Galmes-Panades et al. International Journal of Behavioral Nutrition and Physical Activity (2019) 16:137 https://doi.org/10.1186/s12966-019-0892-4 RESEARCH Open Access Isotemporal substitution of inactive time with physical activity and time in bed: cross-sectional associations with cardiometabolic health in the PREDIMED- Plus study Aina M. Galmes-Panades1,2, Veronica Varela-Mato3, Jadwiga Konieczna1,2, Julia Wärnberg1,4, Miguel Ángel Martínez-González1,5,6, Jordi Salas-Salvadó1,7,8, Dolores Corella1,9, Helmut Schröder10,11, Jesús Vioque11,12, Ángel M. Alonso-Gómez1,13, J. Alfredo Martínez1,14,15, Luís Serra-Majem1,16, Ramon Estruch1,17, Francisco J. Tinahones1,18, José Lapetra1,19, Xavier Pintó1,20,21, Josep A. Tur1,2,22, Antonio Garcia-Rios1,23, Blanca Riquelme-Gallego11,24, José Juan Gaforio11,25, Pilar Matía-Martín26, Lidia Daimiel27, Rafael Manuel Micó Pérez28, Josep Vidal29,30, Clotilde Vázquez1,31, Emilio Ros1,32, Ana Garcia-Arellano1,5,33, Andrés Díaz-López1,7,8, Eva M. Asensio1,9, Olga Castañer1,10, Francisca Fiol34, Luis Alfredo Mira-Castejón35, Anai Moreno Rodríguez1,13, Juan Carlos Benavente- Marín4, Itziar Abete1,14, Laura Tomaino1,16,36, Rosa Casas1,17, F. Javier Barón López4, José Carlos Fernández-García1,18, José Manuel Santos-Lozano1,19, Ana Galera1,20, Catalina M. Mascaró1,2,22, Cristina Razquin1,5, Christopher Papandreou1,7,8, Olga Portoles1,9, Karla Alejandra Pérez-Vega1,10, Miguel Fiol1,2, Laura Compañ-Gabucio12, Jessica Vaquero-Luna1,13, Miguel Ruiz-Canela1,5, Nerea Becerra-Tomás1,7,8, Montserrat Fitó1,10 and Dora Romaguera1,2* Abstract Background: This study explored the association between inactive time and measures of adiposity, clinical parameters, obesity, type 2 diabetes and metabolic syndrome components. It further examined the impact of reallocating inactive time to time in bed, light physical activity (LPA) or moderate-to-vigorous physical activity (MVPA) on cardio-metabolic risk factors, including measures of adiposity and body composition, biochemical parameters and blood pressure in older adults. Methods: This is a cross-sectional analysis of baseline data from 2189 Caucasian men and women (age 55–75 years, BMI 27–40 Kg/m2) from the PREDIMED-Plus study (http://www.predimedplus.com/). All participants had ≥3 components of the metabolic syndrome. Inactive time, physical activity and time in bed were objectively determined using triaxial accelerometers GENEActiv during 7 days (ActivInsights Ltd., Kimbolton, United Kingdom). Multiple adjusted linear and logistic regression models were used. Isotemporal substitution regression modelling was performed to assess the (Continued on next page) * Correspondence: mariaadoracion.romaguera@ssib.es 1 Centro de Investigación Biomédica en Red Fisiopatología de la Obesidad y la Nutrición (CIBEROBN), Institute of Health Carlos III, Madrid, Spain 2 Research Group on Nutritional Epidemiology & Cardiovascular Physiopathology. Health Research Institute of the Balearic Islands (IdISBa), University Hospital Son Espases, Balearic Islands, Spain Full list of author information is available at the end of the article © The Author(s). 2019 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Galmes-Panades et al. International Journal of Behavioral Nutrition and Physical Activity (2019) 16:137 Page 2 of 12 (Continued from previous page) relationship of replacing the amount of time spent in one activity for another, on each outcome, including measures of adiposity and body composition, biochemical parameters and blood pressure in older adults. Results: Inactive time was associated with indicators of obesity and the metabolic syndrome. Reallocating 30 min per day of inactive time to 30 min per day of time in bed was associated with lower BMI, waist circumference and glycated hemoglobin (HbA1c) (all p-values < 0.05). Reallocating 30 min per day of inactive time with 30 min per day of LPA or MVPA was associated with lower BMI, waist circumference, total fat, visceral adipose tissue, HbA1c, glucose, triglycerides, and higher body muscle mass and HDL cholesterol (all p-values < 0.05). Conclusions: Inactive time was associated with a poor cardio-metabolic profile. Isotemporal substitution of inactive time with MVPA and LPA or time in bed could have beneficial impact on cardio-metabolic health. Trial registration: The trial was registered at the International Standard Randomized Controlled Trial (ISRCTN: http://www. isrctn.com/ISRCTN89898870) with number 89898870 and registration date of 24 July 2014, retrospectively registered. Keywords: Inactive time, Light physical activity, Moderate to vigorous physical activity, Time in bed, Cardiometabolic risk, Isotemporal substitution Background this novel study aims to provide new evidence about the as- Cardio-metabolic diseases such as type 2 diabetes (T2D) sociations of inactive time with cardio-metabolic risk factors and the metabolic syndrome (MetS) are increasingly preva- in an aging population. The outcomes were markers of car- lent worldwide [1, 2]. Overweight and obesity are major risk diometabolic health: measures of adiposity and body com- factors for these metabolic alterations [3–5] and the World position, biochemical parameters, blood pressure, obesity, Health Organization (WHO) has projected a significant type 2 diabetes and metabolic syndrome components. The increase by 2030 [1, 3, 6, 7]. A vast body of the literature objectives of the present study were a) to explore cross- suggests that physical activity (PA) and sedentary behav- sectional associations between inactive time and cardio- iours, including inactive time and time in bed, are strongly metabolic risk factors; and b) to assess the impact of and independently associated with markers of obesity, body replacing 30 min per day of inactive time by 30 min of LPA, composition, and the MetS [2, 8–20]. However, current MVPA and time in bed on markers of cardio-metabolic public health guidelines are mostly focused on the health health. benefits of moderate-to-vigorous physical activity (MVPA) and less attention is given to inactive time [21, 22]. Material and methods Research in older adults highlights the health benefits of Study overview and sample MVPA and light physical activity (LPA) [2, 4, 23, 24]. The PREDIMED-Plus study is a 6-year ongoing multi- However, limited research has explored the associations center, randomized clinical trial, with two intervention between time spent inactive, time in bed and cardio- arms for the primary prevention of cardiovascular dis- metabolic health in an aging population [4, 13, 23, 25–28]; ease in Spain. Details of the study’s protocol have been and, limited research has explored these associations at- described elsewhere [32] and are available on the website tending the 24-h finite time of a day in a population with http://www.predimedplus.com/. Briefly, participant’s in chronic conditions. the intervention are receive a multicomponent weight Isotemporal substitution models have been recommended loss intervention that includes an energy-restricted trad- as one of the most appropriate statistical analysis to explore itional Mediterranean Diet (erMedDiet), PA promotion the associations between reallocating activity patterns, par- and behavioural support. Those in the control group ticularly time spent inactive, and health outcomes [4, 13, 23, receive information about the Mediterranean Diet and 26, 28–30]. This type of analyses takes into account that cardiovascular health guidelines only. The aim of the time is finite. Thus, spending time in one behavior (ie. In- study is to prevent cardiovascular disease (a composite active time) results in less time being spent in another (ie. of cardiovascular death, non-fatal myocardial infarction, MVPA) [29], and understands that daily behaviours (sleep, and non-fatal stroke). Eligible participants were men sedentarism and physical activity) are co-dependent [31]. aged 55–75 and women aged 60–75 years, with body This will provide insightful information that will help better mass index (BMI) ≥27 and < 40 kg/m2, who met ≥3 com- understand the impact of reallocating activity patterns in ponents of the MetS [33]. Overall, 6874 men and women cardio-metabolic markers in older adults. This is crucial for were recruited and randomized into the study between the design of effective tailored interventions