Correlation between allostatic load index and cumulative mortality: a register-based study of Danish municipalities

Abstract

Objectives

The aim of this study was to examine population-based allostatic load (AL) indices as an indicator of community health across 14 municipalities in Denmark.

Design

Register-based study.

Setting

Data derived from: the Lolland-Falster Health Study, the Copenhagen General Population Study and the Danish General Suburban Population Study. Nine biomarkers (systolic blood pressure, diastolic blood pressure, pulse rate, total serum cholesterol, high-density lipoprotein cholesterol, waist-to-hip ratio, triglycerides, C-reactive protein and serum albumin) were divided into high-risk and low-risk values based on clinically accepted criteria, and the AL index was defined as the average between the nine values. All-cause mortality data were obtained from Statistics Denmark.

Participants

We examined a total of 106 808 individuals aged 40–79 years.

Primary outcome measure

Linear regression models were performed to investigate the association between mean AL index and cumulative mortality risk.

Results

Mean AL index was higher in men (range 2.3–3.3) than in women (range 1.7–2.6). We found AL index to be strongly correlated with the cumulative mortality rate, correlation coefficient of 0.82. A unit increase in mean AL index corresponded to an increase in the cumulative mortality rate of 19% (95% CI 13% to 25%) for men, and 16% (95% CI 8% to 23%) for women but this difference was not statistically significant. The overall mean increase in cumulative mortality rate for both men and women was 17% (95% CI 14% to 20%).

Conclusions

Our findings indicate the population-based AL index to be a strong indicator of community health, and suggest identification of targets for reducing AL.

Correspondence to Dr Neda Esmailzadeh Bruun-Rasmussen; [email protected]

STRENGTHS AND LIMITATIONS OF THIS STUDY

• Analysis based on data from three population-based health surveys.

• Large sample size representative of Danes aged 40–79 years.

• Access not allowed to individual data on biomarkers.

• Lack of primary neuroendocrine biomarkers.

Introduction

Stress can be defined as a state of threatened homeostasis triggered by extrinsic or intrinsic factors.¹ It is counteracted by a complex range of physiological responses aiming to maintain and re-establish the body equilibrium. The body’s stress response depends on a highly interconnected molecular, neuroendocrine and cellular infrastructure. Long-term dysregulation of the stress response has been linked to increased mortality risk by accelerating biological wear and tear, ageing and chronic disease pathophysiology including fat deposition, inflammation and oxidative stress.² To better understand and characterise the relationship between stress and human biology, McEwen and Stellar developed the concept of allostatic load (AL), defined as a cumulative measure of stress-related physiological adaptations.³ AL can be determined either by subsets of stress biomarkers from the neuroendocrine, cardiovascular, metabolic and immune systems, calculated by summing the dysregulated biomarkers into an AL index, defined by clinical or distributional thresholds^4–12; or via psychosocial indicators.^{13 14}

All-cause mortality is the most comprehensive measure of health, and it is used for comparison between countries, social groups, time periods and so on. However, as death is an irreversible event, all-cause mortality is a too late indicator of the vulnerability of a population and a poor guide for preventive interventions. On this basis, it is pertinent to find alternative measures of health allowing for identification of vulnerable groups. We investigated whether the AL index was a good predictor of community-based vulnerability. We took advantage of population-based health surveys from multiple Danish communities with different levels of all-cause mortality¹⁵; the Copenhagen General Population Study (CGPS)¹⁶ included the municipalities of Gentofte and Rudersdal located north of the capital Copenhagen characterised by low all-cause mortality; the Danish General Suburban Population Study (GESUS)¹⁷ with data from the provincial municipality Næstved represented a more average mortality level, and the Lolland-Falster Health Study (LOFUS)¹⁸ included the Danish municipalities with the highest all-cause mortality. We hypothesised that a high community-based AL index would predict a high community-based all-cause mortality.

Methods

The ‘Strengthening the Reporting of Observational Studies in Epidemiology’ guidelines were followed.¹⁹

Study population

As the association between biological indicators and mortality is expected to be strongest for the adult population, we focused on participants in the health surveys aged 40–79 years at recruitment. The CGPS is an ongoing prospective study initiated in 2003.¹⁶ The targeted population included inhabitants in Allerød, Ballerup, Egedal, Furesø, Gentofte, Gladsaxe, Herlev, Hørsholm, Lyngby-Taarbæk, Rudersdal and Rødovre municipalities. Individuals aged 20+ years were invited at random based on the Danish Civil Registration System. The GESUS was undertaken from 2010 to 2013.¹⁷ Randomly selected individuals aged 30+ years from Næstved municipality were invited. The LOFUS was undertaken from 2016 to 2020. From the municipalities of Lolland and Guldborgsund, a random sample of individuals aged ≥18 years were selected and their entire households were invited. In all three health surveys, participants underwent physical examination, had blood samples drawn and filled in questionnaires.¹⁸ Included municipalities are depicted in online supplemental figure 1.

Patient and public involvement

Patients were not actively involved in any stage of the present study. Once the paper has been published in the international literature, the key results will be reported also in the local press.

