Longitudinal leisure-time physical activity profiles throughout adulthood and related characteristics : a 36-year follow-up study of the older Finnish Twin Cohort

This is a self-archived version of an original article. This version may differ from the original in pagination and typographic details. Author(s): Title: Year: Version: Copyright: Rights: Rights url: Please cite the original version: CC BY 4.0 https://creativecommons.org/licenses/by/4.0/ Longitudinal leisure-time physical activity profiles throughout adulthood and related characteristics : a 36-year follow-up study of the older Finnish Twin Cohort © 2024 the Authors Published version Berntzen, Bram J.; Tolvanen, Asko; Kujala, Urho M.; Silventoinen, Karri; Vuoksimaa, Eero; Kaprio, Jaakko; Aaltonen, Sari Berntzen, B. J., Tolvanen, A., Kujala, U. M., Silventoinen, K., Vuoksimaa, E., Kaprio, J., & Aaltonen, S. (2024). Longitudinal leisure-time physical activity profiles throughout adulthood and related characteristics : a 36-year follow-up study of the older Finnish Twin Cohort. International Journal of Behavioral Nutrition and Physical Activity, 21, Article 47. https://doi.org/10.1186/s12966-024-01600-y 2024 Berntzenetal. Int J Behav Nutr Phys Act (2024) 21:47 https://doi.org/10.1186/s12966-024-01600-y RESEARCH Longitudinal leisure-time physical activity profiles throughoutadulthood andrelated characteristics: a36-year follow-up study oftheolder Finnish Twin Cohort Bram J. Berntzen1, Asko Tolvanen2, Urho M. Kujala3, Karri Silventoinen4, Eero Vuoksimaa1, Jaakko Kaprio1 and Sari Aaltonen1* Abstract Background Personalized interventions aiming to increase physical activity in individuals are effective. However, from a public health perspective, it would be important to stimulate physical activity in larger groups of people who share the vulnerability to be physically inactive throughout adulthood. To find these high-risk groups, we identified 36-year leisure-time physical activity profiles from young adulthood to late midlife in females and males. Moreover, we uncovered which anthropometric-, demographic-, lifestyle-, and health-related characteristics were associated with these physical activity profiles. Methods We included 2,778 females and 1,938 males from the population-based older Finnish Twin Cohort Study, who responded to health and behavior surveys at the mean ages of 24, 30, 40 and 60. Latent profile analysis was used to identify longitudinal leisure-time physical activity profiles. Results We found five longitudinal leisure-time physical activity profiles for both females and males. Females’ profiles were: 1) Low increasing moderate (29%), 2) Moderate stable (23%), 3) Very low increasing low (20%), 4) Low stable (20%) and 5) High increasing high (9%). Males’ profiles were: 1) Low increasing moderate (29%), 2) Low stable very low (26%), 3) Moderate decreasing low (21%), 4) High fluctuating high (17%) and 5) Very low stable (8%). In both females and males, lower leisure-time physical activity profiles were associated with lower education, higher body mass index, smoking, poorer perceived health, higher sedentary time, high blood pressure, and a higher risk for type 2 diabetes. Furthermore, lower leisure-time physical activity was linked to a higher risk of depression in females. Conclusions We found several longitudinal leisure-time physical activity profiles with unique changes in both sexes. Fewer profiles in females than in males remained or became low physically active during the 36-year follow-up. We observed that lower education, higher body mass index, and more smoking already in young adulthood were associated with low leisure-time physical activity profiles. However, the fact that several longitudinal profiles demonstrated a change in their physical activity behavior over time implies the potential for public health interventions to improve leisure-time physical activity levels. Keywords Exercise, Latent classes, Longitudinal, Sports, Working-age individuals Open Access © The Author(s) 2024. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. 