to improve car- 2013 and 2016 across 23 Spanish centers distributed diometabolic health in older people in the future. Therefore, throughout the country’s geography; a subsample of Galmes-Panades et al. International Journal of Behavioral Nutrition and Physical Activity (2019) 16:137 Page 3 of 12 2260 participants wore an accelerometer at baseline. Vienna, Austria) using the open-source R-package GGIR, Participants were asked to wear the accelerometer con- version 1.2–5 (cran.rproject.org/web/packages/GGIR/index. tinuously for at least 7 days. Of those days, we excluded html). This open sources code has been validated in rela- invalid days, i.e. those with less of 10 h of data per day. tion to self-calibrated functions [42]. In addition, we excluded participants with less than 3 days of data [34–36]. Therefore, 2189 participants had valid data, defined as 3 or more days of data with more Outcome assessment than 10 h recorded each day. Out of 2189 participants Obesity with accelerometer, 662 had additional data on body Obesity prevalence and obesity indicators were determined composition obtained from Dual-energy X-ray absorpti- based on anthropometric parameters. Anthropometric vari- ometry (DXA) measurements. All participants provided ables were measured by trained personnel according to written informed consent. The study’s protocol was ap- PREDIMED-Plus protocol [32]. Body weight (kg) and height proved by the Research Ethic Committees from all (cm) were measured in light clothing and without shoes recruiting centers according to the ethical standards of using calibrated scales and a wall-mounted stadiometer. the Declaration of Helsinki. The trial was registered at BMI was calculated by dividing the weight (kg) by height in the International Standard Randomized Controlled Trial meters squared (m2). Obesity was defined as a BMI ≥30 kg/ (ISRCTN: http://www.isrctn.com/ISRCTN89898870). m2, and overweight as a BMI ≥ 27 and < 30 kg/m2 (given the inclusion criteria, all of our participants had a BMI ≥ 27 Exposure assessment and < 40 kg/m2). Waist circumference (WC) was measured Participants were asked to wear an accelerometer on at the middle point between the last rib and the iliac crest. their non-dominant wrist (GENEActiv, ActivInsights All anthropometric variables were determined in duplicate, Ltd., Kimbolton, United Kingdom) continuously for 7 and the average of the two measurements was used. days. The GENEActiv is a triaxial accelerometer with a dynamic range of ±8 g, where g is equal to the Earth’s Body composition gravitational pull. The GENEActiv was set to capture Baseline data on total and regional body composition was and store accelerations at a sampling frequency of 40 Hz measured using two types of DXA equipment belonging to [37]. As these activity counts are time and date stamped, the third-generation scanners from GE Healthcare, Madi- detailed data on the time, volume, and intensity of son – WI, connected with EnCore™ software, depending movements can be derived [38]. on the availability of this material in the recruiting centers. Wrist-worn 3-axial accelerometers do not permit to Total body fat mass (expressed as percentage of total body distinguish standing from sitting or reclining postures, a mass), total body muscle mass (expressed as percentage of fact that has conditioned the use of the term inactivity total body mass) and Visceral Adipose Tissue (VAT) mass (include all postures) instead of sedentarism (only sitting (in kg) were measured. For VAT measures, scans were rea- or reclining postures) in the current study. nalyzed using validated CoreScan software application [43]. Data extracted from the GENEActiv (all in bouts of at These algorithms work through detection of the width of least 1 min) was clustered as: inactive time (cut-off inten- the subcutaneous tissue layer on the lateral part of the sity level used was < 40 mg) for those behaviours during abdomen and the anterior-posterior thickness of the abdo- waking hours equivalent to < 1.5 Metabolic Equivalent men, by X-ray attenuation of the abdominal cavity in the Task, METs; LPA (cut-off intensity level used was ≥40 mg android region. DXA scans were performed by trained and < 100 mg) equivalent to 1.5–3 METs; MVPA (cut-off operators following standard protocol and subject position- intensity level used was ≥100 mg) equivalent to > 3 METs; ing provided by the manufacturer. The DXA was phantom and time in bed (time between going to bed and leaving, calibrated daily according to manufacturer guidelines. calculated using a validated heuristic algorithm from accel- erometer raw data unaided by a sleep diary) [37, 39, 40]. For sensitivity analyses we used accelerometer estimated Biochemical analyses and clinical determinations data on sleep time (calculated from accumulated sustained Blood samples were collected after 12 h of overnight fast inactivity bouts (SIB) during time in bed, excluding short and biochemical analysis were performed on fasting plasma wake periods (min/night). SIB are detected as the absence to determine glucose, glycated hemoglobin (HbA1c), low- of change in arm angle greater than 5 degrees for 5 min or density (LDL)-cholesterol, high-density lipoprotein (HDL)- more [41]); nevertheless, our main models are based on cholesterol and triglycerides concentrations using enzymatic time in bed data, given that this estimation has been vali- methods. Blood pressure was measured three times with a dated when no information from sleep diaries is available. validated semiautomatic oscillometer (Omron HEM-705CP, Raw data files were managed on servers at the University the Netherlands) at 5, 10 and 15 min of rest whilst in a of Malaga and processed with R-package (R Core Team, seated position. Galmes-Panades et al. International Journal of Behavioral Nutrition and Physical Activity (2019) 16:137 Page 4 of 12 Metabolic syndrome accumulating less than 7.6 h/day of inactive time (low MetS was defined according to the International Dia- time spent with inactive behaviours). Tertile 2 (T2) in- betes Federation and the American Heart Association cluded those participants who spent between 7.6 and and National Heart, Lung and Blood Institute [33], as 9.3 h/day inactive (moderate time spent with inactive be- having at least 3 of the following components: abdom- haviours). And tertile 3 (T3) included those participants inal obesity for European individuals (WC ≥88 cm in accumulating between 9.3 and 15.1 h/day inactive (high women and ≥ 102 cm in men), hypertriglyceridemia time spent with inactive behaviours). (≥150 mg/dL) or drug treatment for high plasma trigly- Descriptive characteristics were summarized as means ceride concentration, low HDL (< 50 mg/dL in women and standard deviations (SDs) or as numbers and per- and < 40 mg/dL in men), high blood pressure (systolic centages (%). One-way analysis of variance (ANOVA) blood pressure (SBP) ≥130 mmHg or diastolic blood and Chi-square tests (χ2) were used to assess differences pressure (DBP) ≥85 mmHg) or antihypertensive drug across tertiles of inactive time in hours/day for continu- treatment, or high fasting glucose (≥100 mg/dL) or drug ous and categorical variables respectively. treatment for T2D. The presence of MetS was part of First generalized additive models were applied to ascer- the inclusions criteria. tain about the linearity in the association between our exposures and outcomes. Given that there was no evidence Type 2 diabetes of departure from linearly, multivariate linear regression T2D was defined as meeting any of the following cri- analyses were used to estimate the β-coefficients and 95% teria: self-reported diabetes at inclusion or baseline, confidence intervals (CIs) for the associations between HbA1c ≥ 6.5% or use of antidiabetic medication at base- inactive time (continuous variable: bouts of 30 min; cat- line, such as insulin, metformin (in case of diagnosed egorical variable: sex-specific tertiles) and BMI, WC, body diabetes or Hba1c ≥ 6.5%), and other medication for fat, body muscle mass, VAT, HbA1c, glucose, HDL, LDL, diabetes. triglycerides, SBP and DBP. Our models were adjusted by the minimally sufficient adjustment set of covariables, Covariate assessment determined using Directed Acyclic Graphs (DAGs) imple- Baseline data on sex, age, smoking habits, educational level, mented in