Frailty Index and Sex-Specific 6-Year Mortality in Community-Dwelling Older People: The ActiFE Study

Abstract

Background

Considering that mortality rate and deficit accumulation varies considerably in men and women, we performed a sex-stratified analysis of the association between an estimated frailty index (eFI) with 6-year mortality in the Activity and Function in the Elderly (ActiFE) Study.

Methods

We constructed an eFI using a score (0 [no deficit] to 1 [deficit]) from 32 baseline items representing multiple domains. eFI represents the sum of all scores divided by 32. Cox proportional hazards models adjusted for age, smoking, alcohol intake, and education were used to evaluate this association.

Results

Among 1,204 participants (57.5% men), 18.5% men and 26.0% women were frail (eFI ≥ 0.2) with an age-adjusted mean eFI of 0.13 (95% confidence interval [CI] 0.12, 0.13) and 0.15 (95% CI 0.15, 0.16), respectively. Mortality rate in men (146 deaths) was 34.4 (95% CI 29.3, 40.5) and in women (50 deaths) 15.1 (95% CI 11.5, 19.9) per 1,000 person-years. A 0.1 increment of eFI was associated with a hazard ratio (HR) of 1.94 (95% CI 1.60, 2.35) in men and 2.06 (95% CI 1.58, 2.69) in women. Frail versus nonfrail men and women had a HR of 2.46 (95% CI 1.74, 3.48), and 2.98 (95% CI 1.55, 5.70), respectively. We detected sex differences in the order of the eight common contributor items to the eFI.

Conclusions

We observed a statistically significant difference for the age-adjusted eFI and the frailty prevalence in men and women. However, our analysis does not suggest the presence of effect modification by sex in the association with mortality.

Frailty, Frailty index, Mortality, Older persons, Sex-specific

Frailty is considered as a state characterized by a weakened response to stressors and an increased vulnerability for many adverse health outcomes such as disability, institutionalization, and death (1,2). Aging is associated with multisystem physiological changes, where damage accumulates in cells and tissues. Although these changes may not achieve disease status, they can still put the individual in a state of increased vulnerability.

A stronger evidence base is needed to improve prioritization and informed decision making in the frail older people (3). However, the operationalization of frailty remains controversial. A widely used approach is the accumulation of deficits as a proxy measure for this syndrome. For this approach, a frailty index (FI) score is built representing the proportion of accumulated deficits from a list of potential deficits at a given time point in a given individual. The FI recognizes frailty as multifactorial and dynamic. It includes not only physical function and abilities but also cognitive and social domains and does not reflect the order of appearance of these deficits. In this model, aging can be seen as the random accumulation of deficits and damage to a complex system composed of redundant parts. The more defects accumulate, the faster the system loses its redundancy and increases vulnerability. As a result and above a specific tipping point any new insult could lead to system failure and ultimately death (4). This is known as the redundancy exhaustion hypothesis of the reliability theory of aging (5).

A recent meta-analysis has shown that frailty measured by a FI is significantly associated with mortality (6). It is also known, that women accumulate on average more deficits than men of the same age, but may show at the same time a lower risk of mortality (7). This phenomenon is known as the male–female health-survival paradox (8) and seems counterintuitive with respect to the clear association of deficit accumulation and mortality. Recognizing the need to increase our understanding of sex-specific pathways to frailty across different environments for the further development of management and prevention strategies (9,10), we evaluate the possible presence of effect modification in this relationship. We performed sex-stratified analyses identifying the main contributor items to the FI in men and women, respectively. In addition, we evaluated the association of the concept of deficit accumulation as measured by a cohort-specific FI with 6-year mortality in community-dwelling older adults.