Allostatic load

The AL index was calculated as a summary measure of nine biomarkers across a range of regulatory systems related to disease and mortality and used in most validated AL index constructs.^{2 5} These included biomarkers from the cardiovascular system (systolic blood pressure (SBP), diastolic blood pressure (DBP), and pulse rate (PR)); the metabolic system (total serum cholesterol (TC), high-density lipoprotein cholesterol (HDL-c), waist-to-hip ratio (WHR), triglycerides (TG)); and the inflammatory system (C-reactive protein (CRP), serum albumin). The AL index can be constructed either based on dichotomised or continuous biomarkers; however, previous studies have found that the methods present similar results.²⁰ Therefore, for each biomarker, a dichotomous indicator was created, reflecting those with ‘high-risk’ values (assigned a value of ‘1’) and ‘low-risk’ values (assigned a value of ‘0’). Value assignment as high or low was based on clinically accepted ‘high-risk’ criteria in line with previous studies^{2 5} (table 1). All biomarkers were given equal weight, and the total AL index was defined as the sum of all the biomarkers. In order to compare our findings across communities with different age distributions, the mean AL index was age-standardised using the European standard population,²¹ separately for men and women.

Table 1

Distribution of biomarkers, their clinically-defined ‘high-risk’ criteria for allostatic load by study: Copenhagen General Population Study (CGPS), Danish General Suburban Population Study (GESUS) and Lolland-Falster Health Study (LOFUS)

Biomarker	Number of persons with high-risk values. High cut-off (%)	Mean (SD)	χ2 test
Cardiovascular system
Systolic blood pressure (mmHg), cut-off: ≥140
CGPS	37 981 (48.2)	139 (21)	<0.0001
GESUS	8641 (52.6)	142.4 (21.2)
LOFUS	4311 (37.3)	135.2 (18.5)
Diastolic blood pressure (mmHg), cut-off: ≥90
CGPS	18 748 (23.8)	81 (12)	<0.0001
GESUS	5918 (36.0)	85.8 (11.2)
LOFUS	1359 (11.7)	80.5 (8.2)
Pulse rate (beats/min), cut-off: >90
CGPS	7231 (9.2)	72 (12.0)	<0.0001
GESUS	1420 (8.6)	72.2 (12.3)
LOFUS	302 (2.6)	66.8 (10.4)
Metabolic system
High-density lipoprotein cholesterol (mmol/L), cut-off: ≤1
CGPS	8594 (10.9)	1.4 (0.5)	<0.0001
GESUS	2057 (12.5)	1.6 (0.5)
LOFUS	1701 (14.7)	1.5 (0.5)
Total serum cholesterol (mmol/L), cut-off: ≥5
CGPS	65 548 (83.2)	5.6 (1.1)	<0.0001
GESUS	11 640 (70.9)	5.6 (1.0)
LOFUS	7.050 (61.0)	5.3 (1.1)
Triglycerides (mmol/L), cut-off: ≥2.6
CGPS	12 975 (16.5)	1.7 (1.1)	<0.0001
GESUS	3124 (19.0)	1.9 (1.2)
LOFUS	2121 (18.3)	1.8 (1.2)
Waist-to-hip ratio, cut-off: men ≥0.90; women ≥0.85
CGPS	46 217 (58.6)	0.9 (0.1)	<0.0001
GESUS	11 088 (67.5)	0.9 (0.4)
LOFUS	7919 (68.5)	0.9 (0.1)
Inflammatory system
C-reactive protein (mg/L), cut-off: ≥3
CGPS	14 762 (18.7)	2.4 (4.7)	<0.0001
GESUS	4033 (24.6)	2.7 (5.1)
LOFUS	3174 (27.5)	2.9 (5.4)
Serum albumin (g/L), cut-off: <34
CGPS	1915 (2.4)	40.9 (3.6)	<0.0001
GESUS	7 (0)	45.9 (2.6)
LOFUS	153 (1.3)	39.4 (2.6)

Mortality

Data on the number of deaths and person years by 5-year age group, sex and municipality were obtained from Statistics Denmark from the calendar years in which the blood samples were collected,²² and we calculated cumulative mortality risk from age 40 to 80 by sex and municipality.

Statistics

The analysis of GESUS and LOFUS data was based on individual records, while the CGPS Steering Committee did not allow access to individual records, and the CGPS data were therefore provided in aggregated form. Therefore, we calculated χ² test p values for the homogeneity between the three health surveys in the distribution of high-risk and low-risk values of the biomarkers. The association between the AL index (total, system-specific and individual biomarkers) and cumulative mortality risk was first assessed with Pearson’s correlation. Furthermore, the relation between mean AL index and cumulative mortality risk was analysed by linear regression models, separately with interaction with sex and without sex distinction. As our dependent variable (cumulative mortality risk) has values in (0, 1), linear models are in general not suitable for such analyses. Therefore, to assess the validity of the linear assumption in the range of our sampled AL data (from 1.7 to 3.3), we further analysed the data through beta regression models. These provided similar results as the linear models in the relevant range of AL values (data not reported). In all analyses, the significance level was set to 0.05.

