End of Life Events and Causes of Death in Danish Long-Lived Siblings: Reduced Dementia Risk Compared to Sporadic Long-Livers

Abstract

Background:

Better physical robustness and resilience of long-lived siblings compared to sporadic long-livers has been demonstrated in several studies. However, it is unknown whether long-lived siblings also end their lives better.

Objective:

To investigate end-of-life (EoL) events (dementia diagnosis, medication, hospitalizations in the last 5 years of life), causes of death, and location of death in long-lived siblings compared to matched sporadic long-livers from the Danish population.

Methods:

Long-lived siblings were identified through three nationwide Danish studies in which the inclusion criteria varied, but 99.5% of the families had at least two siblings surviving to age 90 + . Those who died between 2006 and 2018 were included, and randomly matched with sex, year-of-birth and age-at-death controls (i.e., sporadic long-lived controls) from the Danish population.

Results:

A total of 5,262 long-lived individuals were included (1,754 long-lived siblings, 3,508 controls; 63% women; median age at death 96.1). Long-lived siblings had a significantly lower risk of being diagnosed with dementia in the last years of life (p = 0.027). There was no significant difference regarding the number of prescribed drugs, hospital stays, days in hospital, and location of death. Compared to controls, long-lived siblings presented a lower risk of dying from dementia (p = 0.020) and ill-defined conditions (p = 0.030).

Conclusions:

In many aspects long-lived siblings end their lives similar to sporadic long-livers, with the important exception of lower dementia risk during the last 5 years of life. These results suggest that long-lived siblings are excellent candidates for identifying environmental and genetic protective factors of dementia.

Keywords

Aging Alzheimer’s disease causes of death dementia end of life events familial longevity

INTRODUCTION

Across the world, individuals are living longer, and the human life expectancy at birth has more than doubled over the past two centuries [1]. Some individuals, called long-lived individuals, become exceptionally old, and this tends to cluster in families [2, 3].

Family-based studies on long-lived siblings showed better physical robustness (i.e., the ability to resist a deviation from the normal physical state and avoid an adverse health event) and resilience (i.e., the ability to bounce back, and quickly and completely recover after an adverse health event) in long-lived siblings than sporadic long-livers (i.e., individuals living long without a family history of longevity) [4]. We lately reported a lower cumulative incidence of depression and cardiovascular diseases (cerebrovascular diseases, atrial fibrillation, and flutter), and better survival in Danish long-lived female siblings compared to sporadic female long-livers [5]. This survival advantage of long-lived siblings was also demonstrated in the Leiden Longevity study [6]. A US study similarly reported a lower prevalence of Alzheimer’s disease and related disorders, hip fracture, diabetes, depression, heart failure and chronic kidney disease in long-lived siblings compared to similarly aged controls [7]. These results suggest that long-lived siblings have lower morbidity and better survival compared to sporadic long-livers of similar age, and hence provide evidence that long-lived siblings are particularly well suited as a model for exceptional health and survival. However, it is unknown whether long-lived individuals also end their lives better. The physical robustness and resilience found in previous studies of long-lived siblings could result in different health trajectories in the last years of life and different causes of death for these long-lived siblings compared to sporadic long-livers.

While death is the final event, studies suggest that medication, hospitalizations and place of death impact the quality of ending life and dying. Indeed, long, repeated or unplanned hospital admissions at the end of life (EoL), as well as death at the hospital could be an indicator of a low quality of ending life and dying [8 –10]. In addition, studies reported that older adults with dementia receive poorer quality of EoL care concerning several important care areas [11] suggesting that being diagnosed with dementia further contributes to a low quality of ending life and dying [12, 13].

Using data from three nationwide studies on long-lived siblings in Denmark and the Danish national health and civil registers, we investigated EoL events (dementia diagnosis, medication, and hospitalizations), causes of death, and location of death in long-lived siblings compared to matched sporadic long-livers from the Danish population.

