The association of adolescent fitness with cardiometabolic diseases in late adulthood: A 45‐year longitudinal study

The aim of this study was to examine the associations of adolescent cardiorespiratory fitness (CRF), muscular fitness (MF), and speed‐agility fitness (SA) with middle‐aged cardiometabolic disease risk and explore sex differences.


| INTRODUCTION
2][3][4][5][6] It can be divided into components of cardiorespiratory fitness (CRF), musculoskeletal fitness (MF), flexibility, and body composition 7 but does not include components such as speed and agility (SA), probably because only a few previous studies have investigated the associations of SA with health outcomes in the past.4 Results concerning these associations in the youth-adulthood transition are more uncertain.
CMD refers to conditions sharing the same risk factors: overweight, obesity, dyslipidemia, and elevated blood pressure (EBP).CMDs vary in terms of severity, beginning with insulin resistance progressing to CVD and type 2 DM (T2DM). 15,16Body composition measures -body mass index (BMI; kg/m 2 ), waist circumference (WC), and waist-to-hip circumference ratio -are strong risk factors for CMD.
9][20][21][22] However, there are also conflicting findings.23,24 Several studies indicate that higher childhood/adolescent CRF is associated with lower risk for CVD mortality 21 and morbidity, 19,22,25,26 T2DM, 18 MetS 27 and the risk factors such as high BP, 22,28 central obesity, 22 adverse glucose metabolism 20,22,28 and blood lipid profile 22,28 in young adulthood (21-27 years) and middle age (35-64 years).Whether these associations are stronger in males than in females, vice versa, or equal is not clear.Two studies reported similar associations in males and females, 20,22 whereas one study, although conducted using an indirect CRF assessment, indicated a stronger association for females. 28 Th largest and longest duration cohorts 18,19,21,25 included only male participants, underlining the need for long-duration follow-ups involving males and females.The results have remained concordant regardless of the study design, CRF measurement method (indirect VO 2 max or field test), the baseline age (7-18 years), and the length of follow-up (6-46 years) although the association tended to wane as the follow-up time is extended. 23,27The evidence concerning youth MF's association with adulthood cardiometabolic health is more uncertain, even though MF has been shown to be associated with lower CMD morbidity and mortality 9,29 in adulthood.Longitudinal inverse associations have been reported between adolescent MF and adult chronic disability due to CVD, 26 risk for premature death due to CVD, 30 incidence of T2DM, 18 strokes, 19 and risk factors for CMD.20,[31][32][33] However, two of the studies 32,33 reported conflicting findings and three 18,19,26 found adolescent MF to be less influential on future CVD and T2DM than CRF is.No association for adulthood ischemic heart disease was detected in Crump et al. 25 MF in these studies has been ascertained with relative 31,32 and absolute strength tests.[18][19][20] Despite evidence of the association between adolescent fitness and adulthood metabolic health, studies with an objective baseline fitness measurement involving males and females in their late middle age are scarce. Moeover, as previous literature has focused on CRF and MF as the main fitness components, studies concerning SA's role in subsequent health are lacking. To our knowedge, this study is the first to follow its participants in both sexes from adolescence up to age 57-64 years in terms of CRF, MF, and SA and their association with health. Th7 Additionally, extended longitudinal investigation is justifiable as the prevalence of cardiometabolic conditions increases along with age.30,31 2 | METHODS

| Study population
This study is based on the LISE follow-up project launched in April-May 1976, and it includes three measurements: the baseline in 1976 including objectively measured fitness tests and a physical activity (PA) questionnaire; the first follow-up in 2001 including a postal questionnaire; the second follow-up in 2021 including a postal questionnaire and WC self-measurement.At baseline, we used a stratified random sampling, recruiting 56 schools from eastern, western, central and northern Finland, including urban and rural districts.Classes at school were selected randomly and the pupils were drawn either in alphabetical order or by selecting every second or third.The original baseline sample consisted of 9-21-year-old children and adolescents, of which 12-19-year-old adolescents (mean age 14.9) with eligible PF test data were qualified for the final sample (N max = 1794).The first follow-up in 2001 consisted of 1196 (67% of the baseline) participants, then aged 37-44 years, providing eligible baseline PF test data and the first follow-up CMD assessment data.The second follow-up in 2021 consisted of 768 participants (43% of the baseline), then aged 57-64 years, with eligible data from the baseline PF tests and the second follow-up CMD assessment.LISE-project study population and data acquisition in detail are presented in Laakso et al. 36 Participants with eligible baseline PF test data and at least one of the follow-ups qualified for the longitudinal analyses and the exact number of participants in each analysis varied from 341 to 1358 depending on the association studied, that is, available data for each pair of variables examined.The variation was due to the uneven number of test results at baseline, responses per question in the follow-up questionnaires, and the overall number of participants between the two follow-ups.A flow chart of the data structure is presented in Figure 1, and the exact N of each analysis is in Tables 3-5 and Tables S3-S5.