Individual data from GESUS and LOFUS were accessed at a server in Region Zealand. Data analyses and figures were performed in R V.4.1.3²³ with packages tidyverse²⁴ and betareg.²⁵

Results

A total of 106 808 participants aged 40–79 were included; 78 830 (73.8%) from CGPS, 16 418 (15.4%) from GESUS; and 11 560 (10.8%) from LOFUS (table 2). For the three studies together, 54.2% of participants were women and 45.8% were men. Participants had an age distribution of 11.2%, 13.7%, 13.8%, 14.1%, 15.6%, 14.4%, 10.6% and 6.5%, respectively, for the age groups 40–44, 45–49, 50–54, 55–59, 60–64, 65–69, 70–74 and 75–79 years.

Table 2

Baseline characteristics of participants in Copenhagen General Population Study (CGPS), Danish General Suburban Population Study (GESUS) and Lolland-Falster Health Study (LOFUS) aged 40–79 years

Characteristics	CGPS N (%)	GESUS N (%)	LOFUS N (%)	Total N (%)
Overall	78 830 (73.8)	16 418 (15.4)	11 560 (10.8)	106 808 (100)
Gender
Females	43 260 (54.9)	8560 (52.1)	6036 (52.2)	57 856 (54.2)
Males	35 570 (45.1)	7858 (47.9)	5524 (47.8)	48 952 (45.8)
Age at baseline, years
40–44	9008 (11.4)	1972 (12.0)	947 (8.2)	11 927 (11.2)
45–49	11 184 (14.2)	2270 (13.8)	1207 (10.4)	14 661 (13.7)
50–54	10 990 (13.9)	2140 (13.0)	1557 (13.5)	14 687 (13.8)
55–59	11 079 (14.1)	2312 (14.1)	1696 (14.7)	15 087 (14.1)
60–64	12 213 (15.5)	2692 (16.4)	1781 (15.4)	16 686 (15.6)
65–69	11 045 (14.0)	2527 (15.4)	1852 (16.0)	15 424 (14.4)
70–74	8105 (10.3)	1584 (9.6)	1667 (14.4)	11 356 (10.6)
75–79	5206 (6.6)	921 (5.6)	853 (7.4)	6980 (6.5)
Biomarker data
All	88 822 (100)	16 758 (100)	12 453 (100)	118 033 (100)
Incomplete	9992 (11.2)	340 (2.0)	893 (7.2)	11 225 (9.5)

The distributions of persons in high-risk groups according to each biomarker and AL index are presented in table 1. There was some variation in biomarker levels across the health surveys. For the cardiovascular system, the proportions of persons in the high-risk group for SBP, DBP and PR were higher in GESUS (52.6%, 36.0%, 8.6%) and CGPS (48.2%, 23.8%, 9.2%) than in LOFUS (37.3%, 11.7%, 2.6%). For the metabolic system, the proportion of persons in high-risk group for TC decreased from CGPS (83.2%) to GESUS (70.9%) and LOFUS (61.0%), while high TG occurred with almost equal frequency (CGPS (16.5%), LOFUS (18.3%) and GESUS (19.0%)). A low proportion of persons had low HDL-c (LOFUS (14.7%) in GESUS (12.5%) and CGPS (10.9%)). The majority of persons had high level of WHR both in CGPS (58.6%) in GESUS (67.5%) and in LOFUS (68.5%). Higher proportions of persons with elevated CRP were found in LOFUS (27.5%) and GESUS (24.6%) than in CGPS (18.7%). Low serum albumin level was infrequent in all three studies CGPS (2.4%), LOFUS (1.3%) and GESUS (0%).

The data collection on which our AL measurements were based took place between 2003 and 2019. In this approximately 15-year period, life expectancy in Denmark increased by overall 3.8 years.²⁶ Throughout the period, women had higher life expectancy than men though diminishing from 4.7 years in 2003–2004 to 3.9 years in 2018–2019.²⁶ This pattern was reflected in the municipality-based data where the cumulative mortality risk was systematically higher for men than for women (online supplemental table 1). Gladsaxe and Herlev had the highest cumulative mortality risk, but it should be noted that in these municipalities, data collection started already in the early 2000s. Lolland, where data collection started more than 10 years later, was also among the top 5 municipalities in cumulative mortality risk.

The mean AL index was higher for men than for women (table 3). In men, the mean AL indices ranged from 3.3 in Rødovre to 2.3 in Rudersdal. The highest mean AL indices were found in Rødovre (3.3), Herlev (3.2), Næstved (3.2), Ballerup (3.1) and Gladsaxe (3.1). For men, the correlation coefficient between mean AL index and cumulative mortality risk was 0.84 (95% CI 0.55 to 0.95). In women, the mean AL indices ranged from 2.6 in Næstved to 1.7 in Rudersdal and Hørsholm. The highest mean AL indices were found in Næstved (2.6), Herlev (2.5), Ballerup (2.4), Gladsaxe (2.4) and Rødovre (2.4); the same municipalities that topped the list in men. The correlation coefficient between mean AL index and cumulative mortality risk was 0.84 (95% CI 0.55 to 0.95) for women.

