Objectives To determine the association between moderate-to-vigorous intensity physical activity (MVPA) trajectories (course over age and time) through the adult life course and onset of metabolic disease (diabetes and dyslipidaemia).
Methods We analysed prospective community-based cohort data of 5115 participants in the Coronary Artery Risk Development in Young Adults study, who were black and white men and women aged 18–30 years at baseline (1985–1986) at four urban sites, collected through 30 years of follow-up. Individualised MVPA trajectories were developed for each participant using linear mixed models.
Results Lower estimated MVPA score at age 18 was associated with a 12% (95% CI 6% to 18%) higher odds of incident diabetes, a 4% (95% CI 1% to 7%) higher odds of incident low high-density lipoprotein (HDL) and a 6% (95% CI 2% to 11%) higher odds of incident high triglycerides. Each additional annual 1-unit reduction in the MVPA score was associated with a 6% (95% CI 4% to 9%) higher annual odds of diabetes incidence and a 4% (95% CI 2% to 6%) higher annual odds of high triglyceride incidence. Analysing various MVPA trajectory groups, participants who were in the most active group at age 18 (over 300 min/week), but with sharp declines in midlife, had higher odds of high low-density lipoprotein and low HDL incidence, compared with those in the most active group at age 18 with subsequent gains.
Conclusion Given recent trends in declining MVPA across the life course and associated metabolic disease risk, young adulthood is an important time period for interventions to increase and begin the maintenance of MVPA.
- physical activity
Data availability statement
Data may be obtained from a third party and are not publicly available. Data are available upon request and data use agreement with the CARDIA Study (https://www.cardia.dopm.uab.edu/).
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Metabolic disease, including type 2 diabetes and dyslipidaemia (low high-density lipoprotein cholesterol (HDL-C), high low-density lipoprotein cholesterol (LDL-C) and high triglycerides (TG)) are established risk factors for cardiovascular disease (CVD),1–3 the leading cause of mortality in the USA.4 Despite our understanding of the general benefits of moderate-to-vigorous intensity physical activity (MVPA) on preventing type 2 diabetes5 and dyslipidaemia,6–8 there is a paucity of longitudinal data regarding the specific trajectories (the course over age and time) of MVPA during young adulthood and the association of MVPA trajectory groups with adult-onset CVD. In particular, there is a need to understand how MVPA in young adulthood affects the incidence of metabolic disease.
Young adulthood may set the baseline for later life physical activity trajectories and, thus, be an important time window for intervention.9 10 The 2018 US Department of Health and Human Services (HHS) Physical Activity Guidelines recommend a minimum of 150 min of moderate-intensity physical activity per week for adults 18–65 years.11 12 The Physical Activity Guidelines Scientific Report noted that young adults have unique growth and developmental needs similar to adolescents, who are recommended to have 60 min/day (420 min/week) of MVPA.11 However, there was insufficient literature on physical activity and health outcomes in young adulthood to confirm current guidance or to support a change to the current approach.11 Thus, the optimal dose and trajectory patterns of MVPA, particularly in young adulthood, to prevent metabolic disease remains unknown.13 14
The objective of this study was to determine the independent associations between the young adult level of MVPA and subsequent changes in MVPA through the transition to midlife and incidence of metabolic disease (diabetes, high LDL-C, low HDL-C, high TG, dyslipidaemia). Second, we examined if specific MVPA trajectory patterns were associated with metabolic disease onset.