| Assessment of physical fitness
The fitness tests, based on International Fitness Test guidelines, 37 in 56 schools in April-May 1976 were conducted by a trained research team using a consistent measurement protocol.All participants performed an identical warm-up routine and their height (cm) and weight (kg) were measured objectively by a school health care professional.CRF was measured with a 2000-/1500-m run; MF with standing broad jump, sit-ups, and pull-ups/flexed arm hang; and SA with a 50-m dash and a 4 × 10-m shuttle run.Cronbach's alpha for MF was 0.709 and for SA 0.689.For detailed descriptions of the tests, see Table S1.The test scores were subsequently adjusted for age and sex by calculating z-scores for each age group in both sexes separately.

F I G U R E 1
The data structure of the longitudinal analyses.BMI, body mass index; CHD, coronary heart disease; CMDS, cardiometabolic disease score; CRF, cardiorespiratory fitness; DM, diabetes mellitus; EBP, elevated blood pressure; MF, muscular fitness; SA, speed-agility; WC, waist circumference.

| Assessment of physical activity
PA was self-reported by the participants at the second follow-up in March 2021.To eliminate the impact of COVID-19 and follow social contact restrictions, the participants were asked to assess their PA behavior concerning time before the pandemic.MET-hours/day value was calculated from the reported frequency, duration, and intensity of PA.Detailed information on the PA questionnaire used is presented elsewhere. 36 5) WC (cm).Indicators 1-3 were determined by a self-reported (physician-diagnosed) structured postal follow-up questionnaire in 2001 and 2021, including identical questions.The questions were designed for the LISE project but derived mainly from a FinHealth survey. 38The health conditions (1-3) were asked in the following way: "Have you ever been diagnosed with a [condition] which has been ascertained by a physician?"(yes/no).For EBP, participants reporting "ascertained to be >140/90 mmHg" qualified and those reporting "not measured"," not sure if measured" and "ascertained to be <140/90 mmHg" were removed.BMI was computed from self-reported height (cm) and weight (kg).To investigate the incidence of cardiometabolic conditions in middle age, data from the first and the second follow-up was amalgamated regarding the identically measured indicators 1-3 (Figure 1).Finally, a cardiometabolic disease score variable (CMDS) with a scale of 0-3 was calculated as a sum of EBP, CHD, and DM diagnoses to investigate the accumulated burden of the conditions.WC measurement was conducted by a self-measurement in 2021 alongside the postal questionnaire.A measurement tape, with pictorial and verbal instructions (Table S2), was sent to the participants to conduct the measurement.

| Statistical analyses
Analyses were conducted with IBM SPSS (version 28.0.0.0).Continuous composite variables for SA were computed from the 50-m dash and the shuttle run z-scores, and for MF from the broad jump, sit-ups and pull-ups/flexed arm hang z-scores.Z-scores from the 1500-/2000-m run were used as such as a continuous CRF variable.Categorized variables were formed by dividing the continuous variable into tertiles: low, average, and high.
To investigate the associations of continuous and categorized CRF, MF, and SA with continuous CMDS, WC, and BMI, a general linear model (GLM) with adjustment for baseline age and BMI was conducted.GLM parameter estimates were bootstrapped to increase confidence in data representativeness.To investigate the associations of continuous and categorized CRF, MF, and SA with dichotomic CMDS, DM, EBP, and CHD, logistic regression analysis with adjustment for baseline age and BMI was used.Odds ratios with 95% confidence intervals (CI) were calculated to indicate risk differences for CMD indicators in middle age according to adolescent fitness.In addition, relative risks were calculated for analyses using categorized exposures and dichotomic outcomes.Interaction effect with sex was tested in all analyses.In the exploratory analyses, we tested the effect of additional adjustment for PA at the end point (2021).

