Summary of Findings
To the best of our knowledge, this is the first study that examined both the independent and joint effects of BMI and MHS and their interactions on mortality in the same study population stratified into mid- and late-life. In this cohort of 6,252 individuals with measures taken in midlife, and 6,215 individuals with measures taken in late-life, being MU, is independently associated with an elevated risk of mortality, irrespective of BMI. Midlife and late-life overweight and obesityis associated with increased mortality risk only among those who were MU. Conversely, metabolically healthy overweight in late-life is associated with a reduced risk of mortality.
Independent effects of BMI and MHS in mid- and late-life
Fromtheindependent effect models, obesity, but not overweight, in both mid- and late-life was associated with increased mortality risk. The midlife results areconsistent with other large cohort studies [4, 5],but the positive associationbetween late-life obesity and mortality contrasts with many studies on older persons [2, 6-9]. Being MU in either age group was also associated with an elevated risk of mortality, aligning with meta-analyseson mortality risks associated withMetSin a population with a broad spectrum of ages  and older persons .While the direction of the independent effectsof high BMI and MU status on mortality wasbroadly similar in mid- and late-life, the magnitude of the effects fromobesity and being MU in midlife were greater than in late-life.These findings highlight the importance of initiating preventionand interventions to manage obesity and metabolic dysfunction early in adulthood due to its potential long-term impact on survival.
Comparing the independent and joint effects of BMI and MHS
The effects of MHS on mortality remained relatively stable in the joint effect models, whereas the effects of obesity on mortality were attenuated compared to the independent effect models.Therefore, MHS is a stronger predictor of mortality than obesity in both mid- and late-life.In the interaction models, MU individuals had higher risks of mortality than the MH group,regardless of BMI category.Furthermore, individuals belonging to a lower BMI category who were MU carried higher mortality risks than those with a higher BMI and MH,consistent with past studies [35, 36].A study conducted among older personshas also demonstratedthat MetSaccountedfor 71.3% of a BMI and CVD association.Collectively, these results suggest that MU statusmay beaprimary driver of elevated mortality risk.
Effects of BMI and MHS interactions
Even among individuals with normal weight in late-life, being MU (MUN) increased the risk of mortality compared to MHN, consistent with past research [35, 36, 38, 39]. The raised mortality risk observed among those with MUN in late-life may result from reverse causality - weight loss from pre-existing illnesses. However, our findings show that the increased mortality risk in MUN was only slightly attenuated upon adjusting for weight history. This indicatesthat weight loss from overweight or obesity may not be key drivers to the excess mortality risk in MUN in late-life.Since the metabolic dysfunction in MUN is hidden in plain sight, detecting this phenotype is likely challenging. Greater attention to the evaluation of MHS may be necessaryto better assess mortality risks, even in older adults with normal weight.
While individuals with higher BMI were likely to present with unfavourable metabolic profiles, we still found that within the group of people with obesity, 35.7% of the midlife sample and 36.1% of the late-life sample were metabolically healthy. The existence and prognosis of MHO is a subject thatis debated in the literature. Our findings strengthen the evidence of past research demonstrating MHO as a nonsignificant risk of mortality [36, 39, 40]. Nonetheless, other studies have reported an increased [38, 41]or decreased [35, 42] risk of mortality amongindividuals with MHO. Some have attributed the heterogeneous findings to the various criteria and thresholds used in determining MHS and have called for consensus in the definitions [17, 43].
Criteria for defining MHS - findingsfrom sensitivity analyses
It is important to note that only the magnitude of effects changed while the conclusions generally remained the same when various definitions of metabolic health were explored. One exception was the recently proposed criteria of MHS established systematically to distinguish MHO with decreased mortality by Zembic et al..Applying these new definitions did not substantially change our results in the midlife group. However, late-life MHO was significantly associated with a greater risk of mortality, corresponding with the findings in our sensitivity analysis, when metabolic health was defined with stricter criteria (absence of metabolic abnormality). It is noteworthy that the number of persons with MHO in our study dropped substantially when using the new definitions or stricter criteria, which may have explained the rise in mortality risks associated with MHO in late-life. Since these new criteria were derived systematically from a younger population with a mean age of 41.6 years, the risk pattern yielded from our midlife group was consistent with the study , as expected. Nonetheless, the findings from the late-life group were contradictory, thus raising concerns about whether criteria established in a midlife sample apply as well in late-life. Therefore, it may be justified to create age-specific criteria and cut-offs in the definitions of MHS and obesity.