Table 3

Allostatic load and cumulative mortality risk for men and women aged 40–79 years by municipality

Study/municipality/sex	Data collection years	Number of persons	Mean* AL	Cumulative mortality risk
CGPS
Allerød	2013–2014
Males		1797	2.5	0.35
Females		1976	1.9	0.24
Ballerup	2005–2010
Males		4389	3.1	0.52
Females		5174	2.4	0.39
Egedal	2011–2012
Males		3841	2.5	0.39
Females		4349	1.9	0.32
Furesø	2008–2013
Males		2029	2.7	0.36
Females		2398	1.9	0.29
Gentofte	2006–2013
Males		5098	2.8	0.41
Females		6561	2.0	0.30
Gladsaxe	2004–2009
Males		4883	3.1	0.54
Females		5832	2.4	0.40
Herlev	2003–2014
Males		2568	3.2	0.54
Females		3098	2.5	0.40
Hørsholm	2013–2015
Males		1861	2.7	0.35
Females		2341	1.7	0.26
Lyngby-Taarbæk	2009–2015
Males		3573	2.4	0.42
Females		4544	1.8	0.29
Rudersdal	2012–2014
Males		3910	2.3	0.33
Females		4898	1.7	0.27
Rødovre	2008–2010
Males		1621	3.3	0.53
Females		2089	2.4	0.38
GESUS
Næstved	2010–2013
Males		7858	3.2	0.48
Females		8560	2.6	0.34
LOFUS
Lolland	2016–2019
Males		2243	2.7	0.50
Females		2479	2.1	0.39
Guldborgsund	2016–2019
Men		3281	2.6	0.47
Women		3557	2.0	0.33
Correlation†
Males	0.84 (95% CI 0.55 to 0.95)
Females	0.84 (95% CI 0.55 to 0.95)
Combined	0.91 (95% CI 0.82 to 0.96)

*Age-standardised mean.

†Mean allostatic load and cumulative mortality risk.

CGPS, Copenhagen General Population Study; GESUS, Danish General Suburban Population Study; LOFUS, Lolland-Falster Health Study.

The correlation between mean AL for single biomarkers, organ systems and cumulative mortality for men and women is reported in online supplemental table 2 (see also online supplemental figure 2).

From linear regression models, it was found that in the range of AL values investigated in this study, a unit increase in mean AL index corresponds to an increase in cumulative mortality rate of 19% (95% CI 13% to 25%) for men, 16% (95% CI 8% to 23%) for women, with no statistically significant difference between sexes (p value for interaction=0.5). When considering both sexes together, the increase was 17% (95% CI 14% to 20%, adjusted R²=0.82). Similar results were obtained from weighted linear regressions models, with weights equal to the size of each subgroup (defined by sex and municipality) (18% (12% to 24%) for men, 13% (7% to 20%) for women and 17% (14% to 20%) overall) (figure 1).

Figure 1. Cumulative mortality and mean allostatic load by sex for each of the 14 municipalities included in the study. Each municipality is represented by one red (female) and one blue (male) dot.

Discussion

Main finding

In more than 100 000 adults recruited in three population-based health surveys in Denmark, the community-based AL was strongly associated with the cumulative all-cause mortality risk in the community. Men had in general higher mean AL and mortality risk than women. However, the association between AL and cumulative mortality was almost the same between the sexes. Furthermore, we found that a unit increase in mean AL index corresponded to an increase in cumulative mortality rate of 17%. Our study added to the evidence for the AL index being a valid indicator of population vulnerability.

Comparison with other studies

High AL has been associated with poor long-term health outcomes including declines in physical and cognitive functioning,^{4 5} cardiovascular disease (CVD)⁶ and periodontal disease.⁷ In a recent systematic review, Parker et al found that high AL was associated with increased all-cause mortality.² Based on 10 studies, the overall meta-analysis yielded a pooled HR of 1.22 (95% CI 1.14 to 1.30) for all-cause mortality in persons with high AL versus those with low AL. The authors concluded that AL is an emerging and potently modifiable risk factor for all-cause mortality. However, there is a considerable heterogeneity in the definition of AL across studies highlighting the need for standardised measurement. Originally, the AL was measured based on 10 biomarkers: epinephrine, norepinephrine, urinary cortisol, dehydroepiandrosterone sulfate, SBP, DBP, WHR, total cholesterol, HDL-c and glycosylated haemoglobin (HbA1c).²⁷ However, Seeman et al reported that “this original set of 10 parameters was not meant to be comprehensive nor was it offered as a fixed/standard measure of AL”²⁸ as these 10 biomarkers were based on secondary data analyses from the MacArthur Successful Aging Study and limited to biomarkers that were available. Currently, there is still no gold standard measure of the AL index, and most researchers include biomarkers that are available.²⁹ The lack of a fixed definition has thus contributed to the challenge of comparing studies on AL.

