy	MAP	Auto oscillometric	3	None
Gotshall et al. [39]	1994	34 (17); 21.8 (9.4)	NR	Healthy	MAP	Auto oscillometric	2	None
Headid et al. [54]	2020	12 (6); 22.3 (2.0)	23.9 (3.0)	Healthy	SBP, DBP	Auto oscillometric	2.5	None
Kowalsky et al. [44]	2019	14 (12); 53.4 (9.5)	30.9 (4.8)	At increased CVD risk	SBP, DBP	Auto oscillometric	4	SRA
Kruse et al. [43]	2018	13 (3); 38 (3)	29.7 (2)	Inactive, overweight, obese	SBP, DBP	Auto oscillometric	4	Aerobic, standing
Larsen et al. [24]	2014	19 (8); 53.8 (4.8)	31.2 (3.9)	BMI 25–45 kg/m2	SBP, DBP, MAP	Auto oscillometric	5	Aerobic
Lunde et al. [47]	2012	11 (11); 44 (9.3)	30.9 (6.7)	Healthy	SBP, DBP	Auto oscillometric	2	Aerobic
Miyashita et al. [38]	2008	15 (0), 23.4 (3.1)	23.4 (2.3)	Healthy	SBP, DBP	Ascultation	7	Aerobic
Morishima et al. [55]	2017	15 (5); 26.7 (0.5)	25.6 (0.5)	Healthy	MAP	Auto oscillometric	3	None
Morishima et al. [35]	2020	9 (0); 21.2 (2)	22 (3)	Healthy	MAP	Auto oscillometric	3	None
Morishima et al. [65]	2020	19 (0); 20.4 (1.2)a	23.6 (1.9)a	Healthy	MAP	Auto oscillometric	3	None
Morishima et al. [36]	2022	11 (4); 18.6 (0.6)	19.8 (1.5)	Healthy	MAP	Auto oscillometric	3	None
O'Brien et al. [58]	2019	20 (10); 23.5 (2)a	25.4 (2.6)a	Healthy	SBP, DBP, MAP	Volume finger clamp	3	None
O'Brien et al. [56]	2020	18 (18); 23 (3)a	24.0 (2.9)a	Healthy	SBP, DBP, MAP	Volume finger clamp	3	None
Park et al. [48]	2022	14 (7); 23.6 (2.3)	24.4 (3.7)	Healthy	SBP, DBP, MAP	Auto oscillometric	2.5	Aerobic
Peddie et al. [49]	2021	18 (7); 23.5 (5.0)	23.7 (2.6)	Healthy	SBP, DBP	Auto oscillometric	6	Aerobic
Taylor et al. [45]	2020	24 (11); 61.5 (7.8)	32.6 (3.5)	T2D, inactive	SBP, DBP	Auto oscillometric	7	SRA
Vranish et al. [57]	2017	20 (12); 20.8 (0.4)a	24.7 (0.8)a	Healthy	MAP	Ascultation	3	None
Wennberg et al. [50]	2016	19 (9); 59.7 (8.1)	31.5 (4.7)	BMI 25–40 kg/m2, inactive	SBP, DBP	Auto oscillometric	7	Aerobic
Wheeler et al. [25]	2019	67 (35); 67 (7)	31.2 (4.1)	BMI 25–45 kg/m2	SBP, DBP	Auto oscillometric	8	Aerobic
Younger et al. [27]	2015	10 (4); 22.2 (1.3)	25 (2)	Healthy	MAP	Ascultation	5	None

BMI body mass index, BP blood pressure, CVD cardiovascular disease, DBP diastolic blood pressure, F female, h hour, kg kilograms, m meter, MAP mean arterial pressure, n participants, NR not reported, SBP systolic blood pressure, SD standard deviation, SRA simple resistance exercise, T2D type 2 diabetes, y year

^aMean and SD shown is a calculation from two separate groups within a trial (e.g. men and women)

^bParticipant characteristics are included as described in each study

3.2.2. Interruptions

A total of 20 studies observed a sitting interruption in addition to an uninterrupted prolonged sitting condition. Ten studies evaluated two different interruption strategies, resulting in 30 trials of sitting interruptions. Each interruption strategy was categorized as standing, aerobic, or simple resistance exercise. Aerobic interruption strategies ranged from calf raises to stair climbing, with the majority being walking interruptions, and simple resistance exercises were regularly bodyweight exercises that could be easily completed with little to no equipment. There were five standing trials, [26, 34, 42, 43] five simple resistance exercise trials, [37, 40, 44, 45] and 20 trials in 15 articles that tested sitting interruptions with different aerobic interventions [23–26, 37, 38, 41, 43, 46–52].