Methods

Study Population

The ActiFE Ulm (Activity and Function in the Elderly in Ulm) study is a population-based cohort study in community-dwelling older people (≥65 years) randomly selected in Ulm and adjacent regions in Southern Germany. Between March 2009 and April 2010, 1,506 eligible individuals underwent baseline assessments. For the present analysis, 302 participants with missing data for any of the 32 variables identified for the FI were excluded. In total, 1,204 participants were considered in the analyses. All participants provided written informed consent. The ethical committee of Ulm University approved the study (application no. 318/08 and 50/12).

Covariates

Baseline assessments were completed by trained research assistants using standardized methods. Age, sex, socioeconomic factors, and comorbidity were ascertained by interview-based self-report. Height and weight were measured to calculate body mass index (BMI). Duration of school education was binary categorized (≤10 and >10 years of education). Self-report on smoking was categorized into smoker, ex-smoker, and nonsmoker, while self-report on average alcohol drinking was categorized into daily, more than once per week, more than once per month, less than once per month, and never. Multimorbidities were assessed by asking: Have you ever been diagnosed with any of the following diseases: hypertension, myocardial infarction (MI), heart failure (HF), stroke, cancer, diabetes, rheumatism, chronic lung disease, stomach or intestinal ulcers, and migraine. Cardiovascular disease was defined as having either MI, HF, or stroke. Glomerular filtration rate was estimated using the CKD-EPI formula for cystatin C (11), which was measured by immunonephelometry on a Behring Nephelometer II (interassay coefficient of variation 2.9%–3.2%). Activities of daily living (ADL) and instrumental activities of daily living (IADL) were assessed using the ADL (12) and a modified IADL questionnaire based on Lawton et al. (13). The German version of the Hospital Anxiety and Depression Scale (HADS) was used to quantify symptoms of depression and anxiety (14). Social networking was assessed using the Lubben Social Network Scale (15). Self-perception was assessed for the general health and for their own practice of physical activity. Physical performance assessment included handgrip strength (JAMAR dynamometer, Sammons Preston, Bolingbrook, IL) and the chair rising test.

Frailty Index

Based on the principles of the development of a FI, we selected 32 baseline items associated with health status, which accumulate with age but do not saturate too early (16), representing following domains: ADL, IADL, multimorbidities, psychosocial anamnesis, self-perception, risk of fall, and functional measurements (Table 1). The coding of individual variables was performed as recommended (16). For binary variables, 0 indicates the absence and 1 the presence of the deficit. For ordinal variables, the available rank was graded into a score between 0 (no deficit) and 1 (deficit is maximally expressed). For handgrip strength, we applied following cut-points: for women 21 kg, for men 33 kg (17).

Table 1.

Items Included in the Frailty Index

Item	Variable	Score
Level of difficulty by performing following activities of daily living
1	Going up/down stairs	1 point = not possible without help 0 point = none dependency
2	Grooming
3	Dressing
4	Getting in/out of chair
5	Walking around the house
6	Bathing
7	Using (public) transport
8	Shopping
9	Taking medication
10	Doing housework
Have you being said by a doctor that you have any of the following diseases
11	Hypertension	Yes = 1 point No = 0 point
12	Cardiovascular disease (myocardial infarction, heart failure, or stroke)
13	Cancer
14	Diabetes
15	Rheumatism
16	Chronic lung disease
17	Stomach or intestinal ulcers
18	Migraine
Others
19	Do you have problem with urinary continence?	1 point
20	Taking 5 or more medications	1 point
21	Falls in the last 12 months	1 point
Emotions
22	HADS-Depression scale	<8 (0 point); (8, 10) (0.5 points); >10 (1 point)
23	HADS-Anxiety scale	<8 (0 point); (8, 10) (0.5 points); >10 (1 point)
24	Social activity - LUBBEN scale	<12 (1 point)
25	Living alone	1 point
Self-perception: How do you describe your own…
26	Health	Excellent/very good/good (0 point), less good (0.5 point), bad (1 point)
27	Level of physical activity	Very active till average (0 point), less than average (0.5 points), too less (1 point)
Measurements
28	Weight loss in the last three months	More than 1 kg (1 point)
29	Chronic kidney disease (mL/min/1.73 m2)	eGFR < 60 (1 point)
30	BMI (kg/m2)	<18.5 or ≥30 (1 point)
31	Handgrip strength (kg)	male: <33; female: < 21 (1 point)
32	Five-chair rising test (seconds)	<11.2 (0 point), (11.2, 13.7) (0.25 points), (13.7, 16.7) (0.5 points), ≥ 16.7 (0.75 points), unable to perform (1 point)