In our community-based data, the mean AL index in all municipalities was higher in men than in women. In a review of more than 60 studies, Kerr et al found that sex differences in AL were rarely analysed.³⁰ Most studies only added sex as a covariate in the statistical analysis, while stratification by sex or analysis of the potential effect of sex was performed in a few studies only. Seven studies found no sex difference, nine studies found that women had lower AL than men, and the opposite pattern was not found. The reason for the consistency of the sex pattern in our study might be that the same AL measure was used across all 14 communities included in the study.

The correlation between mean AL index and cumulative mortality was stronger than the correlations of individual biomarkers and/or organ systems with cumulative mortality. This is in agreement with the pattern we found previously in analysis of individual data from LOFUS.⁸ Several other studies have also reported the index of AL to be a better predictor of mortality than individual biomarkers.⁵ Castagné et al examined 11-year mortality by using data from the 1958 British birth cohort including 8113 adults. They calculated AL based on 14 biomarkers representing four physiological systems including salivary cortisol t1, salivary cortisol t1–t2, insulin-like growth factor-1, CRP, fibrinogen, immunoglobulin E, HDL-c, low-density lipoprotein, TG, HbA1C, SBP, DBP, PR and peak expiratory flow. They found that higher AL at 44 years old was a significant predictor of mortality 11 years later (HR=3.56) and that AL index was advantageous compared with evaluating each physiological system subscore and biomarker separately.¹¹

Our findings on the higher mortality risk in men compared with women are consistent with previous studies.^{31 32} The average life expectancy has increased over the last decades³¹ and it has been reported that although women in general have higher levels of disability and morbidity, they have longer life expectancy than men. The level of male excess mortality has varied over time, and several explanations for the excess risk have been proposed including biological, social and behavioural factors.^31–34 In a recent large population-based study of 12 longitudinal cohorts including 179 044 individuals, Wu et al reported that men had 60% higher mortality risk than women after adjustment for age (pooled HR: 1.6; 95% CI 1.5 to 1.7). Only smoking and CVD substantially attenuated the effect size (by approximately 22%).³²

Seeman et al investigated the relationship between socioeconomic status, AL and all-cause mortality, and found that higher AL explained 35% of the difference in mortality risk between those of higher socioeconomic status and those of lower socioeconomic status.²⁷ In their analysis, AL retained independent explanatory power even after adjusting for established risk factors and diagnosed disease. Several studies have focused on the association between socioeconomic status and AL, and found low socioeconomic status to be associated with high levels of AL.^{5 35}

During the last decades, several risk algorithms or ‘health metrics’ have been developed in order to identify population vulnerability focusing on estimating risk of negative future health outcomes.^36–39 However, most of these health metrics are complicated, not easily accessible, and challenging to interpret, and also their degrees of validation differ tremendously.^37–39 Therefore, the use of AL as an indicator of population vulnerability, generated by simple biomarkers, may be a more realistic, affordable and powerful tool for assessing risks of morbidity and mortality across different populations.

Strengths and limitations

It was a strength of the present study that biomarker data could be analysed from three population-based health surveys recruiting more than 100 000 adult persons and undertaken in a broad range of Danish municipalities, including both those with the lowest and highest mortality. It was furthermore a strength that data on nine biomarkers across three organ systems were available from all three surveys.

However, our study also had limitations. First, data from the largest health survey CGPS could be provided in aggregate form only, and we were therefore not able to (1) analyse within-community variability and (2) calculate CIs for the age-standardised means of AL. Second, of the persons invited to the health surveys, 36%–40% participated.^40–43 This might indicate selectivity in participation, but when this was investigated halfway through the LOFUS survey, only a moderate gradient was seen in non-participation by, for example, education; relative risks from 1 in high, to 1.13 in medium, to 1.38 in low education.⁴² Third, data on biomarkers from the neuroendocrine system were not available; the neuroendocrine system is known to contribute significantly to allostasis and subsequent AL, as a sequence of physiological changes occurs prior to the onset of initial stress responses (including rapid increase in blood glucose levels and blood pressure, providing the body with extra energy). However, obtaining biomarkers from the neuroendocrine system is challenging, as it is recommended to conduct repeated measurements over 1–2 days which is mostly not possible in health surveys where participants are examined only once. Fourth, HbA1c was not measured in the CGPS data, but still we included the most frequently used biomarkers in the AL literature.^{20 44} Fifth, we only used clinical cut-off values to define high and low levels of the individual biomarkers. In a previous analysis of AL and mortality in LOFUS, we used percentile cut-offs (ie, upper and lower quartiles of the sample distribution),⁸ but this was not possible in this study as we did not have access to individual data from CGPS. Sixth, in the literature, several methods have been used for defining the AL index including the count-based, canonical correlation, z-score and grade of membership method.^{2 5} With the available data, we could use only the count-based method in which a summary index is calculated by summing the number of biomarkers falling within the high-risk and low-risk categories, respectively. Seventh, merging data for men and women might be misleading, and the use of sex-specific biomarker cut-offs has been suggested in previous studies,^{45 46} but is very rarely applied in the AL literature.