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 in a credit line to the data. International Journal of Behavioral Nutrition and Physical Activity *Correspondence: Sari Aaltonen [email protected] Full list of author information is available at the end of the article Page 2 of 15 Berntzenetal. Int J Behav Nutr Phys Act (2024) 21:47 Introduction Long-term physical inactivity in adulthood is a major global health challenge [1], with a substantial economic burden [2]. The physical inactivity related health challenge can partly reflect the findings of previous research that have established heterogeneity and low stability in physical activity (PA) behavior but higher stability in physical inactivity behavior during the life course [3]. Personalized interventions that aim to increase PA could be an effective way to improve health and well-being in adulthood [4]. However, at the population level, it would be more valuable to increase leisure-time PA (LTPA) in larger groups of people who share similar longitudinal LTPA profiles or trajectories during the life course and then tailor interventions by the unique characteristics of these groups. Although PA research increasingly addresses longitudinal PA profiles and trajectories, so far, only a minority of studies have tracked PA over multiple decades. A systematic review of group-based PA trajectories was published in 2019 and included 27 studies [5]. Of these studies, four had a follow-up duration of over three decades which addressed two different cohorts: 1) the Cardiovascular Risk in Young Finns Study (YFS) [6, 7] and 2) the Northern Finland Birth Cohort [8, 9]. These studies with baseline measurements in 1980 uncovered three to five PA trajectories and showed a high proportion of physical inactivity at all ages, which further increased with aging. Published after that review, another 31-year follow-up study in the YFS of children and adolescents aged 9 to 18 at baseline found that about two-thirds of the participants belonged to trajectories with low LTPA levels at the end of follow-up [10]. Recently, Norwegian researchers focused on vigorous LTPA in their 27-year follow-up study [11]. They found four trajectories of vigorous LTPA between ages 13 and 40 years. In order to cover overall PA over the entire lifespan, one study investigated older adults who recalled their PA in young, middle, and older adulthood [12]. This retrospective study found 6 lifecourse PA trajectories; 73% of adults were identified in trajectory classes that indicated persistent low PA and 13% in trajectory classes showed a large decrease in PA over time. Even fewer studies have tried to uncover whether the PA profiles or trajectories over three decades differ between females and males. The existing studies have suggested that more females belong to low PA trajectory classes than males [6, 10]. These results also reflect those of a trajectory study of a somewhat shorter follow-up (i.e., 24 years) [13]. When females and males have been studied separately, consistently high and increasing LTPA trajectories have been linked to healthier diets and less smoking in both sexes [10], as well as to the absence of sleep difficulties [10] and less sedentary behavior measured as television time in females [14]. In studies combining females and males, low or decreasing PA trajectories over three decades have been shown to be associated with low education [6], smoking [6] and depressive symptoms [7]. Studies combining females and males in PA profiles or trajectories with shorter follow-ups – between 3 and 27 years – have shown that low and decreasing PA profiles and trajectories are associated with lower socioeconomic status [5], lower income [11], poorer diet [15, 16], lower alcohol consumption [15–17], smoking [15], poorer subjective health [15, 17], cognitive decline [16, 18], the onset of depression [7], higher body mass index (BMI) [16, 17, 19], incidence of type 2 diabetes (T2D) [20], and an increased risk of cardiovascular disease [21]. Previous evidence also suggests that the health benefits of LTPA may depend on occupational PA [22]. Considering all of this evidence, it seems that only the 31-year YFS study focusing on LTPA from childhood to early midlife has tried to uncover different PA trajectories of more than three decades and simultaneously examine how multiple anthropometric, demographic, and lifestyle factors are associated with these trajectories in females and males separately [10]. Thus, a longitudinal profile or trajectory study covering LTPA over the course of working life is still lacking. However, this would be a key life phase to intervene upon, not only to reduce health complications that come with long-term physical inactivity and aging, but also to improve productivity at work [23]. Therefore, our aim is to identify 36-year LTPA profiles for females and males from young adulthood to late midlife. Furthermore, weaim to uncover which anthropometric-, demographic-, lifestyle-, and health-related characteristics are associated with these longitudinal LTPA profiles. Methods