| RESULTS
The descriptive statistics of the participants involved in the between-measurement analyses are presented in Table 1.Sex-stratified prevalence of CMD in participants involved in between-measurement analyses reported at either the follow-up 1 or 2, is presented in Table 1.
Baseline CRF, MF, and SA z-scores between dropouts and non-dropouts were evaluated using Little's MCAR test.The test indicated that the follow-up 1 non-dropouts had healthier characteristics than dropouts.Such a difference in baseline z-scores was not seen between the follow-up 2 non-dropouts and dropouts (Table 2).

| Cardiorespiratory fitness
The result of GLM, based on continuous CRF variable (Table 3), showed a significant (p = 0.006) inverse BMIindependent association between adolescent CRF and middle-aged CMDS, indicating that a 1 SD increase in the CRF test z-score in adolescence was associated with 0.1 (scale 0-3) lower CMDS in middle age.When GLM was repeated with the categorized CRF variable, the lowest CRF tertile differentiated significantly (p = 0.021) from the highest in terms of CMDS (Table S3).In analysis using CRF and dichotomic CMDS, the baseline age and BMI-adjusted relative risk for one or more cardiometabolic conditions was significantly lower on average (RR = 0.85, 95% CI [0.71, 0.99]) and high (RR = 0.76, 95% CI [0.62, 0.90]) CRF tertile compared to low (Table S3).GLM with continuous CRF (Table 3) showed a significant interaction (p = 0.031) with sex, indicating that the association between adolescence CRF and middle-aged CMDS was stronger among females.This was confirmed in further sex-specific analyses, based on continuous (Table 3) and dichotomic (Table 3) CMDS variables.For example, logistic regression analysis based on the dichotomic CMDS variable showed that in females a 1 SD increase in the CRF test score in adolescence significantly lowered the odds (OR = 0.71, 95% CI [0.56, 0.91]) for one or more cardiometabolic condition in middle-aged (Figure 2).
When the health conditions were studied separately with logistic regression, a significant reduction (OR = 0.72, 95% CI [0.56, 0.91]) in odds for middle-aged EBP by a 1 SD increase in the adolescent CRF test score independent of BMI was detected (Figure 2) among females.Also, a significant association among females was found between adolescent CRF and middle-aged CHD, but due to a low number of cases (n = 2 in Model 2) this result was negligible.The associations between adolescent CRF and middle-aged DM and late middle-aged WC or BMI were not statistically significant in any of the analyses.

| Muscular fitness
The results showed an inverse association between adolescent muscular fitness (MF) and late middle-aged WC and BMI.However, the association was strongly attenuated after controlling for baseline BMI.In GLM with a continuous MF variable (Table 4), a significant inverse association with late middle-aged WC (p = 0.049) and BMI (p < 0.001) in the age and sex-adjusted model was found.In the BMI-adjusted model, the trend for inverse association remained for both outcomes, but did not reach statistical significance.No interactions between MF and sex were detected in associations for WC or BMI.MF in adolescence was not associated with CMDS, separately tested health conditions, or dichotomic CMDS in middle age regarding the two models, nor with males or females combined or separated (Table S4).

| Speed-agility
The association of adolescent SA with cardiometabolic health in middle age was equivocal.GLM for association between SA and WC, based on categorized SA variable, indicated interaction by sex leading to further sex-specific analyses with a categorized SA variable (Table S5).This analysis revealed a significant difference (p = 0.013) between low and high adolescence SA group males' WC in late middle age, persisting throughout the adjustment for   S5) or logistic regression analyses (Table 5; Table S5) in either model could establish a significant association between adolescent SA and middle-aged cardiometabolic health.

| Exploratory analyses
We tested whether the significant associations outlined above persisted after additional adjustment for PA at the end point (Table S6).Overall, the findings were consistent and CRF was significantly associated with CMD, even after adjustment for PA at end point.