Note: BMI = Body mass index; eGFR = Estimated glomerular filtration rate; HADS = Hospital Anxiety and Depression Scale.

Assessment of Deaths

Mortality status and date of death were obtained from the local registration offices for all participants 6 years after the baseline measurement. Time to death or censoring was calculated for each participant.

Statistical Analysis

Knowing that the accumulation of deficits showed different patterns and that the mortality rate varies considerably between men and women (4,16,18), we performed a sex-stratified analysis. Descriptive analyses for numerical variables are expressed as mean ± standard deviation or median (Q1, Q3) depending on their distribution, and categorical variables as percentages. According to their prevalence, we identified the top 10 contributor items to the FI for both sexes. We estimated the sex-specific median increase of the FI with age using linear quantile regression (19–21). Six-year survival estimation was performed using Kaplan–Meier curves for frail (FI ≥ 0.2) versus nonfrail (FI < 0.2) participants (16), where a cut-point of 0.2 has been validated for the identification of frail community-dwelling seniors (22). Differences in survival probability were assessed applying the log-rank test. Cox proportional hazard models evaluated the association between the estimated FI (eFI), as continuous and categorical, and 6-year mortality stratified by sex after adjustment for age, years of school education, smoking, and alcohol consumption. The proportional hazard assumption was evaluated by inclusion of time-dependent covariates in the multivariable models. Hazard ratios (HR) are reported with their 95% confidence intervals (CI). Using Venn Diagrams, we examined the contribution of disease burden (≥2 multimorbidities), impaired functionality (handgrip strength <21 kg in women and <33 kg in men (17); five-chair rising test ≥16.7 seconds or not able to perform) and dependency in ≥1 items to the eFI among the identified frail individuals. All calculations were performed using SAS 9.2 and R software version 3.4.0.

Results

Our study sample includes 1,204 older adults (median age 74.0 years [Q1 70.1, Q3 81.1], 57.5% men). Table 2 shows participants’ baseline characteristics stratified by sex. Men were more likely to be older, current or former smokers, drink more often alcohol, had increased systolic blood pressure and BMI, a higher prevalence of MI, cancer, and diabetes. Women were noted to live more often alone, to have higher prevalence of migraine, urinary incontinence and polypharmacy, more anxiety symptoms, and needed more time on average for the chair rise test. A higher proportion of men attend school for >10 years compared to women.

Table 2.