Participants The participants of this study were from the older Finnish Twin Cohort study, which is a longitudinal populationbased study of twins from same-sex pairs born in Finland before 1958 (N=13,888) [24, 25]. Twins responded to health survey questionnaires in 1975, 1981, 1990 and 2011 (response rates from 89% to 72%). The research was conducted according to the principles of the Declaration of Helsinki, and the data collection was approved by the ethics committee of the Hjelt Institute, University of Helsinki and the ethics committee of the Helsinki and Uusimaa Hospital District, Finland. All participants gave informed consent. To create longitudinal LTPA profiles from young adulthood to late midlife, we used data from all four followups. We included those twins who were between ages 18–31 (mean age 24.1 years; N=11,921 individuals) at Page 3 of 15 Berntzenetal. Int J Behav Nutr Phys Act (2024) 21:47 baseline in 1975. In 1981 and 1990, the participants were at the mean ages of 30.3 years (age range 23–39; N=10,689 individuals) and 39.5 years (age range 32–47; N=7,473 individuals), respectively. At the last follow-up in 2011, the participating twins had reached the mean age of 60.2 years (age range 53–67; N=7,381 individuals). Hereafter the mean ages are referred to as 24, 30, 40 and 60. In total, we had LTPA data at all time points available from 4,716 twin individuals (2,778 females and 1,938 males), including 1,328 complete twin pairs. Assessment ofleisure‑time physical activity Participants reported their LTPA, which was quantified as metabolic equivalent of task (MET) hours expended per day. These values were based on a series of structured and validated items on the frequency, mean duration and mean intensity of LTPA sessions, as well as an item on commuting activity [26, 27]. To calculate MET hours per day, we used the following formula: LTPA frequency (average per day) × LTPA duration (average hours) × LTPA intensity (activity MET score) [28]. The following MET values were used for the intensity of LTPA to obtain a multiple of the resting metabolic rate for each activity: 4 corresponded to walking, 6 corresponded to vigorous walking to jogging, 10 corresponded to jogging, and 13 corresponded to running. The MET value of 4 (walking) was also used for the intensity of commuting-related PA. We further assumed that commuting-related PA was done on 5 days per week. All types of LTPA and commuting-related physical activities were considered when MET hours per day were calculated. The LTPA items were included in all survey questionnaires in the same form, except in 1990 when only one item was used. This item measures combined information on the frequency, duration and intensity of LTPA, including commuting activity. We converted this item to MET-hours/day as well. Assessment ofdemographic, anthropometric andlifestyle characteristics To investigate the associations between longitudinal LTPA profile membership and various characteristics, we selected a group of characteristics, that have been shown to be related to different longitudinal PA profiles and trajectories in previous studies (details given in the introduction) [5–7, 10, 11, 13, 15–22, 29, 30]. The selection of these characteristics was also based on prior large systematic reviews, meta-analyses and guidelines, indicating the disease, health, and lifestyle correlates of PA and physical inactivity behaviors [31–34]. Scientific evidence on the importance of some lifestyle-related characteristics (e.g., sitting) has become available only in recent decades and, therefore, such characteristics were only available from the last follow-up survey questionnaire in 2011. At baseline, the participants reported their financial situation by responding to a structured survey item with an 8-point Likert scale on their monthly income. Higher scores indicate higher monthly income. The measure of education was the self-reported highest educational degree achieved at ages 24 and 30, based on the 1975 and 1981 surveys. The eight categories of level of education ranged from less than compulsory education (1) to tertiary education (i.e., university or polytechnic college) (8) [35]. The participants reported their height and weight in all surveys, and BMI was calculated as the ratio between weight in kilograms and height in square meters (kg/m2). Self-reported BMI has been validated in this cohort [36, 37]. At age 60, the participants were also asked to measure their waist circumference with a measuring tape sent along with the survey questionnaire. Regarding lifestyle