Across the three health surveys, we observed some diversity in the levels of cardiovascular biomarkers. This is to be expected, if the biomarkers reflect the level of health in the respective communities. However, we cannot exclude that also differences in measurement methodology can have played a role. In CGPS, blood pressure and PR were measured by an automatic Digital Blood Pressure Monitor (Kivex) on the left arm, after 5 min of rest and in sitting position.¹⁶ GESUS used two consecutive digital measurements of blood pressure and PR on the left upper arm (apparatus type A&D UA-787, A&D Medical, Tokyo, Japan) after 5 min of rest and registered only the second.¹⁷ In LOFUS, blood pressure and PR were measured after 5 min of rest by three consecutive digital measurements on the upper left arm (apparatus type Welch Allyn Connex proBPO 3400), and the mean value of second and third measurements was used.¹⁸ Some methodological variations in measurements are inevitable when data are analysed from several health surveys.

Interpretation

The notion that the mortality level in a community can be predicted from a very simple index of biomarkers opens avenues for future investigation. Our AL index was composed of blood pressure (SBP and DBP), PR, WHR and five biomarkers measured in blood samples: TC, HDL-c, TG, CRP and serum albumin. The values of each of these biomarkers were dichotomised into high (1) or low (0) risk, based on clinically used cut-off values.

In the range of observed AL means, a linear association was found between mean AL index and cumulative mortality risk. Our regression analysis showed that for men, a one-point change in mean AL index was associated with an approximate 19% absolute change in cumulative mortality risk. To evaluate the potential impact of such a change, we used the current mortality differences across the municipalities. In 2016–2019, the lowest cumulative mortality risk for men aged 40–79 years was 30% found in Allerød municipality, while the highest cumulative mortality risk was 50% and found in Lolland municipality (online supplemental table 1). So, a one-point decrease in mean AL index would basically correspond to elimination of the present geographical differences in mortality across Denmark. The data thus suggested that a very moderate change in the biomarker profile of the population, for example, a switch of 1–2 biomarkers from high-risk to low-risk values, could have a considerable impact on all-cause mortality. However, this change would require interventions comprising a large number of people, and also further studies are needed to understand the regression process of AL and thereby ‘normalisation’ of strained organ systems.

Conclusion

This study based on biomarker data from 100 000 adult persons living in 14 Danish municipalities showed that a community-based AL index based on nine biomarkers was a strong predictor of all-cause mortality in the community both in men and women.

We thank the participants of the Copenhagen General Population Study, the Lolland-Falster Health Study, and the Danish General Suburban Population Study for their willingness to participate.

Data availability statement

Data are available upon reasonable request. Data from LOFUS and GESUS study can be made available via Region Zealand following the Danish Data Protection Regulation. The CGPS data underlying this paper will be shared in reasonable request to the steering committee of the Copenhagen City Heart Study. On request, a data sharing agreement will be made and sent to the Danish Data Protection Agency for approval. Public sharing of the data is not possible under the Danish data agreement.

Ethics statements

Patient consent for publication

Not applicable.

Ethics approval

This study was performed in line with the principles of the Declaration of Helsinki. LOFUS data storage and management were approved by the Regional Data Protection Agency of Zealand (REG-024-2019 & REG-24-2015). LOFUS is registered in Clinicaltrials.gov (NCT02482896). GESUS was approved by the appropriate institutional review boards and ethical committees (SJ-113, SJ-114, SJ-147, SJ-278), and it was reported to the Danish Data Protection Agency. CGPS was approved by a Danish ethics committee (#H-KF-01-144/01) and by Herlev Hospital, Copenhagen University Hospital.

Footnote

Contributors All authors contributed significantly to the study. NEB-R, EL and GN designed the study, interpreted the data and drafted the manuscript. TLFH contributed to the calculation of mortality data. GN performed the statistical analysis. CE, KR and SEB contributed to the interpretation and writing of the manuscript. All authors critically revised and approved the final manuscript. EL is the guarantor.

Funding Region Zealand/University of Copenhagen, Professor grants for Elsebeth Lynge, grant/award number: Not Applicable (2) Nykøbing Falster Hospital, grant/award number: Not Applicable (3) Knud og Edith Eriksens Mindefond, grant/award number: 62786-2021. The funding bodies had no role in the design of the study, neither in the collection, analysis, and interpretation of data nor in the writing of the manuscript.

Map disclaimer The inclusion of any map (including the depiction of any boundaries therein), or of any geographic or locational reference, does not imply the expression of any opinion whatsoever on the part of BMJ concerning the legal status of any country, territory, jurisdiction or area or of its authorities. Any such expression remains solely that of the relevant source and is not endorsed by BMJ. Maps are provided without any warranty of any kind, either express or implied.

Competing interests None declared.

Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this 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.

References

¹ Tsigos C, Kyrou I, Kassi E, et al. Stress: endocrine physiology and pathophysiology. In: Feingold KR, Anawalt B, Blackman MR, eds. Endotext. South Dartmouth (MA): MDText.com, Inc, 2020.

² Parker HW, Abreu AM, Sullivan MC, et al. Allostatic load and mortality: a systematic review and meta-analysis. Am J Prev Med 2022; 63: 131–40. doi:10.1016/j.amepre.2022.02.003

³ McEwen BS, Stellar E. Stress and the individual. Mechanisms leading to disease. Arch Intern Med 1993; 153: 2093–101.