Effects of MHOwin particular
The obesity paradox, the counterintuitive lower risks of mortality among those with high BMI, which tends to manifest in studies among older persons [2, 6-9], was not observed in our study. However, the lack of association between those with MHO and mortality and the negative association between MHOw in late-life and mortality highlight the role of MHS in generating paradoxical relations between high BMI and mortality. For example, an obesity paradox may present in a study population where individuals with higher BMI were metabolically healthier. Such a selection bias for healthier individuals tend to occur in many studies on aging.
This decline in mortality risk associated with MHOw in late-life casts doubts onweight loss recommendations for older persons who are metabolically healthy and overweight. Indeed, the latest nutrition and hydration guidelines in geriatrics established by the European Society for Clinical Nutrition and Metabolism advocated against weight-reducing diets for older individuals who are overweight . The authors mentioned the accumulating evidence of the importance of metabolic risks;notwithstanding, there were no recommendations in assessing MHS as part of risk evaluation related to BMI.
The risk of mortality from individual metabolic componentsfrom sensitivity analysis
Among themetabolic components used to determine MHS, hyperglycaemia had the largestindependent effects on mortality in both age categories, in line with previous literature [38, 45]. However, reports of how the rest of the metabolic components relate to mortality have beenconflicting. While the association between hypertriglyceridemia in late-life and mortality in our study corresponds with the research findings derived from a younger study population , a meta-analysis found hypertriglyceridemia was protective among older persons with a median age of 73 years.In our study, neither hypertension nor low HDL in mid- and late-life was associated with mortality, contradictingpriorevidence[38, 45, 46]. The differences in how individual metabolic components relate to mortality may result from variations in age, pharmaceutical treatments, and the prevalence of metabolic dysfunctions in different study populations. Further investigations are necessary to understand this heterogeneity.
Although there is evidence that the effects of MHS are greater than BMI, the clinical importance of high BMI should not be downplayed. There is still substantial evidence linking high BMI to major non-communicable diseases, like CVD [47, 48], type II diabetes mellitus(T2DM) [48, 49] and cancer [50-52], which in turn predispose individuals to premature death. However, high BMI in itself may not be the proximal cause of mortality, thus explaining the weakened effects of BMI in the joint models and in interactions with MH status. Moreover, the dose-dependent increment in the prevalence of MU status among individuals from higher BMI categories in our study, in line with previous research , likely arises from the strong correlation between high BMI and metabolic dysfunction. Since metabolic parameters outside of the healthy range are well-recognized risk factors for CVD and T2DM , targeting both high BMI and impaired metabolic parameters is likely crucial in primary prevention.
Strengths and Limitations
These findings contribute to understanding the gaps in our knowledge of how mid- and late-life BMI and MHS, independently, jointly, and in interactions, impact mortality.In addition, weight and height used in the derivation of BMI were measured objectively by trained, licensed nurses, thus reducing measurement errors.Moreover, we included weight history in our models to limit reverse causality. Furthermore, since the outcomes data were obtained from linked Tax registries in Sweden, we had comprehensive coverage of mortality.
There are some limitations to our study. Firstly, we included thresholds on non-fasting glucose and lipid levels, which may underestimate the prevalence of hyperglycemia and dyslipidemia. However, the proportion of non-fasting measures was relatively low. Secondly, using age as a timescale in the analysis limits the ability to account for period effects, such as differences in medication use over time on the population level. Thirdly, the mean follow-up time of themidlife group, at 13.9 years,meant our data in the midlife group is at risk of capturing mostly early deaths. Since the sensitivity analyses indicated that CVD in midlife was associated with a nine-fold increase in the risk of death, premature death from CVD events likely accounted for most of the mortality in the midlife group. In addition, the unusually high HR of CVD may be an overestimation caused by the close link between CVD and MHS observed in other studies[55, 56].Moreover, the estimates from the late-life group, with a mean follow-up time of 12 years,is potentially susceptible to reverse causation between lower BMI and mortality.Finally, to correct for potential bias due to weight loss from pre-existing morbidities, we adjusted the models for weight history, which did not drastically change the effects of BMI-MHS phenotypes.
Lastly, this study accounted for BMI and MHS only at baseline and could not capture the impact on mortality from the trajectories of BMI, MHS and the BMI-MHS phenotypes. When we included weight history in the models, the history of overweight and obesity heightenedmortality risks, suggesting cumulative adverse effects from having high BMI.Furthermore, an extensive study of 90,257 women over 30 years supported the transient nature of BMI-MHS and showedthat long periods of obesity increased CVD risk, despite preserved metabolic health.The same study also concluded that many women with MHO transitioned to MUO over time. Future research should identify trajectories of BMI-MHS phenotypes and their impact on mortality.