The Coronary Artery Risk Development in Young Adults (CARDIA) study is a prospective cohort study that recruited black and white young adults at baseline from 1985 to 1986. Participants (N=5115) were recruited from four urban locations (Birmingham, AL; Chicago, IL; Minneapolis, MN and Oakland, CA) and have been followed up for more than 30 years (years 2, 5, 7, 10, 15, 20, 25, 30 with 90%, 86%, 81%, 77%, 74%, 72%, 72% and 71% retention, respectively). The cohort of participants was designed to be diverse in approximately equal parts by sex, race (black and white), age (18–24 years and 25–30 years at baseline) and educational level (high school or less or higher than high school) within each centre. After conducting the baseline examination, one participant requested to be excluded from all further analyses. Further details about the study design have been previously published.15
Self-reported MVPA was assessed by the interviewer-administered CARDIA Physical Activity History Questionnaire at each of the nine examinations.16 17 Participants were asked about the frequency of participation in 13 different activity categories (8 of vigorous and 5 of moderate intensity) within the leisure time and occupational physical activity domains over the prior 12 months. Each activity’s intensity was expressed as metabolic equivalents of task (METs), where one MET is defined as the energy used at rest (approximately an oxygen consumption of 3.5 mL/1 kg of body weight/min). Vigorous intensity activities (≥6 METs) included running, racquet sports, bicycling faster than 10 miles/hour, swimming, vigorous exercise classes, sports (eg, basketball, football), heavy lifting, carrying or digging on the job and home activities, such as snow shovelling and lifting heavy objects. Moderate intensity activities (3–5 METs) included non-strenuous sports (eg, softball), walking, bowling/golf, home maintenance (eg, gardening, raking) and callisthenics.18 Each activity was assigned a frequency based on whether it was performed for ≥1 hour or during any 1 month in the past year, the number of months it was performed at that level and the number of months it was performed on a frequent basis. Intensity scores (3–8 METs) and duration thresholds (2–5 hours/week) were assigned to each activity; activities above these levels of participation were considered frequent.17 An MVPA score was computed by multiplying the frequency (number of months) of participation by the intensity (METs) of the activity with a weighting factor for the months of more frequent participation.19 The MVPA score was the sum of all activities expressed in exercise units (EUs). For reference, an MVPA score of 300 EU estimates the HHS recommendations of approximately 150 min of moderate-intensity activity per week.12 20 Given recent evidence that occupational physical activity does not improve health and may be detrimental, constituting an occupational paradox,21 we excluded the occupational physical activity question (lifting, carrying or digging on the job) from the MVPA score. Convergent validity of the CARDIA Physical Activity History Questionnaire has been established using report-based measures, including physical activity diaries and detailed quantitative recall questionnaires16 19 22 and accelerometers.22–24 It has also been indirectly validated by showing expected relations with physical fitness and measures of body fat,18 19 25 and has demonstrated adequate test–retest reliability.19
Blood was drawn and processed at the central laboratory according to standard procedures at each of the nine CARDIA examinations. Glucose was assayed using the hexokinase method. Diabetes was defined as a fasting glucose ≥126 mg/dL or on diabetic medications but not pregnant for examinations before May 2011. 15 17 Diabetes was defined as fasting glucose ≥126 mg/dL, 2-hour glucose tolerance test ≥200 mg/dL, haemoglobin A1c ≥6.5%, or being on diabetic medications but not pregnant for examinations after May 2011.
Fasting lipid measures were measured at each of the nine examinations. Total cholesterol was measured enzymatically and defined as high if levels were ≥240 mg/dL.15 TG were measured enzymatically and defined as high if levels were ≥200 mg/dL.15 HDL-C was determined after precipitation with dextran sulfate-magnesium chloride and defined as low if levels were <35 mg/dL for males or <45 mg/dL for females.26 LDL-C was calculated using the Friedewald equation and defined as high if levels were ≥160 mg/dL.27 Dyslipidaemia was defined as TG ≥150 mg/dL or HDL <35 mg/dL for males, or TG ≥150 mg/dL or HDL <45 mg/dL for females.
Age (years), race (black or white), sex (male or female), smoking status (never, former or current smoker), alcohol use (mL of alcohol consumed/day), educational attainment (the highest grade of school completed), family history of diabetes or CVD (yes or no), medical history and medications were reported through a questionnaire. The use of diabetes or dyslipidaemia medications was assessed by self-report at each examination. Body mass index (BMI) was calculated based on measured height and weight at each examination.