⁴ D’Amico D, Amestoy ME, Fiocco AJ. The association between allostatic load and cognitive function: a systematic and meta-analytic review. Psychoneuroendocrinology 2020; 121. doi:10.1016/j.psyneuen.2020.104849

⁵ Guidi J, Lucente M, Sonino N, et al. Allostatic load and its impact on health: a systematic review. Psychother Psychosom 2021; 90: 11–27. doi:10.1159/000510696

⁶ Gillespie SL, Anderson CM, Zhao S, et al. Allostatic load in the association of depressive symptoms with incident coronary heart disease: the Jackson heart study. Psychoneuroendocrinology 2019; 109. doi:10.1016/j.psyneuen.2019.06.020

⁷ Sabbah W, Watt RG, Sheiham A, et al. Effects of allostatic load on the social gradient in ischaemic heart disease and periodontal disease: evidence from the third national health and nutrition examination survey. J Epidemiol Community Health 2008; 62: 415–20. doi:10.1136/jech.2007.064188

⁸ Bruun-Rasmussen NE, Napolitano G, Christiansen C, et al. Allostatic load as predictor of mortality: a cohort study from Lolland-Falster, Denmark. BMJ Open 2022; 12: e057136. doi:10.1136/bmjopen-2021-057136

⁹ Shi TH, Wang B, Natarajan S. The influence of metabolic syndrome in predicting mortality risk among US adults: importance of metabolic syndrome even in adults with normal weight. Prev Chronic Dis 2020; 17: E36. doi:10.5888/pcd17.200020

¹⁰ Li Y, Zhong X, Cheng G, et al. Hs-CRP and all-cause, cardiovascular, and cancer mortality risk: a meta-analysis. Atherosclerosis 2017; 259: 75–82. doi:10.1016/j.atherosclerosis.2017.02.003

¹¹ Castagné R, Garès V, Karimi M, et al. Allostatic load and subsequent all-cause mortality: which biological markers drive the relationship? Findings from a UK birth cohort. Eur J Epidemiol 2018; 33: 441–58. doi:10.1007/s10654-018-0364-1

¹² McEwen BS. Allostasis and allostatic load: implications for neuropsychopharmacology. Neuropsychopharmacology 2000; 22: 108–24. doi:10.1016/S0893-133X(99)00129-3

¹³ Fava GA, McEwen BS, Guidi J, et al. Clinical characterization of allostatic overload. Psychoneuroendocrinology 2019; 108: 94–101. doi:10.1016/j.psyneuen.2019.05.028

¹⁴ Piolanti A, Offidani E, Guidi J, et al. Use of the psychosocial Index: a sensitive tool in research and practice. Psychother Psychosom 2016; 85: 337–45. doi:10.1159/000447760

¹⁵ Available: https://www.statistikbanken.dk/10015 [Accessed Feb 2022 ].

¹⁶ Varbo A, Benn M, Tybjærg-Hansen A, et al. Remnant cholesterol as a causal risk factor for ischemic heart disease [published correction appears in J Am Coll Cardiol. 2019 Mar 5;73(8):987-988]. J Am Coll Cardiol 2013: 427–36. doi:10.1016/j.jacc.2012.08.1026

¹⁷ Bergholdt HKM, Bathum L, Kvetny J, et al. Study design, participation and characteristics of the Danish General Suburban population study. Dan Med J 2013; 60: A4693.

¹⁸ Jepsen R, Egholm CL, Brodersen J, et al. Lolland-Falster health study: study protocol for a household-based prospective cohort study. Scand J Public Health 2020; 48: 382–90. doi:10.1177/1403494818799613

¹⁹ Cuschieri S. The STROBE guidelines. Saudi J Anaesth 2019; 13: S31–4. doi:10.4103/sja.SJA_543_18

²⁰ Carbone JT, Clift J, Alexander N. Measuring allostatic load: approaches and limitations to algorithm creation. J Psychosom Res 2022; 163: 111050. doi:10.1016/j.jpsychores.2022.111050

²¹ Waterhouse JAH, Muir CS. Cancer incidence in five continents, 3. Lyon: IARC, 1976: 456.

²² Available: https://www.dst.dk/en/Statistik/emner/borgere/befolkning/doedsfald [Accessed Dec 2022 ].

²³ R Core Team. R: A language and environment for statistical computing. Vienna, Austria R Foundation for Statistical Computing; 2022. Available: https://www.R-project.org

²⁴ Wickham H, Averick M, Bryan J, et al. Welcome to the Tidyverse. JOSS 2019; 4: 1686. doi:10.21105/joss.01686 Available: https://doi.org/10.21105/joss.01686)

²⁵ Cribari-Neto F, Zeileis A. Beta Regression in R. J Stat Softw 2010; 34: 1–24. doi:10.18637/jss.v034.i02

²⁶ Available: https://www.dst.dk/en/Statistik/emner/borgere/befolkning/middellevetid [Accessed Dec 2022 ].