3.3. Methodological Quality Assessment

Methodological quality as assessed using the Cochrane Risk-of-Bias 2 Tool is summarized in the ESM. Studies were assessed as having a ‘low risk of bias,’ ‘some concerns,’ or a ‘high risk of bias’. Ninety-seven percent of the included studies had ‘some concerns’ of bias and the remaining 3% were deemed to have a ‘low risk of bias’. Only one study published a statistical plan prior to the protocol being conducted, the lack of which was the primary source of an increased risk of bias in all other studies [50]. Many studies did not have a sufficiently blinded randomization process, [23, 25–27, 34, 38–44, 46–48, 51–56] and some used a manual oscillometric measurement of BP without a blinded technician [27, 38, 39, 57].

3.4. Synthesis of Results

3.4.1. Systolic Blood Pressure

3.4.1.1. Uninterrupted Sitting

An analysis of 145 data points from 24 trials identified a significant effect of time (p < 0.001). Results from a linear meta-regression suggest that every hour of uninterrupted sitting is associated with a 0.42 mmHg (95% CI 0.18–0.60) increase in SBP. Examination of Cook’s distances and studentized residuals identified five trials as potential outliers; however, the removal of each in turn had no effect on the statistical significance of the observed outcome [25, 37, 48, 56, 58]. Removal of trials utilizing continuous BP measurements had no effect on the statistical significance of the observed outcome but did reduce the magnitude of effect (b = 0.23 mmHg, 95% CI 0.02–0.46, p = 0.04) [56, 58]. The estimated variance components of this model were τ²_{Level 2} = 0.60 and τ²_{Level 3} = 37.9. There was significant heterogeneity in the model (Q(143) = 1259.7, p < 0.01, I² = 90%) [I²_{Level 2} = 1.4%, I²_{Level 3} = 88.7%], justifying a further moderator analysis. Removal of each of the potentially influential trials in turn had no effect on the significance of the observed heterogeneity. A further moderator analysis showed no effect of age (p = 0.59) or BMI (p = 0.57). A further analysis identified a non-linear relationship between time and SBP responses to uninterrupted sitting (p_linearity = 0.049), with SBP increasing from 0 to 120 min before beginning to plateau (Fig. 2).

Fig. 2.

Restricted cubic spline of systolic blood pressure during uninterrupted sitting. Vertical lines, five knots at equally spaced quantiles; circles, mean values reported from individual trials at different sizes corresponding to sample size; shaded area, 95% confidence interval. mins minutes

3.4.1.2. Interrupted Sitting

An analysis of 175 data points from 28 trials identified a significant effect of time (p = 0.022). Results from a linear meta-regression suggest that every hour of interrupted sitting is associated with a 0.24 mmHg (95% CI 0.06–0.42) decrease in SBP. Examination of Cook’s distances and studentized residuals identified five trials as potential outliers [25, 37, 42]. Removal of two trials from one study individually resulted in a loss of significant time effect (b = 0.12, p = 0.203, and b = 0.12, p = 0.22) [37]. In the interests of rigor, the analysis was repeated after excluding both potentially influential trials, resulting in a loss of significant time effect (b = 0.0, p = 0.97). Removal of the final three trials had no effect on the statistical significance of the observed outcome [25, 42]. No included trials included continuous measures of BP, and thus a further sensitivity analysis was not conducted. The estimated variance components of the complete model without outliers were τ²_{Level 2} = 2.16 and τ²_{Level 3} = 25.4. There was significant heterogeneity in the model (Q(154) = 1520, p < 0.001, I² = 89.2%) [I²_{Level 2} = 6.9%, I²_{Level 3} = 81.4%]. Additionally, removal of each of the potentially influential trials in turn had no effect on the significance of the observed heterogeneity. Owing to the significant influence of the trials by Dempsey et al., and the lack of robustness of the overall effect, a further moderator analysis was deemed inappropriate [37]. A further analysis showed that the addition of a restricted cubic spline did not improve the model fit (p_linearity = 0.10, Fig. 3).