characteristics, alcohol consumption, smoking and sleeping were reported by the participants at all time points. The number of monthly alcoholic beverages participants reported to drink was converted into grams of 100% alcohol per month [38, 39]. Smoking status (never/former/current) was defined using responses to two dichotomized (yes/no) items on [40]: 1) the history of ever smoking more than 5 - 10 packs of cigarettes and 2) current or previous daily smoking. The structured response options for sleep time were as follows [41]: 1) 7-point Likert scale from <4 hours (1) to >10 hours per night (7) at baseline and 2) 9-point Likert scale from <6 hours (1) to >10 hours per night (9) at all other follow-up time points. Work-related PA data were collected at baseline and the last follow-up time point. The initial 4-category variable of the physical strain of work was used to create a dichotomized variable for the purpose of analysis: 1) sedentary work that may involve walking and 2) manual work that may involve lifting and carrying heavy objects. At the last follow-up time point, participants also reported how many hours they were sitting (i.e., sedentary behavior) per day 1) in an office, 2) while watching TV or videos at home, 3) at a computer at home, 4) in a vehicle, and 5) elsewhere. The 4 response options given for sitting ranged from <1 hour (1) to >4 hours per day (4), and a sum score of the distinct sitting categories were used to create a final sedentary behavior variable (score ranged 1–20) with higher scores indicating higher overall sitting times [42]. Assessment ofhealth characteristics At the last follow-up time point, the participants rated how they perceived their health in general. The 5-point Likert scale ratings ranged from “very good” (1) to “very poor” (5). For data analysis, we dichotomized the Page 4 of 15 Berntzenetal. Int J Behav Nutr Phys Act (2024) 21:47 subjective health variable so that the response options “very good” and “good” were defined as “good” (1), while the rest of the response options were defined as “poor” (2). In addition to this general health item, the participants also reported whether they had ever been diagnosed by a physician (yes/no) with: high blood pressure (all follow-ups), T2D (all follow-ups), coronary artery disease, including angina pectoris (all follow-ups), or depression (last follow-up). Statistical analysis In order to identify different longitudinal LTPA profiles from young adulthood to late midlife, we used the latent profile analysis, which can be considered as a subset of finite mixture models [43]. In the latent profile analysis, researchers do not impose growth trends on the data. Rather, the profiles are a direct reflection of the data making them a more accurate description of the profiles. We classified participants into distinct profiles based on the means and variances of their LTPA behavior (i.e., there were different means and variances between the identified profiles). The correlation between LTPA variables was set to zero because local independence is the basic assumption underlying the latent profile analysis. Eventually, this analysis of 36-year LTPA behavior across four time points led to longitudinal profiles in which individuals within a profile were more similar than individuals between profiles. Using this approach, we first estimated and compared models with differing numbers of latent profiles to determine which of the models fit the data best, using the Bayesian information criteria (BIC) and Lo-MendelRubin adjusted test (LMR). The lower the BIC value is, the better the model fits the data, while a p-value < 0.05 in the LMR test is used as an indicator to reject the model with fewer profiles (i.e., k number of latent profiles fits the data better than k-1 number of latent profiles). For the best fitting model, we calculated entropy and average latent profile probabilities, indicating the distinctiveness between profiles. We conducted the analyses separately for females and males. If the results suggested an equal number of latent profiles for females and males, we continued our analyses by comparing the profiles between the sexes in successive steps: (1) the equality of mean values, (2) the equality of variances and (3) the equality of latent profile sizes [44]. We estimated all the models using the maximum likelihood method with the Mplus 8.7 statistical package [45]. The maximum likelihood estimation with robust standard errors (robust to non-normality) was utilized. After the final number of LTPA profiles were identified, we continued