Participant Characteristics Stratified by Sex

	Men n = 692	Women n = 512
Age (years), median(Q1, Q3)	74.4 (70.2, 81.6)	73.1 (69.6, 79.5)
School education > 10 y, n (%)	184 (26.7)	65 (12.9)
Smoker, n (%)
Never	228 (33.0)	364 (71.1)
Former	398 (57.5)	128 (25.0)
Current	66 (9.5)	20 (3.9)
Alcohol Consumption, n (%)
Daily	300 (43.6)	72 (14.6)
More than once a week	186 (27.0)	119 (24.1)
More than once a month	152 (22.1)	147 (29.8)
Less than once a month	46 (6.7)	148 (30.0)
Never	4 (0.6)	7 (1.4)
Systolic blood pressure (mmHg), mean (SD)	143.5 (11.6)	139.9 (12.7)
Diastolic blood pressure (mmHg), mean (SD)	79.2 (11.0)	78.8 (10.6)
BMI (kg/m2), mean (SD)	27.8 (3.6)	27.3 (4.7)
eGFR (mL/min/1.73 m2), median (Q1, Q3)	84.0 (67.9, 99.6)	86.3 (68.7, 96.6)
Comorbidities, n (%)
Hypertension	383 (55.4)	264 (51.6)
Cardiovascular disease	186 (26.9)	101 (19.7)
Myocardial infarction	79 (11.4)	26 (5.1)
Heart failure	104 (15.0)	68 (13.3)
Stroke	37 (5.4)	22 (4.3)
Cancer	149 (21.5)	75 (14.7)
Diabetes	107 (15.5)	51 (10.0)
Rheuma	63 (9.1)	62 (12.1)
Chronic lung disease	78 (11.3)	51 (10.0)
Migraine	73 (10.6)	109 (21.3)
Urinary incontinence	163 (23.6)	274 (53.5)
Polypharmacy (≥5 medications), n (%)	225 (32.5)	139 (27.2)
Living alone, n (%)	111 (16.0)	170 (33.2)
HADS-D, median (Q1, Q3)	3 (2, 5)	3 (2, 5)
HADS-A, median (Q1, Q3)	3 (1, 5)	4 (2, 6)
LUBBEN, mean (SD)	16.2 (5.5)	16.3 (5.4)
Fall in the last 12 mo, n (%)	223 (32.2)	194 (37.9)
Handgrip strength (kg), median(Q1, Q3)	39.0 (32.8, 44.0)	23.5 (20.0, 28.0)
Available Chair rise (seconds), median (Q1, Q3)	11 (9, 13)	11 (9, 13)
unable to perform chair rise, n (%)	24 (3.5)	21 (4.1)
Deaths, n (%)	146 (21.1)	50 (9.8)
Mortality rate per 1,000 Person-years (95% CI)	34.4 (29.3, 40.5)	15.1 (11.5, 19.9)
Age-adjusted mean eFI (95% CI)	0.13 (0.12, 0.13)	0.15 (0.15, 0.16)
eFI ≥0.2, n (%)	128 (18.5)	133 (26.0)

Note: BMI = Body mass index; CI = Confidence interval; eFI = Estimated frailty index; eGFR = Estimated glomerular filtration rate; HADS = Hospital Anxiety and Depression Scale.

We observed a total of 196 deaths, 146 among men and 50 among women representing a mortality rate of 34.4 (95% CI 29.3, 40.5) and 15.1 (95% CI 11.5, 19.9) per 1,000 person-years, respectively. Furthermore, there was a statistically significant difference for the age-adjusted eFI, with a mean of 0.13 (95% CI 0.12, 0.13) for men and of 0.15 (95% 0.15, 0.16) for women. We observed a steady increment of the eFI for both sexes, with a steeper course in women compared to men (Figure 1). Using the cut-point of ≥0.2, we identified a significant difference in the proportion of frail individuals between men (n = 128, 18.5% [95% CI 14.8, 21.4]) and women (n = 133, 26.0% [95% CI 19.7, 28.3]).

Figure 1.

Sex-stratified frailty index against age at baseline.

Characteristics of Participants Stratified for Frailty

Those participants identified as frail (eFI ≥ 0.2) (128 men and 133 women) were markedly different from their counterpart (Supplementary Table 1). Overall, frail subjects were noted to be older, had a higher prevalence of the assessed multimorbidities, polypharmacy and falls in the last 12 months, had an increased BMI, higher scores for depression and anxiety, lived more often alone, had lower estimated glomerular filtration rate, handgrip strength, and LUBBEN scores. They also needed more time to perform the chair rise test. Systolic blood pressure was elevated on average only among frail women compared to nonfrail, while diastolic blood pressure was noted to be lower only among frail men compared to nonfrail.