DAGitty software [46] available free on www. erMedDiet, marital status, medical conditions and medica- dagitty.net. The DAGs were built by identifying all known tion use have been evaluated using self-reported question- factors related to inactive time or our outcomes. Therefore naires. Smoking habits was categorized as current, former our main models were adjusted for age, sex, educational and never smoker; educational level was categorized as level, marital status, erMedDiet, MVPA and smoking. higher education/technician, secondary education and not- Logistic regression models were used to assess the as- completed primary education/primary education; marital sociation between categories of inactive time (tertiles) status was categorized as married and not married, which and the prevalence of the MetS’s components, as well as included single/separated/divorced/widow (er). Adherence prevalence of obesity and T2D. Prevalence ratios (PR) to an energy-restricted Mediterranean diet was measured were calculated using the odds ratios (OR) obtained with using a 17-items ErMedDiet score (score range, 0–17; logistic regression model. PR permits to assess the true higher scores indicate greater adherence). This score is a ratios of prevalence in this sample, given the high preva- modified version of the validated 14-item MEDAS (Medi- lence of MetS, obesity and T2D in the present popula- terranean diet Adherence Screener) used in the PRE- tion, to avoid an overestimation of the risk. PR were DIMED study [44]. We also used data on objectively calculated as [(1-P0) + (P0*OR)] (P0 is the prevalence in measured muscle strength. Lower-limb muscle strength the reference category) [47]. was determined at baseline using previously validated in Linear regression modelling using an isotemporal community-dwelling older subjects 30s-chair-stand test substitution was used to quantify the associations of re- [45]. This test consists of counting the number of stand-sit placing 30 min of inactive time for 30 min of time in on a chair cycles within 30 s. Medication use, including bed, LPA or MVPA on cardio-metabolic risk markers. medication for high blood pressure, for high cholesterol, in- Isotemporal substitution has been recommended for use sulin, metformin, and other medications for diabetes treat- in observational research using time-based measures of ment, were self-reported by participants at baseline and physical activity [30]. Prior to running the models, all checked against medical records. activity patterns (time in bed, inactive time, LPA and MVPA) were divided by a constant of 30, which was Statistical analysis considered as an unit of time equivalent to 30 min (ac- Participants were classified in three categories (tertiles), cording to the PA guidelines [21, 22, 48]). Consequently, depending on the inactive time in hours accumulated in every unit increase represents exchanges of 30 min per one day. Tertile 1 (T1) included those participants day of any of these behaviours. To perform the Galmes-Panades et al. International Journal of Behavioral Nutrition and Physical Activity (2019) 16:137 Page 5 of 12 isotemporal substitution models, a variable representing statistically significant higher BMI, WC, total body fat, the total accelerometer wear time was constructed by VAT, HbAc1, glucose, triglycerides and DBP, and lower adding up time in bed, inactive time, LPA and MVPA. total body muscle mass and HDL cholesterol level. This variable of wear time was entered simultaneously in Table 3 shows the prevalence ratios for obesity, T2D the analysis with time in bed, LPA and MVPA. The and the MetS’s components by categories of inactive resulting regression coefficient represents the association time. Those in the most inactive category (T3) showed of re-allocating a unit of inactive time to a unit of time significantly higher obesity prevalence (p for trend = in bed, LPA and MVPA. Finally, the model was adjusted 0.014), significantly higher triglycerides (p = 0.005), and for age, sex, educational level, marital status, erMedDiet higher number of MetS components, ≥4 components (p and smoking. Analyses follow published guideline for for tend = 0.051) and 5 components (p for tend = 0.054), isotemporal substitution [31]. compared to those with less inactive time (T1). Sensitivity analyses were also conducted. Multiple ad- Table 4 shows the β-coefficients (95% CIs) of the iso- justed linear and logistic regression models were ad- temporal substitution models. Figure 1 shows the same justed for LPA, instead of MVPA, given that LPA is the isotemporal substitution models but outcome variables most prevalent type of PA in our population and in had been standardized as z-scores to aid visualization of older adults in general, and some studies had found results. Isotemporal substitution of 30 min a day of in- beneficial effects of LPA on health [4, 5, 26, 27, 49]. In active time with equivalent time in bed was associated addition, multiple adjusted linear regression models were with lower BMI, WC and HbA1c (all p-values < 0.05); further adjusted for WC when assessing as an outcome: reallocating 30 min of inactive time per day with LPA or HbA1c, glucose, HDL, LDL, triglycerides, SBP and DBP. MVPA (i.e., decreasing inactive time at the expense of Finally, linear regression models shown in Table 2 were increasing LPA or MVPA time) was associated with also adjusted for wear time, and results were consistent lower BMI, WC, total body fat, VAT, HbA1c, glucose, (data not shown). triglycerides, and higher total body muscle mass and In order to test whether the results of the isotemporal HDL (all p-values < 0.05). Estimates of association were replacement models remained similar when using a larger in all variables when replacing 30 min a day of in- proxy measure of sleep time, sleep time and time in bed active time by the equal amount of time in MVPA than were included in the analysis conjointly with the covari- when replacing it by LPA or time in bed. ables mentioned above. No significant changes were observed when performing Statistical analyses were performed using Stata v15.0 sensitivity analyses adjusting linear and logistic regression program. P-values <0.05 were deemed as statistically sig- models for LPA instead of MVPA (See Additional file 1: nificant. All analyses were conducted with data from Table S1 and Table S3), or for WC (See Additional file 1: database PREDIMED-Plus with date 2019-March-12. Table S2). When running the isotemporal replacement models with sleep time instead of time in bed results also Results remained similar (See Additional file 1: Table S4). Table 1 presents a comparison of participants’ characteristics among the three categories of inactive time. Participants in Discussion the upper tertile (T3) of inactivity were significantly older Results from this cross-sectional study show that time and had a higher BMI, WC, total body fat, VAT, HbA1c, glu- spent inactive was associated with a number of cardio- cose and triglycerides concentrations, and lower levels of metabolic risk factors in a sample of older adults, inde- total body muscle mass, and HDL cholesterol. Participants in pendent of PA levels. Overall, this study highlights that T3 presented significantly higher prevalence of T2D (38%) replacing 30 min a day of inactive time with an equal and obesity (78%) compared to the other groups, and re- amount of MVPA, LPA and time in bed resulted in a ported higher consumption of medication for the treatment significantly improved cardio-metabolic profile in men of diabetes. Those in the most inactive category accumulated and women with the MetS. the least amount of time in bed, total PA, LPA and MVPA, The results from this study show that inactive time and accumulated less repeats in the chair-stand test (all p worsens the metabolic profile in an aging population with values < 0.001). Lastly, the highest prevalence of smokers was high cardio-metabolic risk, increasing the chances of car- found among those in T3 (p < 0.001). diovascular events. This is similar to other studies where it Table 2 shows the β-coefficients (95% CIs) for the as- has been found that high levels of inactivity, including sociations between total inactive time, (both per 30-min sitting time, are associated with higher rates of obesity, bouts and in tertiles) and