Fig. 3.

Restricted cubic spline of systolic blood pressure during interrupted sitting. Vertical lines, five knots at equally spaced quantiles; circles, mean values reported from individual trials at different sizes corresponding to sample size; shaded area, 95% confidence interval. mins minutes

3.4.2. Diastolic Blood Pressure

3.4.2.1. Uninterrupted Sitting

Analysis of 145 data points from 24 trials identified a significant effect of time (p = 0.015). Results from a linear meta-regression suggest that every hour of uninterrupted sitting is associated with a 0.24 mmHg (95% CI 0.06–0.42) increase in DBP. Examination of Cook’s distances and studentized residuals identified three trials as potential outliers; however, the removal of each in turn had no effect on the statistical significance of the observed outcome [37, 43, 49]. Removal of trials utilizing continuous BP measurements resulted in a loss of statistical significance (p = 0.07) [56, 58]. The estimated variance components of this model were τ²_{Level 2} = 0.56 and τ²_{Level 3} = 26.9. There was significant heterogeneity in the model (Q(143) = 1644.8, p < 0.01, I² = 90.5%) [I²_{Level 2} = 1.84%, I²_{Level 3} = 88.6%], justifying a further moderator analysis. Removal of each of the potentially influential trials in turn had no effect on the significance of the observed heterogeneity. A further moderator analysis showed no effect of age (p = 0.64) or BMI (p = 0.94). A further analysis identified a non-linear relationship between time and DBP responses to uninterrupted sitting (p_linearity = 0.02), with DBP appearing to decrease within the first 60 min before then increasing beyond baseline values (Fig. 4).

Fig. 4.

Restricted cubic spline of diastolic blood pressure during uninterrupted sitting. Vertical lines, five knots at equally spaced quantiles; circles, mean values reported from individual trials at different sizes corresponding to sample size; shaded area, 95% confidence interval. mins minutes

3.4.2.2. Interrupted Sitting

An analysis of 175 data points from 28 trials identified a significant effect of time (p = 0.001). Results from a linear meta-regression suggest that every hour of interrupted sitting is associated with a 0.24 mmHg (95% CI 0.12–0.42) decrease in DBP. Examination of Cook’s distances and studentized residuals identified six trials as potential outliers; however, the removal of four trials in turn had no effect on the significance of the observed outcome [25, 26, 37, 43]. Removal of the same two trials from one study that significantly influenced the SBP analysis resulted in a loss of overall significance (b = −0.002, p = 0.12) [37]. No included trials included continuous measures of BP, and thus a further sensitivity analysis was not conducted. The estimated variance components of this model were τ²_{Level 2} = 1.0 and τ²_{Level 3} = 19.5. There was significant heterogeneity in the model (Q(173) = 1481.1, p < 0.01, I² = 88.6%) [I²_{Level 2} = 4.4%, I²_{Level 3} = 84.2%], justifying a further moderator analysis. Removal of each of the potentially influential trials in turn had no effect on the significance of the observed heterogeneity. Because of the significant influence of the trials by Dempsey et al., and the lack of robustness of the overall effect, a further moderator analysis was deemed inappropriate [37]. A further analysis identified a non-linear relationship between time and DBP responses to uninterrupted sitting (p_linearity < 0.001), with DBP appearing to decrease within the first 90 min before plateauing (Fig. 5).

Fig. 5.

Restricted cubic spline of diastolic blood pressure during interrupted sitting. Vertical lines, five knots at equally spaced quantiles; circles, mean values reported from individual trials at different sizes corresponding to sample size; shaded area, 95% confidence interval. mins minutes