²⁷ Seeman TE, Crimmins E, Huang M-H, et al. Cumulative biological risk and socio-economic differences in mortality: MacArthur studies of successful aging. Soc Sci Med 2004; 58: 1985–97. doi:10.1016/S0277-9536(03)00402-7

²⁸ McCrory C, McLoughlin S, Layte R, et al. Towards a consensus definition of allostatic load: a multi-cohort, multi-system, multi-biomarker individual participant data (IPD) meta-analysis. Psychoneuroendocrinology 2023; 153. doi:10.1016/j.psyneuen.2023.106117

²⁹ Seeman T, Epel E, Gruenewald T, et al. Socio-economic differentials in peripheral biology: cumulative allostatic load. Ann N Y Acad Sci 2010; 1186: 223–39. doi:10.1111/j.1749-6632.2009.05341.x

³⁰ Kerr P, Kheloui S, Rossi M, et al. Allostatic load and women’s brain health: a systematic review. Front Neuroendocrinol 2020; 59: 100858. doi:10.1016/j.yfrne.2020.100858

³¹ Crimmins EM, Shim H, Zhang YS, et al. Differences between men and women in mortality and the health dimensions of the morbidity process. Clin Chem 2019; 65: 135–45. doi:10.1373/clinchem.2018.288332

³² Wu Y-T, Niubo AS, Daskalopoulou C, et al. Sex differences in mortality: results from a population-based study of 12 longitudinal cohorts. CMAJ 2021; 193: E361–70. doi:10.1503/cmaj.200484

³³ Mackenbach JP, Valverde JR, Bopp M, et al. Determinants of inequalities in life expectancy: an international comparative study of eight risk factors. Lancet Public Health 2019; 4: e529–37. doi:10.1016/S2468-2667(19)30147-1

³⁴ Rogers RG, Everett BG, Onge JMS, et al. Social, behavioral, and biological factors, and sex differences in mortality. Demography 2010; 47: 555–78. doi:10.1353/dem.0.0119

³⁵ Johnson SC, Cavallaro FL, Leon DA. A systematic review of allostatic load in relation to socioeconomic position: poor fidelity and major inconsistencies in biomarkers employed. Soc Sci Med 2017; 192: 66–73. doi:10.1016/j.socscimed.2017.09.025

³⁶ May HT, Anderson JL, Muhlestein JB, et al. Improvement in the predictive ability of the intermountain mortality risk score by adding routinely collected laboratory tests such as albumin, bilirubin, and white cell differential count. Clin Chem Lab Med 2016; 54: 1619–28. doi:10.1515/cclm-2015-1258

³⁷ Ganna A, Ingelsson E. 5 year mortality predictors in 498,103 UK Biobank participants: a prospective population-based study. Lancet 2015; 386: 533–40. doi:10.1016/S0140-6736(15)60175-1

³⁸ Clift AK, Le Lannou E, Tighe CP, et al. Development and validation of risk scores for all-cause mortality for a smartphone-based “General Health Score” App: prospective cohort study using the UK biobank. JMIR Mhealth Uhealth 2021; 9: e25655. doi:10.2196/25655

³⁹ Weng SF, Vaz L, Qureshi N, et al. Prediction of premature all-cause mortality: a prospective general population cohort study comparing machine-learning and standard epidemiological approaches. PLoS One 2019; 14: e0214365. doi:10.1371/journal.pone.0214365

⁴⁰ Fuchs A, Kühl JT, Sigvardsen PE, et al. Arterial hypertension and morphologic abnormalities of cardiac chambers: results from the copenhagen general population study. J Hypertens 2021; 39: 703–10. doi:10.1097/HJH.0000000000002705

⁴¹ Timpson NJ, Harbord R, Davey Smith G, et al. Does greater adiposity increase blood pressure and hypertension risk?: Mendelian randomization using the FTO/MC4R genotype. Hypertension 2009; 54: 84–90. doi:10.1161/HYPERTENSIONAHA.109.130005

⁴² Jepsen R, Wingstrand A, Abild SL, et al. Socio-economic determinants of participation in the Lolland-Falster health study. J Public Health (Berl) 2020; 28: 657–64. doi:10.1007/s10389-019-01095-z

⁴³ Heltberg A, Andersen JS, Kragstrup J, et al. Social disparities in diabetes care: a general population study in Denmark. Scand J Prim Health Care 2017; 35: 54–63. doi:10.1080/02813432.2017.1288702

⁴⁴ Rodriquez EJ, Kim EN, Sumner AE, et al. Allostatic load: importance, markers, and score determination in minority and disparity populations. J Urban Health 2019; 96: 3–11. doi:10.1007/s11524-019-00345-5

⁴⁵ Juster R-P, Pruessner JC, Desrochers AB, et al. Sex and gender roles in relation to mental health and allostatic load. Psychosom Med 2016; 78: 788–804. doi:10.1097/PSY.0000000000000351

⁴⁶ Kudielka BM, Kirschbaum C. Sex differences in HPA axis responses to stress: a review. Biol Psychol 2005; 69: 113–32. doi:10.1016/j.biopsycho.2004.11.009