| DISCUSSION
In this long follow-up study, CRF in adolescence was inversely associated with the accumulated burden of CMD in late adulthood.MF and SA in adolescence, in turn, were only weakly associated with WC and BMI after 45 years, and not consistently by sex and after adjustment for baseline BMI.The inverse association between CRF and burden of CMD was independent of body mass and PA and was stronger in females than in males.
The finding that CRF in adolescence was associated with CMD in middle age independent of BMI is in line with previous studies [18][19][20]25 and highlights several explanatory behind the phenomenon. Firt, CMD is regulated by multiple factors in adulthood, of which CRF has been acknowledged as a substantial and independent one.39,40 Nevertheless, as adiposity remains a strong early  2).b Interaction CHD*sex: Model 1: OR = 0.41; 95% CI = 0.17, 0.96; p = 0.040; Model 2: OR = 0.28; 95% CI = 0.08, 0.97; p = 0.045 (Figure 2).predictor for 41,42 combining sufficient levels of CRF and low adiposity in youth seems to form a robust base for cardiometabolic health across the life course according to our findings and previous research.21,25 Second, the BMI-independent association possibly illustrates the earlier suggested 43 genetic regulation of maximal oxygen uptake in adolescence which becomes visible among fatbut-fit youth who tend to maintain normal aerobic capacity   and gain no additional risk for subsequent CMD 44 despite the risk of excess adiposity 45 Despite evidence establishing the inverse association, the mechanism through which sufficient CRF in youth elicits the lowered risk for CMD in adulthood is not clear.Gutin 46 proposed that this may be related to stem-cell differentiation in youth because of vigorous aerobic training.Another hypothesis favors CRF-associated metabolism, which is associated with lower CMD risk.47 Finally, we traced the inverse association between adolescent CRF and middle-aged CMD to originate mainly from the low CRF in females.The finding is important for PA promotion in youth while it also supplements the previous results concerning males, 19,25,26 demonstrating that the adolescents with the lowest CRF are at the highest risk for adulthood CMD.The analyses showed persistent longitudinal CRF-CMD association even after controlling for PA, which is suggesting that regardless of the activity levels performed in late adulthood (i.e., PA assessed at the end point, when participants were aged ~ 60 years), CRF in pediatric ages has a prognostic value for CMD. Heritbility of CRF has been reported to be roughly 50% 43 leaving room for environmental factors, including PA and exercise, to produce individual variation on CRF levels.Regardless of its genetic component, CRF is a modifiable factor and PA, and particularly that of high intensity is known to have an impact on CRF.48 In fact, previous evidence from twin studies supports that long-term leisure-time PA also during adulthood is beneficial for health even after controlling for other genetic factors.49 Therefore, this evidence together with the findings of this study suggests that enhancing CRF at pediatric ages could have long-lasting effects on future CMD.
Although we found associations between adolescent CRF and CMD, they were not strong.Presumably, this refers to the low overall prevalence of CMD in the study population, and on the other hand, the relatively low number of participants with data on adolescent CRF and middle-aged CMD.According to the literature, the low prevalence of CVD at the age of our study population is not uncommon, as the prevalence of CVD and DM in the population under 65 years remains relatively low compared to those over 65 years. 34,35econd, presumably the self-reports of CMD likely underestimated the prevalence because of undiagnosed cases, which in the case of DM can be substantial. 50Hence, implementing more sensitive methods, such as blood samples as in the reference studies, 20,32 to analyze dyslipidemia and adverse glucose profile would have revealed pre-existing conditions and affected the longitudinal associations.The finding that CRF was inversely associated with the risk for hypertension, and accordingly a higher accumulated burden of CMD particularly in females supports the finding of Hamer et al. 28 who reported a stronger link among females between childhood CRF and middle-aged cardiometabolic health.However, another long follow-up with males and females 22 could not identify sex differences.Given that CVD in middle age tends to develop earlier in males, 46 finding an explanation for the sex difference in literature, 24 and our data calls for further investigation.Finally, the large follow-up time explains the parsimonious associations and is endorsed by previous research. 23,27Long duration presumably highlights the multifactorial and slowly progressive nature of CMD.Nevertheless, we detected an association between CRF and CMD across a substantial period, which is an important result.For future research, we suggest following participants over the age of 65 years whereupon the prevalence of CMD is shown to increase. 34,35e found a sex-independent inverse association between adolescent MF and late middle-aged WC and BMI which, however, was dependent on baseline BMI.This finding is in line with previous long-duration follow-ups indicating MF in youth to be less influential for cardiometabolic health than in CRF. 25,28,29,32,33Given that such a strong genetic regulation as in CRF 43 affecting likely also later life cardiometabolic health 44,45 has not been established for MF, the absence of the individual longitudinal association is conceivable.Nevertheless, the association found (although BMI-dependent) seems to support the recent finding 29 suggesting the intertwined association between MF and body composition over the life course.The association of adolescent SA with middle-aged cardiometabolic health was inconsistent in the present study.Our results indicated no longitudinal association between SA and CMD.Nevertheless, we found an association between SA and late middle-aged WC in males.An explanation for this link is difficult to find as reference studies are completely missing.Diminutive cross-sectional evidence among children suggests a moderate association between SA and CMD risk.According to our results and the literature, 31,32 SA has similarities with MF as both seem to be dependent on body mass.Hence, in the SA-WC association found in this study, extracting the contribution of SA from BMI tracking is not straightforward.