3.4.3. Mean Arterial Pressure

3.4.3.1. Uninterrupted Sitting

Analysis of 132 data points from 27 trials identified a significant effect of time (p < 0.001). Results from a linear meta-regression suggest that every hour of uninterrupted sitting is associated with a 0.66 mmHg (95% CI 0.36–0.90) increase in MAP. Examination of Cook’s distances and studentized residuals identified five trials as potential outliers; however, the removal of each in turn had no effect on the statistical significance of the observed outcome [39, 41, 56, 59]. Removal of trials utilizing continuous BP measurements had no effect on the statistical significance of the observed outcome but did reduce the magnitude of effect slightly (b = 0.54 mmHg, 95% CI 0.24–0.78, p < 0.001) [56, 58]. The estimated variance components of this model were τ²_{Level 2} = 1.6 and τ²_{Level 3} = 26.6. There was significant heterogeneity in the model (Q(130) = 3323.7, p < 0.01, I² = 97.6%) [I²_{Level 2} = 5.7%, I²_{Level 3} = 92%], justifying a further moderator analysis. Removal of each of the potentially influential trials in turn had no effect on the significance of the observed heterogeneity. A further moderator analysis showed no effect of age (p = 0.07) or BMI (p = 0.06). A further analysis identified a non-linear relationship between time and MAP responses to uninterrupted sitting (p_linearity < 0.001), with MAP appearing to increase within the first 120 min before plateauing (Fig. 6).

Fig. 6.

Restricted cubic spline of mean arterial pressure during uninterrupted sitting. Vertical lines, five knots at equally spaced quantiles; circles, mean values reported from individual trials at different sizes corresponding to sample size; shaded area, 95% confidence interval. mins minutes

3.4.3.2. Interrupted Sitting

Analysis of 60 data points from ten trials identified a non-significant effect of time (p = 0.69). Examination of Cook’s distances and studentized residuals identified two trials as potential outliers; however, the removal of each in turn had no effect on the statistical significance of the observed outcome [34, 41]. As there was no significant effect of time and no effect of influential or outlying trials, no further analysis was conducted. Additionally, a further analysis showed that the addition of a restricted cubic spline did not improve the model fit (p_linearity = 0.76, Fig. 7).

Fig. 7.

Restricted cubic spline of mean arterial pressure during interrupted sitting. Vertical lines, five knots at equally spaced quantiles; circles, mean values reported from individual trials at different sizes corresponding to sample size; shaded area, 95% confidence interval. mins minutes

4. Discussion

The objectives of this meta-analysis were (1) to identify the temporal association between the duration of uninterrupted sitting and BP in adults and (2) to determine whether regular sitting interruptions moderate the association between uninterrupted sitting and BP. The results of this analysis show that (1) a significant relationship exists between sitting duration and increases in SBP, DBP, and MAP and (2) regularly interrupting sitting offsets those deleterious effects.

4.1. Limitations

While this meta-analysis provides novel insights not identified by previous analyses, there are still limitations to consider in contextualizing the present findings. A major limitation of this regression model is the large amount of heterogeneity found in most calculations within the results. While several variables including BMI, age, and length of sitting bout were extracted from the included studies to evaluate covariates that could impact the heterogeneity, none explained a significant portion of the observed heterogeneity. However, the use of a three-level model allowed for the inclusion of all data from relevant studies without violating the statistical assumptions of the meta-analysis model. The inclusion of all available data allowed for an accurate representation of the effect of sitting duration on sitting-induced changes in BP.

A further consideration beyond the sitting duration itself is the issue of validity. Typically, studies that use shorter sitting durations (< 4 h) are geared towards high internal validity. As such, participant movement (bathroom breaks, fidgeting) is typically more tightly regulated and reduced. This level of restricted movement is only feasible with shorter sitting bouts. Alternatively, studies that utilize longer sitting durations (> 4 h) likely aim to achieve a greater level of ecological validity in the form of a simulated workday or similar. As such, participant movement in these studies is typically less tightly controlled than in trials with shorter sitting bouts. This increased movement may maintain muscle activity and thus reduce venous blood pooling, one of the hypothesized mechanisms for sitting-induced dysfunction, and thereby maintain or improve BP [17]. Alternatively, participants involved in longer sitting bouts may be able to interact with external stimuli (e.g., work e-mails), which may affect mental stress levels and increase BP more than if they were sitting in isolation. These potential issues are speculative in nature; however, they are important considerations when interpreting sitting studies of such differing durations.