Summarising physical activity
MVPA trajectories were modelled among all CARDIA participants. We developed a linear mixed model (LMM) for repeated measures of MVPA in order to generate succinct summaries of exercise patterns over time. The MVPA slopes use all observations of the MVPA scores prior to metabolic disease onset in order to use as much of the data for each participant as possible and to stabilise the best linear unbiased predictions. The LMM included fixed effects for a four-level categorisation of sex and race, with age as continuous, and their interactions, as well as random effects for participant and age, with unstructured covariance. Our inclusion of these covariates, along with observed outcomes, makes the LMM assumption of missingness at random more plausible, though this assumption is not ultimately verifiable. From the fixed and random effects estimates provided by this model, we calculated the expected MVPA level at age 18 and annual change for each participant. For ease of interpretation, we changed the sign of both summaries, so as to capture the associations of lower level and faster decline in MVPA with increased metabolic risk.
Modelling the association of lower MVPA with incident metabolic disease
Unadjusted cumulative incidence of metabolic diseases (diabetes, high LDL-C, low HDL-C, high TG, or dyslipidaemia) by sex and race/ethnicity were estimated using Kaplan-Meier methods. The data for each participant were then expanded to include a record for each age between study entry and either metabolic disease onset, which was assumed to occur at the first visit at which it was detected, or at censoring by the end of the study of loss to follow-up. Pooled logistic models were used to estimate the independent associations of the expected MVPA at age 18 and subsequent annual change with onset of metabolic disease, adjusting for potential confounders, including sex, race, family history of diabetes or CVD, years of education, smoking status, alcohol use and BMI (smoking status, alcohol use and BMI were time varying, with the last observation carried forward), which have been adjusted for in prior analyses of physical activity and CVD risk (directed acyclic graphs are shown in online supplemental appendix 1).17 28 We scaled the estimated MVPA score at age 18 from high to low per 100 EUs, corresponding to 0.45 SDs, for ease of interpretation. We kept annual reduction in total MVPA per 1 EU, corresponding to 0.23 SDs. We tested if BMI category or sex and race modified the effect of MVPA (level and change) on incident metabolic disease. Pooled logistic models estimated the associations of meeting various MVPA thresholds at age 18 (<150, 150–300, 300–600, >600 EU) in combination with annual change categories (gain, loss of <2.5 EU/year, of loss >2.5 EU/year) and onset of metabolic disease, adjusting for confounders (sample sizes shown in online supplemental appendix A). We used Stata V.16.0 (StataCorp) for all analyses.
Table 1 shows the baseline demographic and health characteristics of 5114 participants included in the sample. The sample was 51.6% black and 45.5% male. Demographic and health characteristics of the sample at baseline are shown in table 1. Average MVPA declines from young adulthood in all race and sex groups, particularly in black men (figure 1). Incidence of diabetes and cholesterol outcomes by race and sex are presented in online supplemental figures A–E.
Table 2 shows pooled logistic regression model estimates for the associations of the two MVPA summaries (estimated MVPA level at age 18 and subsequent declines in MVPA) with metabolic disease onset. Model 2.1 adjusted for age only, whereas model 2.2 adjusted for age, race, sex, education, family history, smoking status, alcohol and BMI. In the fully adjusted model (model 2.2), lower estimated MVPA score (per 100 units) at age 18 was associated with a 12% (95% CI 6% to 18%) higher odds of incident diabetes, a 4% (95% CI 1% to 7%) higher odds of incident low HDL, a 6% (95% CI 2% to 11%) higher odds of incident high TG and a 3% (95% CI 0% to 6%) higher odds of incident dyslipidaemia. Each additional annual 1-unit reduction in the MVPA score was associated with a 6% (95% CI 4% to 9%) higher annual odds of diabetes incidence, a 4% (95% CI 2% to 6%) higher annual odds of high triglyceride incidence and a 2% (95% CI 0% to 3%) higher annual odds of dyslipidaemia incidence. Pooled logistic regression model estimates stratified by BMI category (online supplemental appendix B) and race and sex (online supplemental appendix C) are shown in online supplemental appendix 1.