analyses by examining associations between latent profile membership and various demographic, anthropometric, lifestyle and health characteristics using the one-step Bolck-Croon-Hagenaars method for continuous variables and the two-step model-based approach test for categorical variables [46, 47]. Regarding these association results, the p-value of < 0.001 corresponds to a multiple-test corrected Bonferroni p-value < 0.05 (i.e., 0.05/45 tests=0.001). Moreover, because we had complete twin pairs in our data, the non-independency of data was possible (the observations and their error terms between the co-twins of a twin pair can be correlated). Therefore, we used the type=complex-option to calculate unbiased standard errors and p-values when examining the associations between latent profile membership and demographic, anthropometric, lifestyle and health characteristics. Results The descriptive statistics of leisure-time physical activity, demographic, anthropometric, lifestyle, and health characteristics are given by sex and age in Supplementary Table1. We identified 6-profile models to be the best models for both females and males based on the BIC values (Table1). However, according to the LMR test, for both sexes these 6-profile solutions did not fit better than 5-profile solutions. A closer inspection also further revealed that the 6-profile solution divided one of the 5-profile solutions into two profiles that were equal in their shapes both in females and males. Therefore, to follow the general statistical principle of parsimony, we chose the 5-profile solution as final models for both females and males. Because these results suggested an equal number of profiles for females and males, we further statistically tested the similarity of the 5-profile solutions between females and males. The BIC value for the freely estimated model increased from 89975.01 to 90033.30, when setting latent profile means equal. This indicated that latent profile solutions needed to be analyzed separately for females and males. Longitudinal LTPA profiles We labeled the longitudinal LTPA profiles based on their unique elements 1) at baseline, 2) during the transition toward the final time point, and 3) at the end of followup (Fig.1 and Supplementary Table2). The largest proportions of both females (n=792, 29%; Profile 1) and males (n=564, 29%; Profile 3) belonged to Low increasing moderate profiles. As our labeling indicates, these profiles in both sexes were characterized by low LTPA levels at baseline, then increasing their LTPA levels first slightly and then more steeply to a mean of 4.5 (females) and 4.3 (males) MET hours/day. The second largest proportion of females (n=628, 23%) was assigned to Profile 4, Moderate stable, characterized by a stable level of Page 5 of 15 Berntzenetal. Int J Behav Nutr Phys Act (2024) 21:47 LTPA (between 3.4 and 3.9 MET hours/day) throughout the entire follow-up. In males, the corresponding moderate-level LTPA profile was Profile 5 (n=405, 21%) that we labeled as Moderate decreasing low. This profile was characterized by a steady LTPA decrease over time (from mean 4.5 MET hours/day to 2.4 MET hours/day). The third largest proportions of females belonged to Profile 2 (n=553, 20%), Very low increasing low. These women had Table 1 The comparison of models with differing numbers of latent profiles to determine which of the models fits the data best BIC Bayesian information criteria, LMR Adjusted Lo-Mendell-Rubin likelihood ratio test, AvePP Average latent profile probabilities for the most likely latent profile membership Number of latent profiles BIC LMR Entropy AvePP Profile sizes Females 1 55979.93 NA 1.0 1.0 2778 2 50078.58 < 0.001 0.838 0.963; 0.939 2149; 629 3 48693.20 < 0.001 0.790 0.873; 0.924; 0.931 777; 1565; 436 4 48250.09 0.002 0.725 0.746; 0.848; 0.946; 0.876 542; 752; 367; 1117 5 47914.81 0.002 0.690 0.798; 0.802; 0.931; 0.830; 0.760 792; 553; 256; 628; 549 6 47743.52 0.078 0.695 0.754; 0.794; 0.841; 0.785; 0.912; 0.815 524; 499; 197; 804; 155; 599 Males 1 43138.70 NA 1.0 1.0 1938 2 37605.22 < 0.001 0.857 0.958; 0.961 670; 1268 3 36391.14 < 0.001 0.840 0.906; 0.953; 0.934 445; 510; 983 4 36002.79 0.095 0.745 0.882; 0.797; 0.804; 0.956 435; 544; 530; 429 5 35601.98 0.007 0.773 0.846; 0.926; 0.838; 0.929; 0.835 502; 147; 563; 321; 405 6 35284.26 0.167 0.765 0.917; 0.788; 0.828; 0.932; 0.823; 0.827 113; 315; 458; 247; 474; 331 Fig. 1 Longitudinal leisure-time physical activity profiles from young adulthood to late midlife in females and males. MET=metabolic equivalent of task Page 6 of 15 