anthropometric measurements, triglycerides and MetS [50, 51] and premature mortality body composition, biochemical parameters and blood and diabetes [51, 52] across different populations. pressure. Higher inactive time was associated with a Isotemporal substitution analyses have public health worse adiposity and cardio-metabolic profile, including implications [4, 9, 13, 23, 25, 28, 29, 31]. Comparative Galmes-Panades et al. International Journal of Behavioral Nutrition and Physical Activity (2019) 16:137 Page 6 of 12 Table 1 Baseline characteristics of the study population across categories of inactive time measured by accelerometer Tertiles of inactive time (h/day)a Total n All T1 n = 735 T2 n = 729 T3 n = 725 P - value Age, years 2189 65.0 (4.95) 64.1 (4.77) 65.2 (4.94) 65.8 (4.98) < 0.001 Women, n (%) 2189 1032 (47.1) 345 (46.9) 343 (47.1) 344 (47.5) 0.979 Anthropometric measures BMI (kg/m2) 2189 32.6 (3.46) 32.2 (3.29) 32.4 (3.37) 33.2 (3.63) < 0.001 Waist circumference (cm) 2189 107 (9.54) 106 (9.41) 107 (9.20) 109 (9.78) < 0.001 DXA Body composition Total body fat (%) 662 40.8 (6.98) 39.4 (6.83) 40.7 (6.84) 42.3 (6.98) < 0.001 VAT (kg) 651 2.27 (0.90) 2.12 (0.85) 2.32 (0.90) 2.38 (0.93) 0.006 Total body muscle mass (%) 662 56.1 (6.62) 57.4 (6.50) 56.2 (6.50) 54.7 (6.61) < 0.001 Clinical parameters HbA1c (%) 2009 6.15 (0.87) 6.08 (0.81) 6.12 (0.79) 6.26 (0.99) < 0.001 Glucose (mg/dL) 2164 114 (29.5) 113 (27.9) 113 (27.8) 117 (32.4) 0.035 HDL (mg/dL) 2173 47.8 (11.9) 49.1 (12.0) 47.7 (11.9) 46.6 (11.6) < 0.001 LDL (mg/dL) 2127 121 (46.5) 122 (31.9) 122 (51.0) 119 (53.7) 0.459 Triglycerides (mg/dL) 2178 152 (78.9) 141 (74.9) 151 (71.4) 165 (87.7) < 0.001 SBP (mmHg) 2175 140 (17.4) 140 (17.0) 141 (17.3) 140 (18.0) 0.293 DBP (mmHg) 2175 80.3 (10.1) 80.6 (9.88) 80.0 (9.92) 80.2 (10.4) 0.430 Diabetes prevalence, n (%) 2189 749 (34.2) 229 (31.2) 246 (33.7) 274 (37.8) 0.027 Obesity prevalence, n (%) 2189 1610 (73.6) 519 (70.6) 528 (72.4) 563 (77.7) 0.007 Metabolic Syndrome High blood pressure, n (%) 2189 1611 (73.6) 534 (72.7) 542 (74.3) 535 (73.8) 0.755 High triglycerides, n (%) 2178 901 (41.4) 244 (33.4) 305 (42.0) 352 (48.9) < 0.001 Low HDL cholesterol, n (%) 2173 888 (40.9) 263 (36.0) 307 (42.3) 318 (44.3) 0.004 High glucose, n (%) 2189 1686 (77.0) 566 (77.0) 561 (77.0) 559 (77.1) 0.998 High waist circumference, n (%) 2189 2098 (95.8) 706 (96.1) 693 (95.1) 699 (96.4) 0.408 Physical activity behavioursb Time in bed (h/day) 2189 8.06 (1.29) 8.28 (1.25) 8.23 (1.21) 7.68 (1.33) < 0.001 Inactive time (h/day) 2189 8.30 (1.98) 6.25 (0.90) 8.21 (0.56) 10.5 (1.31) < 0.001 LPA (h/day) 2189 2.54 (1.07) 3.45 (0.99) 2.45 (0.74) 1.72 (0.63) < 0.001 MVPA (h/day) 2187 0.67 (0.54) 0.93 (0.59) 0.66 (0.49) 0.43 (0.40) < 0.001 Total PA (h/day) 2187 3.22 (1.33) 4.38 (1.22) 3.11 (0.84) 2.15 (0.78) < 0.001 Chair test 30s (repeats) 2189 13.3 (5.01) 13.8 (5.12) 13.3 (5.00) 12.7 (4.83) < 0.001 Sociodemographic/lifestyle data Smoking habits, n (%) 2181 < 0.001 Never 942 (43.2) 328 (44.8) 329 (45.4) 285 (39.4) Current 251 (11.5) 61 (8.3) 75 (10.3) 115 (15.9) Former 988 (45.3) 343 (46.9) 321 (44.3) 324 (44.7) Educational level, n (%) 2171 < 0.001 Higher education/technician 474 (21.8) 113 (15.5) 176 (24.3) 185 (25.8) Secondary education 610 (28.1) 210 (28.8) 198 (27.4) 202 (28.1) Primary education/illiterate 1087 (50.1) 406 (55.7) 350 (48.3) 331 (46.1) Alcohol intake (g/day) 2187 11.4 (15.4) 11.7 (15.5) 11.0 (14.7) 11.5 (16.1) 0.649 Energy intake (kcal/day) 2187 2407 (633) 2444 (640) 2413 (629) 2364 (628) 0.051 Galmes-Panades et al. International Journal of Behavioral Nutrition and Physical Activity (2019) 16:137 Page 7 of 12 Table 1 Baseline characteristics of the study population across categories of inactive time measured by accelerometer (Continued) Tertiles of inactive time (h/day)a Total n All T1 n = 735 T2 n = 729 T3 n = 725 P - value MedDiet score (17 points) 2189 8.56 (2.70) 8.64 (2.75) 8.61 (2.63) 8.42 (2.71) 0.243 Marital status 2179 0.004 Not Married 546 (25.1) 161 (22.0) 174 (23.9) 211 (29.3) Married 1633 (74.9) 570 (78.0) 553 (76.1) 510 (70.7) Medication use for High blood pressure 2189 1700 (77.7) 554 (75.4) 563 (77.2) 583 (80.4) 0.065 Cholesterol 2189 1159 (53.0) 378 (51.4) 384 (52.7) 397 (54.8) 0.437 Insulin 2189 111 (5.07) 36 (4.90) 35 (4.80) 40 (5.52) 0.796 Metformin 2189 544 (24.9) 160 (21.8) 184 (25.2) 200 (27.6) 0.035 Other medications for diabetes 2189 514 (23.5) 151 (20.5) 171 (23.5) 192 (26.5) 0.028 Data shown is mean (SD), unless otherwise specified; Abbreviations: BMI body mass index, VAT visceral adipose tissue, SBP systolic blood pressure, DBP diastolic blood pressure, LDL low density lipoproteins, HDL high density lipoproteins, HbA1c glycated haemoglobin, PA physical activity, LPA light physical activity, MVPA moderate-to-vigorous physical activity, MedDiet Mediterranean diet. a T1, Men (min 3.79 / max 7.6), Women (min 2.64 / max 7.15). T2 Men (min 7.61 / 9.34max); Women (min 7.16 / 8.73max). T3, Men (min 9.35 / 17.2 max), Women (min 8.74/ 15.15max). Tertiles were calculated using the total sample of 2189, and the sample size shown corresponds to the distribution of these 2189 individuals within tertiles; the sample size within tertiles varied for outcome variables with different total sample size. Sample sizes in tertiles of body composition variables determined by DXA were: T1, n = 225; T2, n = 219; T3, n = 218. bData shown of time in bed, physical activity and inactive time has been recorded by accelerometer. High waist circumference was defined has a circumference ≥ 120 cm in men and ≥ 88 cm in women. High glucose was defined as ≥110 mg/dl of glucose in blood or antidiabetic treatment with metformin or insulin. High levels of triglycerides was defined as ≥150 mg/dl of blood. Low levels of HDL cholesterol was defined as ≤40 mg/dl in men and ≤ 50 mg/dl in women. Obesity was defined with a BMI ≥30 kg/m2. Hypertension was defined with a SBP ≥90 mmHg and a SBP ≥140 mmHg. It was considered diabetes if the participant has a previous diagnosis of diabetes, with a HbA1c (%) ≥6.5, or with the use of medication for diabetes treatment as insulin and metformin Table 2 Associations of total inactive time with adiposity indicators and cardio-metabolic risk factors Tertiles of inactive time (h/day)a p for trend Continuous p-value (per 30 min/d of inactive time) Outcome T1 n = 735 T2 n = 729 T3 n = 725 Anthropometric measures BMI (kg/m2) 0 (ref.) 0.22 (−0.14;0.58) 0.90 (0.52;1.29) < 0.001 0.11 (0.06;0.15) < 0.001 Waist circumference (cm) 0 (ref.) −0.05 (− 0.99;0.89) 1.97 (0.96;2.97) < 0.001 0.26 (0.15;0.37) < 0.001 Body Composition Total body fat (%) 0 (ref.) 0.69 (−0.14;1.53) 1.41 (0.50;2.32) 0.002 0.17 (0.07;0.27) 0.001 VAT (Kg) 0 (ref.) 0.16 (0.00;0.31) 0.23 (0.06;0.39) 0.013 0.03 (0.01;0.05) 0.002 Total body muscle mass (%) 0 (ref.) −0.66 (−1.46;0.14) −1.30 (−2.17;-0.44) 0.003 − 0.16 (− 0.25;-0.06) 0.002 Clinical Parameters HbA1c (%) 0 (ref.) 0.01 (−0.09;0.10) 0.11 (0.01;0.21) 0.032 0.02 (0.01;0.03) < 0.001 Glucose (mg/dL) 0 (ref.) 0.08 (−3.06;3.22) 2.68 (−0.68;6.05) 0.116 0.47 (0.11;0.83) 0.010 HDL (mg/dL) 0 (ref.) −1.02 (−2.19;0.15) − 1.64 (− 2.89;-0.39) 0.010 −0.20 (− 0.34;-0.07) 0.003 LDL (mg/dL) 0 (ref.) 1.16 (−3.87;6.19) − 1.23 (−6.63;4.17) 0.652 −0.19 (− 0.76;0.39) 0.526 Triglycerides (mg/dL) 0 (ref.) 8.58 (0.25;16.9) 20.0 (11.0;28.9) < 0.001 2.15 (1.20;3.11) < 0.001 SBP (mmHg) 0 (ref.) 1.43 (−0.41;3.27) 0.34 (−1.63;2.32) 0.740 −0.00 (− 0.21;0.21) 0.981 DBP (mmHg) 0 (ref.) 0.02 (−1.03;1.06) 0.74 (−0.38;1.86) 0.195 0.14 (0.02;0.26) 0.022 Values shown are β (95% CI). Abbreviations: BMI body mass index, VAT visceral adipose tissue, HbA1c glycated haemoglobin, HDL high density lipoprotein, LDL low density lipoprotein, SBP systolic blood pressure, DBP; diastolic blood pressure. Linear regression models were used to assess the association between inactive time and each cardiometabolic risk outcomes, adjusting for age, sex, education level, marital status, erMedDiet, moderate-vigorous physical activity and smoking. a T1, Men (min 3.79 / max 7.6), Women (min 2.64 / max 7.15). T2 Men (min 7.61 / 9.34max); Women (min 7.16 / 8.73max). T3, Men (min 9.35 / 17.2 max), Women (min 8.74/ 15.15max). Tertiles were calculated using the total sample of 2189, and the sample size shown corresponds to the distribution of these 2189 individuals within