Frailty and 6-Year Mortality

The mortality rate for frail men was 96.4 (95% CI 75.3, 123.4) versus 23.2 (95% CI 18.7, 28.7) per 1,000 person-years in nonfrail ones (log-rank test, p < .001). Women showed a mortality rate of 40.1 (95% CI 28.3, 56.7) and 7.2 (95% CI 4.5, 11.4) per 1,000 person-years when comparing frail versus nonfrail participants (log-rank test, p < .001; Figure 2). We observed an HR of 1.94 (95% CI 1.60, 2.35) in men and a HR of 2.06 (95% CI 1.58, 2.69) in women for 6-year mortality for every 0.1 increment of the FI after adjustment for age, years of school education, smoking, and alcohol consumption. The multivariable analysis also showed a higher risk for 6-year mortality in both sexes for those with eFI ≥0.2 when compared with those with eFI <0.2 (HR 2.46 [95% CI 1.74, 3.48] in men, HR 2.98 [95% CI 1.55, 5.70] in women; Table 3).

Table 3.

Cox Proportional Hazards Model for the Association Between Frailty and 6-Year Mortality

	HR (95% CI)
	Men (n = 684, 146 deaths)	Women (n = 487, 47 deaths)
eFI as continuous for a 10% incrementa	1.94 (1.60, 2.35)	2.06 (1.58, 2.69)
eFI ≥ 0.2a	2.46 (1.74, 3.48)	2.98 (1.55, 5.70)

Note: CI = Confidence interval; eFI = Estimated frailty index.

^aAdjusted for age, years of school education, smoking, and alcohol consumption.

Figure 2.

Sex-stratified survival curves according to frailty status. (a) Men, (b) women.

Table 4 shows the list of the top 10 contributor items to the eFI for both sexes according to their prevalence. Although men and women share eight of those, their ranking position differs. While hypertension is the number one contributor among men, urinary incontinence was the most prevalent deficit among women. Three of the common eight contributors referred to physical functioning: chair rise test, handgrip strength, and falls in the last 12 months. Polypharmacy is also present as a surrogate for the presence of multimorbidities. While living alone shows a higher prevalence in women, having decreased social interaction plays a more important role in men.

Table 4.

Top 10 Contributing Items to the Frailty Index

Men	Women
1. Hypertension	1. Urinary incontinence
2. Chair rise test	2. Hypertension
3. Polymedication	3. Chair rise test
4. Falls in the last 12 mo	4. Falls in the last 12 mo
5. Cardiovascular disease	5. Living alone
6. Handgrip strength	6. Handgrip strength
7. Body mass index	7. Polymedication
8. Urinary incontinence	8. Body mass index
9. Cancer	9. Migraine
10. LUBBEN scale	10. Cardiovascular disease

Figure 3a shows the relative contribution of functionality, multimorbidity, and dependency items to frailty among all 261 participants identified as frail (eFI ≥ 0.2). The combination of multimorbiditiy and impaired functionality is present in 35.3%, while all three domains are affected in 28.0% of the identified frail participants. Multimorbidity, impaired functionality, and dependency contribute as separate entities by 23.4, 1.5, and 0.4%, respectively. We observed in 20.3% of 128 frail men, the concurrent contribution of these three domains, whereas 46.1% were categorized as frail due to concomitant deficits in functionality and multimorbidities. Only 1.6% men were categorized as frail due to a deficit in function (Figure 3b), whereas 23.4% men showed ≥2 morbidities. No men reported deficit with respect to dependency. Among 133 frail women, all three factors were present in 35.3%, whereas the combination of deficits in functionality and multimorbidity was noted in 24.8%. We identified 3.8% with deficit in functionality and dependency, as well as 9.0% women with ≥ 2 morbidities and dependency. Similar to men impaired multimorbidity, impaired functionality, and dependency contribute as separate entities by 23.3, 1.5, and 0.8%, respectively (Figure 3c).

Figure 3.