Associations between various MVPA thresholds at age 18 combined with categories of subsequent annual change in MVPA and onset of metabolic disease are shown in table 3. In fully adjusted models treating MVPA as additive (model 3.1), reductions in MVPA in midlife (compared with gains in MVPA) were associated with onset of all metabolic disease outcomes for any given MVPA threshold level at age 18. In fully adjusted models allowing MVPA categories to interact (model 3.2), certain MVPA combinations are notable. For instance, among participants with over twice the minimum recommended MVPA level (>600 EU) at age 18, those with steep subsequent losses (>2.5 EU/year) had 4.74 higher odds (95% CI 1.55 to 14.50) of low HDL incidence and 3.22 higher odds (95% CI 1.23 to 8.47) of high LDL than those with gains in MVPA. In addition, participants with 300–600 EU at age 18 and subsequent gains in MVPA did not have higher odds of most metabolic disease onset compared with participants with >600 EU at age 18 and subsequent gains in MVPA.
In this prospective cohort study with 30 years of follow-up, we found that a high level of MVPA in young adulthood is a critical starting point for maintaining lifetime metabolic health. Young adult MVPA is associated with lower incidence of diabetes, low HDL-C and high TG, independent of MVPA levels across later adulthood. Maintaining high levels of MVPA in the adult life course is also important; for any given young adult MVPA set point, decline in MVPA through the adult life course is also associated with incident diabetes and high TG.
These findings add to prior literature on physical activity and metabolic disease5–8 by leveraging a large longitudinal cohort with 30 years of follow-up data to develop MVPA trajectories throughout the life course. Using these individualised trajectories, we find independent associations with the young adult level of MVPA and subsequent declines in the later adult level of MVPA with diabetes and high TG. It is notable that both estimated MVPA level and slope were independently associated with diabetes and triglyceride onset, even after adjusting for a number of potential confounders.
Our findings indicate that young adult MVPA levels provide protection from subsequent metabolic disease, independent of MVPA levels up through midlife. Thus, young adulthood is an important period for intervention to ensure adequate MVPA levels. Furthermore, our findings indicate that protective levels of activity are higher than the currently recommended minimum. This is an important finding since the HHS Physical Activity Guidelines Scientific Committee Report noted there was insufficient literature on physical activity to inform guidelines in young adults.11 Physical activity typically declines in the transition from adolescence to young adulthood due to educational, economic and social transitions.9 10 29 For instance, young adults may have fewer opportunities for team or organised sports when they transition to the workforce or college, compared with adolescents, who have physical activity requirements and more opportunities for organised team sports in school.30 The transition to parenthood may also displace leisure time for physical activity.31
The MVPA trajectory analysis also identified notable MVPA patterns by race and sex through the life course. For instance, black women have the lowest MVPA levels through the adult life course. Although black men start with high average levels of MVPA in young adulthood, their levels persistently decline throughout the adult life course, similar to findings from the National Health and Nutrition Examination Surveys (NHANES).32 Physical activity interventions and messaging may particularly focus on black women through adulthood and preventing declines in black men. We also found that black adults have higher diabetes incidence throughout the life course compared with white adults, similar to findings in NHANES.4 33
Clinical and public health implications
We find that reductions in MVPA during midlife are associated with an increased incidence of metabolic disease across several outcomes. Public health campaigns and clinicians should focus messaging on maintaining adequate levels of MVPA and preventing declines throughout the adult life course. It is also noteworthy that young adults who were in the most active group at age 18 (over 300 min/week), but had sharp declines in MVPA in midlife, had higher odds of high LDL and low HDL incidence, compared with those in the most active group at age 18 and subsequent gains. Young adults may be more willing to take on immediate risk (eg, not being physically active) if they perceive the outcome is too far into the future, a concept referred to as temporal discounting.34 Young adults may respond to messaging that promotes optimising health beyond just the need to avoid risk.35
Limitations and strengths
Limitations and strengths of this study should be noted. While we adjusted for several potential confounders including age, race, sex, education, family history, smoking status, alcohol and BMI (and behavioural and BMI covariates were time-varying), there is the possibility of unmeasured confounders, such as neighbourhood or genetic factors.36 In addition, mixed models can produce biased effect estimates in the presence of time-varying confounding affected by prior exposure, which would require causal methods to adjust for this potential bias, such as inverse probability weighting.37 There may be attenuation bias due to measurement error in MVPA and residual confounding due to measurement error in some confounders such as smoking and alcohol. There was a possibility of selection bias due to censoring, including losses-to-follow-up and competing risks. MVPA was measured simultaneously with other proposed confounders in one visit, which may lead to over-adjustment bias (adjustment for mediators rather than confounders). MVPA was based on self-report and may be subject to information and prevarication bias, and it did not collect information regarding activity intensity. Nonetheless, the same questionnaire was used across the 30-year follow-up period, which is a distinct strength. Because we scaled the estimated MVPA score at age 18 per 100 EUs for ease of interpretation, the ORs for the two MVPA estimates (level at age 18 and annual reduction) are not standardised and should be interpreted in different scales. The sampling design of CARDIA was not representative of all races or ethnicities in the USA, which may limit generalisability; however, the study specifically focused on participants identifying as black or white race.15 Given the larger proportion of the sample with metabolic disease by age 60, the number of eligible participants in the analysis drops with age.