tertiles; the sample size within tertiles varied for outcome variables with different total sample size. Sample sizes in tertiles of body composition variables determined by DXA were: T1, n = 225; T2, n = 219; T3, n = 218 Galmes-Panades et al. International Journal of Behavioral Nutrition and Physical Activity (2019) 16:137 Page 8 of 12 Table 3 Prevalence Ratio of clinical and metabolic syndrome parameters according to tertiles of inactive time Tertiles of inactive time (h/day)a p for trend Outcome T1 n = 735 T2 n = 729 T3 n = 725 Obesity prevalence 1 (ref.) 1.02 (0.95;1.09) 1.09 (1.02;1.15) 0.014 Diabetes prevalence 1 (ref.) 1.05 (0.89;1.21) 1.14 (0.96;1.32) 0.125 Metabolic Syndrome Components High blood pressure 1 (ref.) 0.97 (0.93;1.00) 1.01 (0.98;1.03) 0.580 High triglycerides 1 (ref.) 1.11 (1.01;1.21) 1.16 (1.05;1.26) 0.005 Low HDL cholesterol 1 (ref.) 1.08 (0.96;1.22) 1.05 (0.90;1.19) 0.533 High glucose 1 (ref.) 0.99 (0.93;1.05) 0.99 (0.92;1.05) 0.674 High waist circumference 1 (ref.) 0.99 (0.96;1.01) 1.00 (0.97;1.02) 0.954 ≥ 4 components metabolic syndrome 1 (ref.) 1.09 (0.97;1.21) 1.13 (1.00;1.25) 0.051 5 components metabolic syndrome 1 (ref.) 1.08 (0.82;1.40) 1.30 (0.99;1.67) 0.054 Values shown are prevalence ratios (95% CI). Abbreviations: HDL high-density lipoprotein. Logistic regression models were adjusted for age, sex, educational level, marital status, erMedDiet, moderate-vigorous physical activity and smoking status. Tertiles were calculated using the total sample of 2189, and the sample size shown corresponds to the distribution of these 2189 individuals within tertiles; the sample size within tertiles varied for outcome variables with different total sample size a T1, Men (min 3.79 / max 7.6), Women (min 2.64 / max 7.15). T2 Men (min 7.61 / 9.34max); Women (min 7.16 / 8.73max). T3, Men (min 9.35 / 17.2 max), Women (min 8.74/ 15.15max). Tertiles were calculated using the total sample of 2189, and the sample size shown corresponds to the distribution of these 2189 individuals within tertiles. Table 4 Isotemporal substitution of inactive time (30 min/day) with time in bed and physical activity on cardio-metabolic risk Outcome Inactive time with time in bed p-value Inactive time with LPA p-value Inactive time with MVPA p-value Anthropometric measures BMI (kg/m2) −0.12 (−0.18;-0.05) < 0.001 −0.19 (− 0.27;-0.11) < 0.001 − 0.40 (− 0.55;-0.25) < 0.001 Waist circumference (cm) − 0.29 (− 0.46;-0.13) 0.001 −0.42 (− 0.62;-0.21) < 0.001 −1.11 (−1.49;-0.72) < 0.001 Body composition Total body fat (%) −0.07 (− 0.21;0.08) 0.391 −0.43 (− 0.62;-0.25) < 0.001 −0.69 (− 1.03;-0.34) < 0.001 VAT (Kg) − 0.03 (− 0.05;0.00) 0.052 −0.06 (− 0.09;-0.03) 0.001 −0.06 (− 0.12;0.01) 0.075 Total body muscle mass (%) 0.05 (−0.09;0.20) 0.465 0.40 (0.22;0.56) < 0.001 0.62 (0.29;0.95) < 0.001 Clinical parameters HbA1c (%) −0.02 (− 0.04;-0.01) 0.006 − 0.03 (− 0.05;-0.01) 0.002 −0.08 (− 0.12;-0.05) < 0.001 Glucose (mg/dL) − 0.09 (− 0.64;0.45) 0.736 −1.05 (− 1.73;-0.38) 0.002 −2.15 (−3.44;-0.87) 0.001 HDL (mg/dL) 0.11 (− 0.10;0.31) 0.306 0.27 (0.01;0.52) 0.039 1.15 (0.67;1.63) < 0.001 LDL (mg/dL) 0.30 (−0.58;1.19) 0.502 0.34 (−0.75;1.43) 0.538 1.43 (−0.63;3.48) 0.174 Triglycerides (mg/dL) −0.89 (−2.35;0.58) 0.235 −1.98 (−3.78;-0.17) 0.032 −9.40 (−12.8;-5.99) < 0.001 SBP (mmHg) 0.09 (−0.24;0.41) 0.597 −0.02 (− 0.42;0.38) 0.906 0.35 (− 0.41;1.11) 0.363 DBP (mmHg) −0.07 (− 0.26;0.11) 0.431 − 0.08 (− 0.30;0.15) 0.505 0.36 (− 0.07;0.79) 0.105 Values shown are β (95% CI). These represent the change in outcome variables when substituting 30 min/day of inactive time with time in bed and physical activity. Abbreviations: LPA light physical activity, MVPA moderate-vigorous physical activity, BMI body mass index, VAT visceral adipose tissue, HbA1c glycated haemoglobin, HDL high density lipoprotein, LDL low-density lipoprotein, SBP; systolic blood pressure, DBP diastolic blood pressure. Linear regression models were used to assess isotemporal substitution of inactive time with: time in bed, light PA and MVPA, adjusting for age, sex, educational level, marital status, erMedDiet, smoking and total wear time a T1, Men (min 3.79 / max 7.6), Women (min 2.64 / max 7.15). T2 Men (min 7.61 / 9.34max); Women (min 7.16 / 8.73max). T3, Men (min 9.35 / 17.2 max), Women (min 8.74/ 15.15max). Tertiles were calculated using the total sample of 2189, and the sample size shown corresponds to the distribution of these 2189 individuals within tertiles Galmes-Panades et al. International Journal of Behavioral Nutrition and Physical Activity (2019) 16:137 Page 9 of 12 Fig. 1 Isotemporal substitution of inactive time (30 min/day) with time in bed and physical activity on standardized cardio-metabolic risk. Values shown are β (95% CI). These represent the change in outcome variables (z-scores) when substituting 30 min per day of inactive time with time in bed and physical activity. Abbreviations: LPA: light physical activity; MVPA: moderate-vigorous physical activity; BMI: body mass index; VAT: visceral adipose tissue; HbA1c: glycated haemoglobin; HDL: high density lipoprotein; LDL: low-density lipoprotein; SBP: systolic blood pressure; DBP: diastolic blood pressure. Linear regression models were used to assess isotemporal substitution of inactive time with time in bed, light PA and MVPA, adjusting for age, sex, educational level, marital status, erMedDiet, and smoking.*indicates p value < 0.05 research in older adults is limited, especially in a popula- however findings remain ambiguous [4, 9, 10, 13, 23–25, tion with chronic conditions, such as MetS [25]. Thus, 23] and although there are some studies in adult popula- the present results are in line with previous research tion [4, 25, 28, 29], few research has been conducted in conducted in adults (18–79 years) [4, 9, 13, 23, 25], older adults [23, 26] and, as far as we know, none in indi- which shows the beneficial effects of exchanging an viduals with overweight/obesity and metabolic syndrome. unit of time spent inactive with equal amounts of PA Given the prevalence of MetS, and the prevalence of or sleep on cardio-metabolic risk factors, including an population aging worldwide, effective and sustain- obesity and lipid profile. This study shows that re- able long term actions are needed. Understanding the placing inactive time with any other behaviors has beneficial effects of substituting time inactive with dif- beneficial effects on cardio-metabolic risk, and these ferent activity levels and sleep in high risk and aging benefits increase proportionally. For instance, re- populations is of importance as it will help defining placing 30 min/d of inactive time with 30 min/d of future tailored health interventions. Multicomponent time in bed was associated with a lower WC of − interventions to increase PA and decrease inactive time, 0.26 cm, whereas replacing this amount of inactive using a multidisciplinary approach are recommended. time with LPA resulted in a WC of − 0.45 cm and According to our results, the promotion of MVPA with MVPA in a WC of − 1.08 cm. would be of most benefit, however in older adults de- Previous studies in adults have highlighted the benefits signing health interventions focused on LPA and sleep of replacing inactive time with MVPA, with the greatest might be more appropriate. Interventions focused on benefits on improved BMI [9], T2D [9, 23], triglycerides, LPA and sleep might result more feasible, appealing HbA1c [23] and glucose. Similarly, this study shows that and might help improve attrition and sustainability in interchanging 30 min of time inactive by MVPA was sig- the long term, as they will not need continuous supervi- nificantly associated with improvements in BMI, WC, body sion and are easy to implement at home or care homes. fat, muscle mass, HbA1c, glucose, HDL and triglycerides. A marked strength of this study was the use of a large Furthermore, the present study shows that health benefits cohort of older men and women, with overweight/obesity are also attained when time inactive is replaced by LPA or and MetS. It is important to highlight that only objective time in bed, with improvements on: BMI, WC, body fat, and validated measurements were used for this study for VAT, muscle mass, HbA1c, glucose, HDL and triglycerides both exposure and outcome variables. This reduces any (LPA); and BMI, WC and HbA1c (time in bed). This is of potential bias or measurement error and increases the op- interest, as research on this area continuous to mount, portunities for comparison across the literature. In terms Galmes-Panades et al. International Journal of Behavioral Nutrition and Physical Activity (2019) 16:137 Page 10 of 12 of limitations, the cross-sectional design prevents the as- Tailored health intervention research with a focus on sessment of causality. Given that exposure and outcome sleep, LPA and MVPA are recommended. variables were measured simultaneously, we cannot rule out reverse causation, i.e. our outcomes, such as obesity, Supplementary information may have preceded inactivity, and not the other way Supplementary information accompanies this paper at https://doi.org/10. 