Venn Diagrams with contributing domains to frailty among (a) frail participants (n = 261), (b) frail men (n = 128), and (c) frail women (n = 133).

Discussion

To our knowledge, this is the first study evaluating the presence of effect modification by sex in the association between frailty, measured as a FI based on the concept of accumulation of deficits, and 6-year mortality in community-dwelling older people. Our results showed a statistically significant difference for age-adjusted eFI by men and women, with a respective significant difference in the frailty prevalence when using a cutoff of 0.2. A different ranking order between both sexes was also noted for the top 10 contributor items to the eFI. However, and contrary to our expectations, our analysis does not suggest the presence of effect modification by sex in the association of the eFI and 6-year mortality.

The observed monotonic increase of the eFI with age can be considered as an indicator of construct validity (23). Our results are also consistent with the literature showing women accumulating more deficits than men, even though men are known to have higher mortality rate (24–26). The overall identified frailty prevalence of 21.7% (18.5% for men and 26.0% for women) is slightly higher than the 17% prevalence observed in the German ESTHER cohort study (23), which also used a deficit accumulation approach. Reported levels of frailty prevalence in community-dwelling older adults vary according to the definition used. A systematic review showed that for the definition according to the physical phenotype, the prevalence varies from 4.0% to 17.0% with a weighted average prevalence of 9.9%, while using broader frailty definitions is associated with a prevalence range from 4.2% to 59.1% and a weighted average prevalence of 13.6% (27).

Longer life span among women has been described since medieval age (28). Although women live longer than men, they tend to have poorer health status reflected by greater levels of disability, more physical and psychological comorbidities and worse self-rated health (4), a phenomenon known as the male–female health-survival paradox. In our sample we also observed a steeper increase of the eFI with age in women compared to men, and, in parallel, a slightly higher 6-year mortality risk for each 10% increment of the eFI in women compared to men.

For the observed differences in the ranking of the top 10 contributors to the eFI, following considerations should be mentioned: reporting bias may play a role where men could tend to under-report medical conditions, especially the ones related to anxiety and depression, while at the same time the lifetime risk of developing these mood disorders is higher in women than in men (29). They also tend to report fewer limitations in their ADLs (30). With respect to multimorbidities, and as seen in our data, men tend to experience more cardiovascular diseases, with a higher mortality risk, while women seem to be more affected by other chronic conditions (31). Nevertheless we observed overall a higher prevalence of all addressed diseases in frail individuals compared with nonfrail ones. The identification of living alone as an important contributor item in women goes to the fact that women have been noted to have higher levels of social vulnerability than men (32). However, not living alone but rather a decreased social interaction measured through the LUBBEN scale was an important contributor item in men.

When evaluating multimorbidities, impaired functionality and dependency as driving components of the eFI, 28% of our study population were affected in all three domains. In this regard, the concomitant presentation of ≥ 2 diseases seems to be the main component either isolated, present in 23.4% of the frail individuals, or in combination, affecting other 71.3% of them. Impaired functionality play an important role in combination with multimorbidity and dependency, affecting in this way 65.9% of the frail subjects. Similar distributions can be observed when analyzing men and women separately. Isolated deficits in functionality or dependency were almost not observed in our study sample. One can argue that this is related to a relative healthy population, arising from the community. On the other hand, our observed data can be explained in the context of the (extended) disablement process model by Verbrugge and Jette (33), which describes the pathway from disease to disability. According to this model, frailty should be seen as the consequence of pathophysiological changes observed by aging—represented by a decline in the functional reserves till lastly the appearance of multimorbidities, which accelerates the development of functional decline, leading finally to disability (33). In this context, the occurrence of diseases can be seen as the accumulation of biological abnormalities that become clinically relevant, and therefore should be included in the evaluation of frailty. The rapid occurrence of multimorbidities, with a possible acceleration of the development of functional decline, could contribute to a rapid development of frailty. Thanks to its multidimensional construct, a FI based on a deficit accumulation model allows also the identification of frail individuals without multimorbidities.