Berntzenetal. Int J Behav Nutr Phys Act (2024) 21:47 a mean MET hours/day of 0.7 at baseline, slightly increasing from age 30 to reach the mean level of 2.1 MET hours/day at the end of the follow-up. In males, the corresponding profile, but with an opposing trend, was Profile 1 (n=502, 26%), Low stable very low, whichremained fairly stable with a minimal decrease, going from 1.6 MET hours/day to 1.2 MET hours/day. The fourth largest female LTPA group was Profile 5 (n=549, 20%), Low stable. This profile was characterized by a stable low level of LTPA (between mean 2.0 and 1.5 MET hours/day) throughout the follow-up, having the lowest LTPA level at the end of follow-up. In males, the profile with the lowest LTPA at all timepoints was Profile 2 (n=147, 8%), Very low stable. This LTPA profile portrayed a very inactive group of young male adults (mean 0.4 MET hours/day), who slightly increased their LTPA level, but kept it fairly constant until the end of follow-up (at the highest point the mean level was 1.2 MET hours/day). The smallest proportion of females (n=256, 9%) belonged to the profile of the highest level of LTPA, Profile 3, High increasing high. Females in this profile were those with the highest LTPA level in young adulthood (mean 6.0 MET hours/ day), and they even increased their LTPA level over time (mean 8.9 MET hours/day). The corresponding highlevel LTPA profile in males was Profile 4 (n=320, 17%), High fluctuating high. These males consistently kept their mean MET hours/day over 7.4, despite some fluctuation over time. Longitudinal LTPA profile associations withdemographic andanthropometric characteristics The results of the comparisons of demographic, anthropometric and lifestyle characteristics between longitudinal LTPA profiles in females and males are presented in Figs.2, 3, 4 and Supplementary Tables3 and 5. Although we detected no significant age or income differences between the five longitudinal LTPA profiles in either females or males, educational differences between profile memberships were found. Females in the Very low increasing low profile were significantly more likely to have lower education relative to those in the Low increasing moderate (p-value=0.007), Moderate stable (p=0.001), and Low stable profiles (p=0.019) at age 24, as well as to all other profiles at mean age 30. Males in the Very low stable profile reported significantly lower educational attainment than those in other profiles (p-values <0.001). The members of the Low stable very low and Low increasing moderate profiles also reported somewhat lower educational attainment than those who belonged to the High fluctuating high profile (p-values <0.011). For weight and BMI, significant differences were detected by longitudinal LTPA profiles throughout the follow-up in females (p<0.001): those females in the High increasing high profile had the lowest body weight and BMI at each time point, while those in the Very low increasing low and Low stable profiles consistently had the highest body weight and had the highest mean BMI. The same pattern occurred in males, with a significantly higher body weight and BMI in the two lowest LTPA profiles, Low stable very low and Very low stable, compared to the highest LTPA profiles, High fluctuating high and Low increasing moderate (p-values <0.027). Moreover, waist circumference measured at the last follow-up time point was smaller among those females who belonged to the High increasing high and Low increasing moderate profiles (p-values <0.001). Males in the High fluctuating high profile also had a significantly smaller mean waist circumference at the last follow-up time point compared to other profiles (p-values <0.003). These anthropometric differences between longitudinal LTPA profiles remained significant in females after Bonferroni correction (Supplementary Table3). Longitudinal LTPA profile associations withlifestyle characteristics Regarding lifestyle characteristics, females and males in different longitudinal LTPA profiles had similar alcohol consumption habits within their own sex groups, with the exception of age 40 when males in the Low increasing moderate profile used less alcohol than male individuals in the Low stable very low (p=0.010) and Very low stable profiles (p=0.035). Females in the Moderate stable profile more likely had never smoked compared to those females in profiles who started off with low LTPA at age 24 (p-values <0.047), but no differences appeared at age 30 (Fig.4). In females at age 40 and 60, the Very low increasing low profile had the most current female