In conclusion, MVPA level in young adulthood and declines in later adulthood are each significantly and independently associated with later life metabolic disease onset. Public health and clinical programmes should emphasise, prioritise and develop interventions to promote MVPA in young adulthood, the time when individuals establish an MVPA set point. Future research could examine the mechanisms and mediators by which MVPA may be related to metabolic disease, such as insulin sensitivity. Regardless of young adult MVPA level, interventions to sustain or increase MVPA across adulthood remain another priority for lifetime metabolic health.
What are the findings?
In this prospective longitudinal study with regular 30-year follow-up, we found that low young adult moderate-to-vigorous intensity physical activity (MVPA) was associated with higher odds of diabetes, high triglycerides and low HDL incidence.
Reductions in MVPA during midlife are associated with an increased incidence of metabolic disease.
How might it impact on clinical practice in the future?
Given recent trends in declining MVPA across the life course and associated metabolic disease risk, young adulthood and midlife are important time periods for interventions to increase and maintain MVPA.
Data availability statement
Data may be obtained from a third party and are not publicly available. Data are available upon request and data use agreement with the CARDIA Study (https://www.cardia.dopm.uab.edu/).
Patient consent for publication
The institutional review boards at each study site approved all study procedures. All participants provided written informed consent. The University of California, San Francisco institutional review board deemed this secondary data analysis project exempt (IRB 18-26523).
The authors thank Samuel E Benabou for editorial assistance. An abstract corresponding to this manuscript was presented at the Society for Adolescent Health and Medicine Annual Meeting on 10 March 2021.
Contributors JMN conducted the literature search, interpreted findings, and wrote, revised, and edited the manuscript. EV conceptualised the study, analysed data and edited the manuscript. KPG, SS, AKG, KB-D conceptualised the study, interpreted the data and provided critical revisions on the manuscript. AEM, JSR and JPR provided critical revisions on the manuscript. All authors approved the final draft.
Funding JMN is supported by the National Heart, Lung, and Blood Institute (K08HL159350) and the American Heart Association (CDA34760281). KB-D is supported by the National Institutes of Health (K24DK103992). The funders had no role in the study design. The Coronary Artery Risk Development in Young Adults Study (CARDIA) is conducted and supported by the National Heart, Lung, and Blood Institute in collaboration with the University of Alabama at Birmingham (HHSN268201800005I and HHSN268201800007I), Northwestern University (HHSN268201800003I), the University of Minnesota (HHSN268201800006I) and the Kaiser Foundation Research Institute (HHSN268201800004I). This manuscript has been reviewed by CARDIA for scientific content.
Disclaimer The views expressed in this manuscript are those of the authors and do not necessarily represent the views of the National Heart, Lung, and Blood Institute; the National Institutes of Health or the US Department of Health and Human Services.
Competing interests The authors have no conflicts of interest to report.
Patient and public involvement statement Patients or the public were not involved in the design, conduct, reporting or dissemination plans of our research.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.