1186/s12966-019-0892-4. round. In addition, due to its cross-sectional study design, the isotemporal replacement model used in this study is Additional file 1: Table S1. Associations of total inactive time with not based on actual replacements of one activity for an- adiposity indicators and cardio-metabolic risk factors (with further other and should be interpreted at the population level; adjustment for light physical activity, instead of moderate-to-vigorous physical activity). Table S2. Associations of total inactive time with longitudinal studies are necessary to confirm the results adiposity indicators and cardio-metabolic risk factors (with further obtained in this study. Selection of older subjects with adjustment for light physical activity, instead of moderate-to-vigorous overweight/obesity and MetS for the study cohort limits physical activity and waist circumference as an indicator of adiposity). Table S3. Prevalence ratio of metabolic syndrome parameters according extrapolation of findings to other populations, including to tertiles of sedentary time (with further adjustment for light physical younger, leaner or healthier subjects. Moreover, this study activity, instead of moderate-to-vigorous physical activity). Table S4. was limited to Caucasians, hence the associations found Isotemporal substitution of inactive time (30 min/day) with sleep time and physical activity on cardio-metabolic risk. may not be applicable to other ethnic groups. Thus, repli- cating this research in different ethnic groups with differ- ent lifestyles and fat distribution would be of interest. Abbreviations BMI: Body mass index; CIs: Confidence intervals; DAG: Directed acyclic Methodological limitations of differentiating between graphs; DBP: Diastolic blood pressure; DXA: Dual-energy X-ray absorpti- sitting, standing are also important to consider. The wrist- ometry; erMedDiet: Energy-restricted traditional Mediterranean Diet; worn 3-axial accelerometers used in this study quantify HbA1c: Glycated hemoglobin; HDL: High-density lipoprotein cholesterol; LDL: Low-density lipoprotein cholesterol; LPA: Light physical activity; time spent in different intensities of activity based on spe- METs: Metabolic equivalent tasks; MetS: Metabolic syndrome; cific count thresholds. This method works reasonably well MVPA: Moderate-to-vigorous physical activity; OR: Odds ratios; PA: Physical for identifying inactive, LPA and MVPA but it is limited activity; PR: Prevalence ratios; SBP: Systolic blood pressure; SDs: Standard deviations; T2D: Type 2 diabetes; VAT: Visceral adipose tissue; WC: Waist in its capacity to distinguish between standing and reclin- circumference; WHO: World health organization ing postures. Thus, throughout this paper we refer to “in- activity” (activities of < 1.5 METs during day-time) and Acknowledgements The authors thank the participants for their enthusiastic collaboration, and not “sedentarism” (meaning activities of < 1.5 METs/day the PREDIMED-Plus personnel and investigators, as well as all affiliated in seated or reclining positions). Another limitation is the primary care centers, for their excellent work. Authors also thaks the IdISBa use of the cutoff intensity level points to cluster data as in- funding with the Liberi program for publishing in open acces journals. active, LPA or MVPA time. Cut points are normally popu- Authors’ contributions lation and protocol specific, limiting the possibility of AMGP and DR conceived of the study. AMGP and VVM completed the comparison across studies and populations [35]. Finally, statistical analysis. AMGP, VVM and DR drafted the manuscript. DR supervised the study. All authors were involved in oversight of recruitment, data although we used validated algorithms to estimate time in collection, revision of the manuscript and read and approved the final bed from accelerometer data without the use of sleep manuscript. diaries, sleep time estimates were less accurate, which Authors’ information prevented us from using sleep time in the main analyses. Not applicable. This issue has been overcame by using several sophisti- cated analysis to assess the complex inter-relationships Funding between different lifestyle behaviours in relation to cardio- The PREDIMED-Plus trial was supported by the official funding agency for biomedical research of the Spanish government, ISCIII through the Fondo de metabolic risk factors. Investigación para la Salud (FIS), which is co-funded by the European Regional Development Fund (four coordinated FIS projects led by Jordi Conclusion Salas-Salvadó and Josep Vidal, including the following projects: PI13/00673, PI13/00492, PI13/00272, PI13/01123, PI13/00462, PI13/00233, PI13/02184, These results add to the growing literature using Isotem- PI13/00728, PI13/01090, PI13/01056, PI14/01722, PI14/00636, PI14/00618, poral Replacement methods and it is one of the few fo- PI14/00696, PI14/01206, PI14/01919, PI14/00853, PI14/01374, PI16/00473, cused on older adults with the metabolic syndrome. PI16/00662, PI16/01873, PI16/01094, PI16/00501, PI16/00533, PI16/00381, PI16/00366, PI16/01522, PI16/01120, PI17/00764, PI17/01183, PI17/00855, Results from this cross-sectional study indicate that re- PI17/01347, PI17/00525, PI17/01827, PI17/00532, PI17/00215, PI17/01441, placing inactive time with any PA and time in bed was as- PI17/00508, PI17/01732, PI17/00926); the Especial Action Project entitled: sociated with improved cardio-metabolic factors in older Implementación y evaluación de una intervención intensiva sobre la actividad física Cohorte PREDIMED-PLUS grant to Jordi Salas-Salvadó; the adults with overweight or obesity and the MetS. Our find- European Research Council grant (Advanced Research Grant 2013–2019; ings support the notion that PA and inactive time are both 340918) to Miguel Ángel Martínez-Gonzalez; the Recercaixa grant to Jordi linked with health outcomes and that both behaviors Salas-Salvadó (2013ACUP00194); grants from the Consejería de Salud de la Junta de Andalucía (PI0458/2013; PS0358/2016; PI0137/2018); the should be included in public health guidelines. Future PROMETEO/2017/017 grant from the Generalitat Valenciana; Grant of support intervention studies are needed to confirm causality. to research groups no. 35/2011 (Balearic Islands Gov. & FEDER funds) to Galmes-Panades et al. International Journal of Behavioral Nutrition and Physical Activity (2019) 16:137 Page 11 of 12 Catalina M. Mascaró and Josep A. Tur; FPU Programme, PhD fellowship Victoria Hospital, Department of Endocrinology, Instituto de Investigación (Spanish Ministry of Science, Innovation & Universities) to Catalina M. Biomédica de Málaga (IBIMA), University of Málaga, Málaga, Spain. 19 Mascaró; the Astra Zeneca Young Investigators Award in Category of Obesity Department of Family Medicine, Research Unit, Distrito Sanitario Atención and Diabetes 2017 to Dora Romaguera; Juan de la Cierva-formación research Primaria Sevilla, Sevilla, Spain. 20Lipids and Vascular Risk Unit, Internal grant (FJCI-2015-24058) of the Spanish Ministry of Economy, Industry and Medicine, Hospital Universitario de Bellvitge, Hospitalet de Llobregat, Competitiveness and European Social Funds to Jadwiga Konieczna; the Barcelona, Spain. 21Department of Medicine, Universidad de Barcelona, ‘FOLIUM’ programme within the FUTURMed project from the Fundación Barcelona, Spain. 22Research Group on Community Nutrition & Oxidative Instituto de Investigación Sanitaria Illes Balears (financed by 2017annual plan Stress, University of Balearic Islands, Palma de Mallorca, Spain. 