smokers, while the Moderate stable profile had the fewest (p<0.001 at age 40 and p=0.005 at age 60). In terms of male smoking, there were significantly more males in the High fluctuating high profile who were never smokers and fewer current smokers compared to other profiles at all time points (p-values <0.001 and <0.010, respectively). By age 60, the High fluctuating high profile had also the lowest proportion of former male smokers (p-values <0.024). Differences in sleep did not emerge between longitudinal LTPA profiles in males, whereas females in the Very low increasing low profile slept significantly more per night than females in all other profiles (p=0.022) at age 30. We did not find any differences in work-related PA levels between the longitudinal LTPA profiles but sedentary behavior (i.e., sitting) differences at age 60 existed. Females in the Low stable profile sat significantly longer per day than those in Low increasing moderate (p=0.018) and High increasing high profiles (p=0.001), and females in the Very low increasing low profile sat longer than Page 7 of 15 Berntzenetal. Int J Behav Nutr Phys Act (2024) 21:47 Fig. 2 Mean body mass index, percentages of individuals with high blood pressure and percentages of individuals with type 2 diabetes from young adulthood to late midlife by longitudinal leisure-time physical activity profiles. Left panels (F) represent females and right panels (M) males. BMI=body mass index; kg=kilogram; m=meter; T2D=type 2 diabetes Page 8 of 15 Berntzenetal. Int J Behav Nutr Phys Act (2024) 21:47 Fig. 3 Mean waist circumference, mean sedentary time and percentages of individuals with poor subjective health in late midlife (i.e., the last follow-up at age 60) by longitudinal leisure-time physical activity profiles. Left panels (F) represent females and right panels (M) males. cm=centimeter; h=hours Page 15 of 15 Berntzenetal. Int J Behav Nutr Phys Act (2024) 21:47 52. Park JH, Moon JH, Kim HJ, Kong MH, Oh YH. Sedentary lifestyle: overview of updated evidence of potential health risks. Korean J Fam Med. 2020;41(6):365–73. 53. Barone Gibbs B, Sternfeld B, Whitaker KM, Brach JS, Hergenroeder AL, Jacobs DR Jr, Reis JP, Sidney S, White D, Pettee Gabriel K. Bidirectional associations of accelerometer-derived physical activity and stationary behavior with self-reported mental and physical health during midlife. Int J Behav Nutr Phys Act. 2021;18:74. https:// doi. org/ 10. 1186/ s1296602101145-4. 54. Lee S. Cross-lagged associations between physical activity, self-rated health, and psychological resilience among older American adults: a 3-wave study. J Phys Act Health. 2023;20:625–32. 55. Pontzer H. Exercise is essential for health but a poor tool for weight loss: a reply to Allison and colleagues. Int J Obes. 2023;47:98–9. 56. Werneck AO, Oyeyemi AL, Szwarcwald CL, Sardinha LB, Silva DR. Body mass index trajectories and noncommunicable diseases in women: the role of leisure time physical activity. Am J Hum Biol. 2021;33(3): e23492. https:// doi. org/ 10. 1002/ ajhb. 23492. 57. Smith AD, Crippa A, Woodcock J, Brage S. Physical activity and incident type 2 diabetes mellitus: a systematic review and dose-response metaanalysis of prospective cohort studies. Diabetologia. 2016;59(12):2527–45. 58. Teychenne M, Ball K, Salmon J. Physical activity and likelihood of depression in adults: a review. Prev Med. 2008;46(5):397–411. 59. Pearce M, Garcia L, Abbas A, Strain T, Schuch FB, Golubic R, et al. Association between physical activity and risk of depression: a systematic review and meta-analysis. JAMA Psychiatry. 2022;79(6):550–9. 60. Kim H, Jeong W, Kwon J, Kim Y, Jang SI, Park EC. Sex differences in type of exercise associated with depression in South Korean adults. Sci Rep. 2020;10:18271. https:// doi. org/ 10. 1038/ s4159802075389-8. 61. Wennman H, Borodulin K. Associations between physical activity types and reaching the physical activity guidelines: The FinHealth 2017 Study. Scand J Med Sci Sports. 2021;31:418–26. 62. Official Statistics of Finland (OSF): Time use 2009, Cultural and sports hobbies 1981-2009. Helsinki: Statistics Finland; 2009. 63. Borodulin K, Laatikainen T, Juolevi A, Jousilahti P. Thirty-year trends of physical activity in relation to age, calendar time and birth cohort in Finnish adults. Eur J Public Health. 2008;18(3):339–44. 64. Engberg E, Alen M, Kukkonen-Harjula K, Peltonen JE, Tikkanen HO, Pekkarinen H. Life events and change in leisure time physical activity: a systematic review. Sports Med. 2012;42(5):433–47. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.