Methodological differences between studies could represent small aspects of variation and heterogeneity. Two studies (four trials) were noted for measuring BP using volume-clamp finger plethysmography as opposed to the more commonly used oscillometry method [56, 58]. Volume-clamp plethysmography allows for continuous BP measurement as opposed to discrete oscillometric measures; however, previous reviews have concluded that these measures can differ significantly [14, 60–62]. However, in the interest of capturing and analyzing all available data, trials using these different methods of BP assessment were included. A sensitivity analysis was conducted for the trials where continuous volume-clamp plethysmography was used and the uninterrupted DBP model was the only model that lost significance because of the removal of four trials from two studies.

Another methodological difference and potential limitation within this analysis was the various differences in the 24 studies (72%) that fed participants before and/or during sitting bouts. The studies fed participants on average every 4.43 h during sitting, and every study with a sitting bout > 3 h fed their participants. A slight hypotensive effect could be expected after feeding compared with a fasted state [63, 64]. Neither a moderator analysis nor a sensitivity analysis was performed for studies that did or did not feed participants because of the large differences in studies that fed participants at different times, with different foods, or not at all. Most studies controlled the type of food consumed both between conditions and also between participants. A few studies allowed participants to consume their favorite lunch [34], a choice from multiple snacks [35, 55, 65], or a high-fat meal [52]. The one outlier identified in the analysis did include feeding, although the meal was similar to those used in other studies and was standardized to recommended macronutrient targets, suggesting that differences in feeding did not introduce errant variation into the models [37].

4.2. Comparison with the Literature

4.2.1. Prolonged Uninterrupted Sitting

This meta-analysis found that SBP, DBP, and MAP increased over the course of uninterrupted prolonged sitting bouts. Systolic BP increased at a rate of 0.42 mmHg/h, DBP increased at a rate of 0.24 mmHg/h, and MAP increased at a rate of 0.66 mmHg/h. Following the working model proposed by our group, lower-limb blood pooling may be an important contributor to sitting-induced changes in BP [17]. Four (12%) of the included studies measured sitting-induced increases in blood pooling and all four observed that prolonged-sitting increased calf circumference and/or decreased gastrocnemius tissue oxygenation as determined by near-infrared spectroscopy [43, 46, 54, 57]. Increases in lower-limb blood pooling may contribute to reduced venous return and, subsequently, reduced cardiac output [66]. This reduction in cardiac output may then contribute to a reduction in shear stress within the aorta, contributing to acute endothelial dysfunction, resulting in increased arterial stiffness [66]. The aforementioned reduction in cardiac output may also contribute to reduced renal perfusion pressure, thus stimulating the renin–angiotensin–aldosterone system, further increasing BP [17]. A further contributor to the acute sitting-induced increases in BP may be lower-limb vascular dysfunction. A previous meta-analysis reported that acute bouts of prolonged uninterrupted sitting result in lower-limb vascular dysfunction, as indicated by flow-mediated dilation [15]. Indeed, 15 trials in this analysis reported sitting induced lower-limb vascular dysfunction [27, 35, 36, 43, 48, 52–59, 65, 67]. Decreases in lower-limb vascular function may present as increased arterial stiffness, contributing to increased total peripheral resistance and augmented pulse wave propagation, thus driving an increase in BP.

Subsequent spline analyses identified significant non-linear associations between BP and sitting duration, with the greatest increases in SBP and MAP occurring within 60–120 min, after which values plateaued. It is conceivable, therefore, that the greatest increases in blood pooling occur in this early window, after which the rate of pooling decreases. Very few studies have reported changes in blood pooling, and of those studies, most report pre-sitting versus post-sitting changes, which do not allow for investigation of this phenomenon [43, 46, 48, 54, 56, 57, 67]. Counterintuitively, DBP appears to decrease within the first hour of sitting before increasing steadily from that point onwards. The main contributing factors to DBP are total peripheral resistance and arterial compliance [68]. Increases in total peripheral resistance (TPR) result in increases in DBP whereas decreases in compliance (i.e., increases in stiffness) decrease DBP. Whilst it has been shown that TPR increases during bouts of uninterrupted sitting, less is known about changes in arterial compliance [39, 69]. It is possible that acute decreases in compliance secondary to acute declines in endothelial function may drive a temporary decrease in DBP before increased TPR induces the opposite effect [15]. However, this is largely speculative and further investigation is warranted. To fully elucidate how blood pooling may contribute to sitting-induced increases in BP as well as detriments in other cardiovascular markers, studies of differing durations with estimates of venous pooling are necessary. By addressing this gap in the literature, we may move closer towards establishing a biologically plausible model by which sedentary behaviors may contribute to an increased cardiovascular disease risk.