23Department of the sustainable tourism tax and at 50% with charge to the ESF Operational of Internal Medicine, Maimonides Biomedical Research Institute of Cordoba Program 2014–2020 ofthe Balearic Islands) to Jadwiga Konieczna. JR17/00022 (IMIBIC), Reina Sofia University Hospital, University of Cordoba, Cordoba, contract to Olga Castaner from ISCIII. CIBERobn (Centros de Investigación Spain. 24Department of Preventive Medicine and Public Health, University of Biomedica en Red: Obesidad y Nutrición), CIBEResp (Centros de Investigación Granada, Granada, Spain. 25Departamento de Ciencias de la Salud, Centro de Biomedica en Red: Epidemiología y Salud Publica) and CIBERdem (Centros Estudios Avanzados en Olivar y Aceites de Oliva, Universidad de Jaén, Jaén, de Investigación Biomedica en Red: Diabetes y Enfermedades). None of the Spain. 26Department of Endocrinology and Nutrition, Instituto de funding sources took part in the design, collection, analysis or interpretation Investigación Sanitaria Hospital Clínico San Carlos (IdISSC), Madrid, Spain. 27 of the data and in writing the manuscript, or in the decision to submit the Nutritional Genomics and Epigenomics Group, IMDEA Food, CEI UAM + manuscript for publication. CSIC, Madrid, Spain. 28Cátedrea Cronicidad Universidad Miguel Hernández. Fundación SEMERGEN, Elche, Spain. 29CIBER Diabetes y Enfermedades Availability of data and materials Metabólicas (CIBERDEM), Instituto de Salud Carlos III (ISCIII), Madrid, Spain. 30 There are restrictions on the availability of data for the PREDIMED-Plus trial, Department of Endocrinology, Institut d` Investigacions Biomédiques due to the signed consent agreements around data sharing, which only August Pi Sunyer (IDIBAPS), Hospital Clinic, University of Barcelona, Barcelona, allow access to external researchers for studies following the project Spain. 31Department of Endocrinology and Nutrition, Hospital Fundación purposes. Requestors wishing to access the PREDIMED-Plus trial data used in Jimenez Díaz Instituto de Investigaciones Biomédicas IISFJD, University this study can make a request to the PREDIMED-Plus trial Steering Autonoma, Madrid, Spain. 32Lipid Clinic, Department of Endocrinology and Committee chair: jordi.salas@urv.cat. The request will then be passed to Nutrition, Institut d’Investigacions Biomèdiques August Pi Sunyer (IDIBAPS), members of the PREDIMED-Plus Steering Committee for deliberation. Hospital Clínic, Barcelona, Spain. 33Emergency Department, Complejo Hospitalario de Navarra, Servicio Navarro de Salud (Osasunbidea), Pamplona, Ethics approval and consent to participate Spain. 34Public Health Center Son Serra-La Vileta, Primary Care Management, Research Ethics Committees from all recruitment centers approved the study Balearic Islands Health Service, Palma, Spain. 35Raval Public Health Center, protocol, according to the ethical standards of the Declaration of Helsinki. All Primary Care Management, Elche, Spain. 36Department of Clinical Health and participants provided written informed consent. All centers have the ethics Community Sciences (DISCCO), Università degli Studi di Milano, Milan, Italy. approval and consent from all the ethic committees. The trial was registered at the International Standard Randomized Controlled Trial (ISRCTN: http:// Received: 15 July 2019 Accepted: 27 November 2019 www.isrctn.com/ISRCTN89898870) with number 89898870 and registration date of 24 July 2014, retrospectively registered. Consent for publication References Not applicable. 1. Saklayen MG. The global epidemic of the metabolic syndrome. Curr Hypertens Rep. 2018;20:12. Competing interests 2. Falconer CL, Cooper AR, Walhin JP, Thompson D, Page AS, Peters TJ, et al. The authors declare that they have no competing interests. Sedentary time and markers of inflammation in people with newly diagnosed type 2 diabetes. Nutr Metab Cardiovasc Dis. 2014;24:956–62. Author details 3. Chooi YC, Ding C, Magkos F. The epidemiology of obesity. Metabolism. 1 Centro de Investigación Biomédica en Red Fisiopatología de la Obesidad y 2019;92:6–10. la Nutrición (CIBEROBN), Institute of Health Carlos III, Madrid, Spain. 2Research 4. Colley RC, Michaud I, Garriguet D. Reallocating time between sleep, Group on Nutritional Epidemiology & Cardiovascular Physiopathology. Health sedentary and active behaviours: associations with obesity and health in Research Institute of the Balearic Islands (IdISBa), University Hospital Son Canadian adults. Heal Reports. 2018;29:1–13. Espases, Balearic Islands, Spain. 3School of Sport, Exercise and Health Science, 5. Chastin SFM, Palarea-Albaladejo J, Dontje ML, Skelton DA. Combined effects Loughborough University, Loughborough, UK. 4School of Health Sciences, of time spent in physical activity, sedentary behaviors and sleep on obesity University of Málaga-Institute of Biomedical Research in Malaga (IBIMA), and cardio-metabolic health markers: a novel compositional data analysis Málaga, Spain. 5Department of Preventive Medicine and Public Health, approach. Devaney J, editor. PLoS One. 2015;10:e0139984. IDISNA, University of Navarra, Pamplona, Spain. 6Department of Nutrition, 6. Breda J, Jewell J, Webber L, Galea G. WHO projections in adults to 2030. In: Harvard T.H. Chan School of Public Health, Boston, MA, USA. 7Universitat European congress on obesity; 2015. https://www.karger.com/Article/Pdf/3 Rovira I Virgili, Departament de Bioquímica i Biotecnología, Unitat de Nutrició 82140. 02 May 2019. Humana, Reus, Spain. 8Institut d’Investigació Sanitària Pere Virgili (IISPV), 7. Kelly T, Yang W, Chen C-S, Reynolds K, He J. Global burden of obesity in Hospital Universitari Sant Joan de Reus, Unitat de Nutrició, Reus, Spain. 2005 and projections to 2030. Int J Obes. 2008;32:1431–7. 9 Department of Preventive Medicine, University of Valencia, Valencia, Spain. 8. Lin X, Zhang X, Guo J, Roberts CK, McKenzie S, Wu WC, et al. Effects of 10 Unit of Cardiovascular Risk and Nutrition, Institut Hospital del Mar exercise training on cardiorespiratory fitness and biomarkers of d’Investigacions Mèdiques (IMIM), Barcelona, Spain. 11CIBER de Epidemiología cardiometabolic health: A systematic review and meta-analysis of y Salud Pública (CIBERESP), Instituto de Salud Carlos III, Madrid, Spain. randomized controlled trials. J Am Heart Assoc. 2015;4:1–28. 12 Miguel Hernandez University, ISABIAL-FISABIO, Alicante, Spain. 13Bioaraba 9. Rosique-Esteban N, Díaz-López A, Martínez-González MA, Corella D, Goday Health Research Institute; Osakidetza Basque Health Service, Araba University A, Martínez JA, et al. Leisure-time physical activity, sedentary behaviors, Hospital, University of the Basque Country UPV/EHU, Vitoria-Gasteiz, Spain. sleep, and cardiometabolic risk factors at baseline in the PREDIMED-PLUS 14 intervention trial: a cross-sectional analysis. Song Y, editor. PLoS One. 2017; Department of Nutrition, Food Sciences, and Physiology, Center for Nutrition Research, University of Navarra, Pamplona, Spain. 15Precision 12:e0172253. Nutrition and Cardiometabolic Health program, IMDEA Food, CEI UAM + 10. Galmes-Panades AM, Konieczna J, Abete I, Colom A, Rosique-Esteban N, CSIC, Madrid, Spain. 16Nutrition Research Group, Research Institute of Zulet MA, et al. Lifestyle factors and visceral adipose tissue: results from the Biomedical and Health Sciences (IUIBS), University of Las Palmas de Gran PREDIMED-PLUS study. PLoS One. 2019;14:1–15. Canaria, Las Palmas de Gran Canaria, Spain. 17Department of Internal 11. Ekelund U, Steene-Johannessen J, Brown WJ, Fagerland MW, Owen N, Medicine, Institut d’Investigacions Biomèdiques August Pi Sunyer (IDIBAPS), Powell KE, et al. Does physical activity attenuate, or even eliminate, the Hospital Clinic, University of Barcelona, Barcelona, Spain. 18Virgen de la detrimental association of sitting time with mortality? A harmonised meta- Galmes-Panades et al. International Journal of Behavioral Nutrition and Physical Activity (2019) 16:137 Page 12 of 12 analysis of data from more than 1 million men and women. Lancet. 2016; 32. Martínez-González MA, Buil-Cosiales P, Corella D, Bulló M, Fitó M, Vioque J, 388:1302–10. et al. Cohort profile: design and methods of the PREDIMED-Plus 12. Nagai K, Tamaki K, Kusunoki H, Wada Y, Tsuji S, Ito M, et al. Isotemporal randomized trial. Int J Epidemiol. 