Currently, there is an urgent need to change our approach so that medical management of older adults should rather implement therapeutic strategies to retain or improve functionality and not only focus on the cure of specific diseases (34). The use of such a multidimensional FI could help to identify those relevant sex-specific areas of deficit associated with the development of frailty in the ambulatory settings.

Strengths and Limitations

Our study is based on a cohort of 1,204 participants aged ≥ 65 years. Even though the ActiFE-cohort was intended to be a representative cohort of community-dwelling older adults, the observed health status was above average; therefore, we have to assume some underrepresentation of the most ill people. Although this may have led to an underestimation of frailty prevalence compared to the general population, our estimates are in concordance with the literature. We compared those included in the development of the FI with those excluded (n = 302). Having a higher mortality rate excluded subjects showed significant differences in only four items: they were older, lived more often alone with lower scores for the LUBBEN scale, and had a higher prevalence for urinary incontinence (Supplementary Table 2). In this context, the observed distribution of the eFI, and its variability along ages can be seen as indicators for a good construct of our index. In addition, our eFI, contrary to most frailty research, not only focuses on disease and functionality, but also takes into account the participant perspective by including information related to self-perception of health and their own practice of physical activity (3). Due to the study design with a balanced presence of both sexes across all ages group we have been able to evaluate not only the possible presence of effect modification by sex in the association between frailty and 6-year mortality but also differences in the characterization of frail men and women.

Conclusion

Our prospective study in community-dwelling older adults showed significant differences in the age-adjusted median eFI, as well as in the prevalence of frailty in men and women. Although frail men seem to have a higher 6-year mortality risk than women, we do not detect differences in the observed risk for every 0.1 increment of the eFI between both sexes. Furthermore, 8 of the top 10 contributor items to the eFI were the same for men and women, but differences in their ranking order could be noted. Multimorbidity could be identified as an important contributor domain to frailty in our study sample, while decreased functionality and dependency played combined an important role. Further longitudinal studies analyzing the contribution of these three domains to a FI among frail participants could help to further deepen our understanding about possible pathophysiological mechanisms behind the development of frailty in older adults.

Funding

This work was supported by a grant from the Ministry of Science, Research and Arts, state of Baden-Wuerttemberg, Germany, as part of the Geriatric Competence Center, Ulm University. Michael Denkinger and Kilian Rapp received a “Forschungskolleg Geriatrie” grant from the Robert Bosch Foundation, Stuttgart, Germany. None of the sources had any influence on the content.

Conflict of Interest

None reported.

Author notes

The ActiFE study group consists of further members: B. Böhm, Lee Kong Chian School of Medicine, Nanyang Technological University, Singapore and Imperial College London, UK; H. Geiger, Institute for Molecular Medicine, Ulm University, Germany; A. Lukas, Malteser Krankenhaus Seliger Gerhard Bonn/Rhein-Sieg, Germany; J. Stingl, Institute of Pharmacology of Natural Products & Clinical Pharmacology and Federal Institute for Drugs and Medical Devices, Bonn, Germany; M. Riepe, Division of Gerontopsychiatry, Department of Psychiatry and Psychotherapy II, Ulm University Medical Center, Germany; K. Scharffetter-Kochanek, Department of Dermatology and Allergology, Ulm University Medical Center, Germany; W. Koenig, Deutsches Herzzentrum München, Germany; R. Laszlo, J.M. Steinacker, Department of Internal Medicine II - Sports- and Rehabilitation Medicine, Ulm University Medical Center, Germany; A. Ludolph, C. von Arnim, Department of Neurology, Ulm University Medical Center, Germany; G. Nagel, G. Weinmayr, Institute of Epidemiology and Medical Biometry, Ulm University, Germany; R. Peter, Institute of the History, Philosophy and Ethics of Medicine, Ulm University, Germany.