4.2.2. Interrupted Sitting

In sitting trials where an interruption strategy was used, this analysis found a non-significant change in MAP, and trivial significant decreases in SBP and DBP (−0.24 mmHg/h). These results should be treated with caution, however, as the exclusion of two trials from one study resulted in a loss of overall significance in both SBP and DBP models [37]. It should be noted, however, that while the models excluding these trials may not have shown a statistically significant decrease in BP, the lack of a significant increase should be interpreted as a positive outcome. While the aforementioned study fits the inclusion criteria and thus should not be excluded from this analysis, the study sample, patients with type 2 diabetes, may have influenced the observed result. Patients with type 2 diabetes may have greater responses to acute bouts of physical activity, thus producing the large effects observed within this analysis. Because of the lack of robustness within these analyses, any conclusions should be drawn with caution. These results suggest that regularly interrupting bouts of sitting with physical activity interruption strategies may offset the deleterious effects of prolonged uninterrupted sitting on BP.

The beneficial effects of regularly interrupting sitting are likely a result of a reduction in lower-limb pooling via maintained use of the skeletal muscle pump. Seven studies included in this review evaluated lower-limb blood pooling, five using calf circumference as a proxy, and two using near-infrared spectroscopy [43, 46, 48, 54, 56, 57, 67]. All of the studies found significant increases in venous pooling measurements with time, and two of three studies with interruption arms reported less venous pooling [43, 46, 48]. Further, muscle activity is likely to stimulate increased blood flow and thus nitric oxide [17]. This may prevent local vascular dysfunction, particularly in the lower limbs, reducing total peripheral resistance and potentially reducing the effect of increased arterial stiffness of wave reflection timing [70]. While investigating the effect of different interruption strategies was inappropriate owing to the lack of robustness of the main analyses, it is conceivable that strategies that stimulate greater skeletal muscle pump activity may produce greater beneficial effects on BP as well as other cardiovascular markers. Previous reviews have identified that the most efficacious interruption strategies may be aerobic (e.g., walking) or simple resistance activities (e.g., calisthenics); however, to better optimize sitting interruption strategies, it is necessary to investigate whether a dose response to sitting exists [14, 15]. Understanding whether a dose response exists may allow researchers to identify the latest timepoint at which an interruption should be implemented. From that point, researchers can begin to identify optimal interruption timing, type, and intensity.

5. Conclusions

Epidemiological evidence suggests that increased exposure to sedentary behaviors, such as prolonged sitting, may contribute to an increased cardiovascular disease risk and all-cause mortality. Experimental research has identified that acute bouts of prolonged uninterrupted sitting may negatively affect several cardiometabolic variables, including BP. It is hypothesized that repeated and persistent exposure to this acute dysfunction may contribute to an increased risk of cardiovascular disease and all-cause mortality. However, prior to this meta-analysis, it was unclear how sitting duration may impact the magnitude of BP increases and whether regularly interrupting sitting may offset any relationship. This meta-analysis showed that prolonged uninterrupted sitting is associated with a statistically significant increase in MAP, SBP, and DBP and that regularly interrupting sitting may result in maintenance of resting BP, or slight decreases in BP. Building on the purported mechanism of sitting-induced cardiovascular dysfunction (lower-limb blood pooling), future work should investigate whether sitting duration influences lower-limb blood pooling and in turn, whether that pooling affects the degree of sitting-induced cardiovascular dysfunction.

Key Points.

A previous review identified that a bout of prolonged uninterrupted sitting can have detrimental effects on several cardiovascular measures, including peripheral blood pressure. However, it is unclear how sitting duration may impact the magnitude of blood pressure increases.

This meta-regression shows that sitting duration is positively associated with increases in peripheral blood pressure, suggesting that longer sitting bouts are likely to result in greater increases in peripheral blood pressure. However, regularly interrupting sitting may offset these negative effects.

Future work should investigate the mechanisms responsible for this relationship, notably lower-limb blood pooling.

Availability of data and material

The data analyzed for this meta-analysis and the corresponding R script for analysis are available from the corresponding author on reasonable request.