| Strengths and limitations
This study has some limitations which should be acknowledged.First, cardiometabolic conditions were assessed with physician-diagnosed self-reports, which might introduce a source of error.Although evidence indicates satisfactory concurrent validity between selfreports and medical records in CVD and DM, 52 for EBP the lack of repeated measures comprised a possible threat to validity. 53Second, due to questionnaire-based cardiometabolic ascertainment in follow-ups, we were not able to conduct a time-adjusted survival analysis for the longitudinal associations.Third, the tests used for the baseline fitness measurement might have impacted the validity as field tests involve a risk for measurement error due to motivational issues.However, the criterion validity of field running tests to measure CRF has been shown to be good, 54 and measuring fitness in a rather large adolescent population other than by field tests would be difficult.Fourth, objective assessment of the fitness status in the follow-up measurements was not available which is a limitation considering the recent study of Kokkinos et al. 8 suggesting that change in CRF level over time moderates the longitudinal association between CRF and mortality.Fifth, skewness of the data in follow-up 1 (Table 2) might have impaired data representativeness.However, we could confirm that this did not change the main result as exploratory analysis (Table S6), conducted with not skewed follow-up 2 data only, showed identical results compared with the main analysis.Finally, no indicators of blood lipid profile were measured in this study.This is a limitation because low HDL, high LDL cholesterol, and triglycerides are risk factors for CMD 16 and MetS. 55 major strength of this study is that it involved both males and females which allowed to investigate the role of sex as a moderator in the associations between fitness components and CMD.Moreover, the 45-year follow-up design, based on a baseline random sampling across the whole country, connecting mid-adolescence with late adulthood (up to ~ 60 years of age) makes this study unique in the field.Another strength refers to the objective assessment of physical fitness, and the investigation of the speed-agility component in relation to future CMD, which was not available in previous studies in this field.Finally, in this study, we were able to explore the role of PA in the associations between physical fitness components and CMD as opposed to recent long cohorts. 18,19,21

| Conclusions
The findings from this prospective study support that low CRF, but not MF and SA fitness, in adolescence is associated with increased risk for CMD in middle age, and that this association is stronger in females than in males.Assessment of fitness, particularly CRF, in youth, is clinically informative about the risk of future CMD in midadulthood.Moreover, enhancing CRF in the first decades of life might be beneficial for future life.

| Perspective
In this study, adolescent objectively measured CRF inversely, and regardless of sex and BMI, associated with the accumulated burden of cardiometabolic conditions up to 45 years.In females, adolescent CRF was inversely associated with middle age elevated BP.Moreover, adolescent speed agility, a previously understudied component in prospective cohorts, was inversely linked with late middle age WC in males.The findings extend previous evidence 18,19,25 suggesting youth low objectively measured cardiorespiratory and MF inversely associates with adulthood cardiometabolic disease risk in males.Physical fitness, particularly CRF, in youth may be beneficial in preventing CMD across the adulthood.Assessment and promotion of physical fitness in youth is clinically informative.

AUTHOR CONTRIBUTIONS
Perttu T. T. Laakso designed the study and wrote the manuscript with support from Francisco B. Ortega, Pertti Huotari,

T A B L E 1 Follow
Characteristics of the participants at the different time points.

F I G U R E 2
Odds ratios for CMD indicators in middle age by 1 SD increase in adolescent mean CRF-score (sex-separated analyses adjusted for baseline age and BMI).Males = blue; females = red; y-axis = odds ratio; error bars = 95% confidence intervals.CHD, coronary heart disease; CMDS_di, dichotomized cardiometabolic disease score (DM, EBP, CHD); DM, diabetes mellitus; EBP, elevated blood pressure.T A B L E 4 MF in adolescence and cardiometabolic disease in middle age (unstandardized coefficients; all continuous variables are expressed as standard deviation increments).

2.4 | Ascertainment of cardiometabolic health
Missing value analysis of follow-up and non-followup participants.