2018:1–17. substitution of sedentary time with physical activity and its associations 33. Alberti KGMM, Eckel RH, Grundy SM, Zimmet PZ, Cleeman JI, Donato KA, et with frailty status. Clin Interv Aging. 2018;13:1831–6. al. Harmonizing the metabolic syndrome. Circulation. 2009;120:1640–5. 13. Yates T, Henson J, Edwardson C, Dunstan D, Bodicoat DH, Davies MJ, et al. 34. Brazendale K, Beets MW, Bornstein DB, Moore JB, Pate RR, Weaver RG, et al. Objectively measured sedentary time and associations with insulin Equating accelerometer estimates among youth: the Rosetta stone 2. J Sci sensitivity: importance of reallocating sedentary time to physical activity. Med Sport. 2016;19:242–9. Prev Med. 2015;76:79–83. 35. Rowlands AV, Edwardson CL, Davies MJ, Khunti K, Harrington DM, Yates T. 14. Cooper AR, Sebire S, Montgomery AA, Peters TJ, Sharp DJ, Jackson N, et al. Beyond cut points. Med Sci Sports Exerc. 2018;50:1323–32. Sedentary time, breaks in sedentary time and metabolic variables in people 36. Herrmann SD, Barreira TV, Kang M, Ainsworth BE. How many hours are with newly diagnosed type 2 diabetes. Diabetologia. 2012;55:589–99. enough? Accelerometer wear time may provide bias in daily activity 15. Same RV, Feldman DI, Shah N, Martin SS, Al Rifai M, Blaha MJ, et al. estimates. J Phys Act Health. 2013;10:742–9. Relationship between sedentary behavior and cardiovascular risk. Curr 37. Rowlands AV, Mirkes EM, Yates T, Clemes S, Davies M, Khunti K, et al. Cardiol Rep. 2016;18:6. Accelerometer-assessed physical activity in epidemiology. Med Sci Sports 16. Hu FB, Leitzmann MF, Stampfer MJ, Colditz GA, Willett WC, Rimm EB. Exerc. 2018;50:257–65. Physical activity and television watching in relation to risk for type 2 38. Healy GN, Matthews CE, Dunstan DW, Winkler EAH, Owen N. Sedentary diabetes mellitus in men. Arch Intern Med. 2001;161:1542–8. time and cardio-metabolic biomarkers in US adults: NHANES 200306. Eur 17. Hu FB, Li MY, Colditz GA, Willett WC, Manson JE. Television watching and Heart J. 2011;32:590–7. other sedentary behaviors in relation to risk of obesity and type 2 diabetes 39. van Hees VT, Sabia S, Jones SE, Wood AR, Anderson KN, Kivimäki M, et al. mellitus in women. JAMA. 2003;289:1785. Estimating sleep parameters using an accelerometer without sleep diary. Sci 18. Chau JY, Grunseit A, Midthjell K, Holmen J, Holmen TL, Bauman AE, et al. Rep. 2018;8:12975. Cross-sectional associations of total sitting and leisure screen time with 40. Hildebrand M, Van Hees VT, Hansen BH, Ekelund U. Age group cardiometabolic risk in adults. Results from the HUNT Study, Norway. J Sci comparability of raw accelerometer output from wrist- and hip-worn Med Sport. 2014;17:78–84. monitors. Med Sci Sports Exerc. 2014;46:1816–24. 19. Abete I, Konieczna J, Zulet MA, Galmés-Panades AM, Ibero-Baraibar I, Babio 41. van Hees VT, Sabia S, Anderson KN, Denton SJ, Oliver J, Catt M, et al. A N, et al. Association of lifestyle factors and inflammation with sarcopenic novel, open access method to assess sleep duration using a wrist-worn obesity: data from the PREDIMED-Plus trial. J Cachexia Sarcopenia Muscle. accelerometer. PLoS One. 2015;10:e0142533. 2019. https://doi.org/10.1002/jcsm.12442. 42. van Hees VT, Fang Z, Langford J, Assah F, Mohammad A, da Silva ICM, et al. 20. Salas-Salvadó J, Díaz-López A, Ruiz-Canela M, Basora J, Fitó M, Corella D, et Autocalibration of accelerometer data for free-living physical activity al. Effect of a lifestyle intervention program with energy-restricted assessment using local gravity and temperature: an evaluation on four Mediterranean diet and exercise on weight loss and cardiovascular risk continents. J Appl Physiol. 2014;117:738–44. factors: one-year results of the PREDIMED-plus trial. Diabetes Care. 2018;42: 43. Kaul S, Rothney MP, Peters DM, Wacker WK, Davis CE, Shapiro MD, et al. dc180836. Dual-energy X-ray absorptiometry for quantification of visceral fat. Obesity 21. Murphy MH, Broom DR, Gill JMR, Gray CM, Jones A, Steele J, et al. UK (Silver Spring). 2012;20:1313–8. physical activity guidelines: draft review and recommendations for adults 44. Álvarez-Álvarez I, Martínez-González MÁ, Sánchez-Tainta A, Corella D, Díaz- (aged 19-64 years). 2018. https://www.bristol.ac.uk/media-library/sites/sps/ López A, Fitó M, et al. Adherence to an energy-restricted Mediterranean documents/cmo/adults-technical-report.pdf. 30 April 2019. diet score and prevalence of cardiovascular risk factors in the PREDIMED- 22. Piercy KL, Troiano RP, Ballard RM, Carlson SA, Fulton JE, Galuska DA, et al. plus: a cross-sectional study. Rev Esp Cardiol (English Ed). 2019;72:925–34. The physical activity guidelines for Americans. JAMA. 2018;9762:1–9. 45. Jones CJ, Rikli RE, Beam WC. A 30-s chair-stand test as a measure of lower 23. Hamer M, Stamatakis E, Steptoe A. Effects of substituting sedentary time with body strength in community-residing older adults. Res Q Exerc Sport. 1999; physical activity on metabolic risk. Med Sci Sports Exerc. 2014;46:1946–50. 70:113–9. 46. Textor J, van der Zander B, Gilthorpe MS, Liśkiewicz M, Ellison GT. Robust 24. Maher CA, Mire E, Harrington DM, Staiano AE, Katzmarzyk PT. The causal inference using directed acyclic graphs: the R package “dagitty”. Int J independent and combined associations of physical activity and sedentary Epidemiol. 2016;45:1887–94. behavior with obesity in adults: NHANES 2003-06. Obesity. 2013;21:E730–7. 47. Zhang J, Yu KF. What’s the relative risk? JAMA. 1998;280:1690. 25. Buman MP, Winkler EAH, Kurka JM, Hekler EB, Baldwin CM, Owen N, et al. 48. World Health Organization. World recommendations on physical activity for Reallocating time to sleep, sedentary behaviors, or active behaviors: health. 2010. https://www.who.int/dietphysicalactivity/global-PA-recs-2010. associations with cardiovascular disease risk biomarkers, NHANES 2005– pdf. 02 May 2019. 2006. Am J Epidemiol. 2014;179:323–34. 49. Ekelund U, Tarp J, Steene-Johannessen J, Hansen BH, Jefferis B, Fagerland 26. Lerma NL, Cho CC, Swartz AM, Miller NE, Keenan KG, Strath SJ. Isotemporal MW, et al. Dose-response associations between accelerometry measured substitution of sedentary behavior and physical activity on function. Med physical activity and sedentary time and all cause mortality: systematic Sci Sports Exerc. 2018;50:792–800. review and harmonised meta-analysis. BMJ. 2019. https://doi.org/10.1136/ 27. Falconer CL, Page AS, Andrews RC, Cooper AR. The potential impact of bmj.l4570. displacing sedentary time in adults with type 2 diabetes. Med Sci Sports 50. Varela-Mato V, O’Shea O, King JA, Yates T, Stensel DJ, Biddle SJ, et al. Cross- Exerc. 2015;47:2070–5. sectional surveillance study to phenotype lorry drivers’ sedentary 28. Biddle G, Edwardson C, Henson J, Davies M, Khunti K, Rowlands A, et al. behaviours, physical activity and cardio-metabolic health. BMJ Open. 2017;7: Associations of physical behaviours and behavioural reallocations with e013162. markers of metabolic health: a compositional data analysis. Int J Environ Res 51. Wilmot EG, Edwardson CL, Achana FA, Davies MJ, Gorely T, Gray LJ, et al. Public Health. 2018;15:2280. Sedentary time in adults and the association with diabetes, cardiovascular 29. Edwardson CL, Henson J, Bodicoat DH, Bakrania K, Khunti K, Davies MJ, et al. disease and death: systematic review and meta-analysis. Diabetologia. 2012; Associations of reallocating sitting time into standing or stepping with 55:2895–905. glucose, insulin and insulin sensitivity: a cross-sectional analysis of adults at 52. Edwardson CL, Gorely T, Davies MJ, Gray LJ, Khunti K, Wilmot EG, et al. risk of type 2 diabetes. BMJ Open. 2017;7:e014267. Association of sedentary behaviour with metabolic syndrome: a meta- 30. Mekary RA, Willett WC, Hu FB, Ding EL. Isotemporal substitution paradigm analysis. O’Connor KA, editor. PLoS One. 2012;7:e34916. for physical activity epidemiology and weight change. Am J Epidemiol. 2009;170:519–27. 31. Dumuid D, Pedišić Ž, Stanford TE, Martín-Fernández J-A, Hron K, Maher CA, Publisher’s Note et al. The compositional isotemporal substitution model: a method for Springer Nature remains neutral with regard to jurisdictional claims in estimating changes in a health outcome for reallocation of time between published maps and institutional affiliations. sleep, physical activity and sedentary behaviour. Stat Methods Med Res. 2019;28:846–57.