METHODS

Study population

The identification of long-lived siblings was undertaken in three consecutive, nationwide studies in Denmark: the Danish Oldest Siblings (DOS) pilot study, the Genetics of Healthy Ageing (GeHA) study [14], and the Danish part of the Long Life Family Study (LLFS) [15, 16]. Recruitment for these studies ran sequentially during the years 2004 to 2009, where all individuals born before April 2, 1918, and alive in 2004 were identified in the Danish Civil Registration System (DCRS). Long-lived siblings were defined in different ways depending on the study: i) recruitment to DOS was conditional on both siblings being alive and 88 years or older; ii) recruitment to GeHA required both siblings to be alive and above age 90, and iii) the LLFS recruited only families with a family longevity selection score (FLoSS) above 7 [16]. Despite differences in the criteria to define long-lived individuals, 99.5% of the enrolled families included at least two siblings who survived to age 90+, whereas for the few remaining families, one sibling survived to age 90 + and at least one other sibling survived to age 89.

A total of 3,972 siblings from 659 families were enrolled in either DOS (659 siblings from 114 families), GeHA (2,736 siblings from 469 families), or LLFS (577 siblings from 76 families). Among them, 33.6% were interviewed, while the other members of each sibship were identified through the Danish civil and health registration system, and full sibships were included.

Long-lived siblings enrolled in these three studies who died between 2006 and 2018 were included. For each sibling, two controls (i.e., sporadic long-liver controls) matched on sex, year-of-birth and age-at-death were randomly selected from the Danish population.

Danish National Registers

The information used in this study was based on pedigree data obtained in the three studies (DOS, GeHA, LLFS) and combined with data from the Danish national population-based registers presented below.

The Danish Civil Registration System (DCRS)

The DCRS, which covers the entire population alive and residing in Denmark since April 2, 1968, contains information on each resident’s vital status, sex, place and date of birth and death [17, 18]. All persons registered in DCRS are assigned a unique personal identification number used in all national registers, enabling accurate linkage between all national registers. Once a person has been assigned a unique personal identification number, the same number will not be given to other persons, and this number follows the person afterwards.

Health registers

The Danish National Patient Register (NPR) is a health register established in 1977 [19, 20] covering inpatient hospitalizations and somatic wards. Since 1995, NPR also covers outpatient, emergency room visit, and psychiatric wards. The reported data are administrative (e.g., patient’s municipality, identification of hospital ward, date and time of activity, and information on accidents leading to hospital contact) and clinical (e.g., diagnoses and surgical procedures). Different types of diagnoses are recorded: primary diagnoses (the main reason for hospitalization), secondary diagnoses (supplementing the primary diagnosis), referral diagnoses (reason for referral), temporary diagnoses and complications.

The Danish National Prescription Registry (DNPR) has provided individual-level information on dispensed prescriptions for each Danish resident since 1995 [21, 22]. DNPR contains information on all prescription drugs dispensed at Danish community pharmacies as well as prescriptions dispensed to residents of long-term care institutions (e.g., nursing homes). The register records information related to drug user and to prescriber, as well as drug and pharmacy information.

Danish Register of Causes of Death (DRCD)

The DRCD covers all deaths among citizens dying in Denmark since 1970 and contains information on the deceased person, the death (date, place), the manner of death, postmortem examinations, surgical intervention, cause of death (underlying cause, contributory causes) and the physician issuing the death certificate [23].

Outcomes

Regarding EoL events, we focused on dementia diagnosis, the number of prescribed drugs, and hospitalizations in the 5 last years of life. Dementia diagnosis in the 5 last years of life was identified in the NPR and DRCD using ICD10 codes (F00.0-F03.9; G30.0-G30.9) based on a previous study [24]. The number of prescribed drugs at death, 1 year, 2 years, 3 years, 4 years, and 5 years prior to death was identified through the DNPR containing individual-level information on all prescription drugs dispensed at Danish community pharmacies. Drugs were considered different when their Anatomical Therapeutic Chemical (ATC) codes differed at the fourth coding. Several hospitalization outcomes were studied: having been hospitalized, number of hospital stays, number of inpatient days at hospital, and number of emergency room (ER) visits during the year, the second year, the third year, the fourth year, and fifth year prior to death. Hospitalization outcomes were identified using NPR.

The cause and location of death were also investigated through DRCD. Only underlying causes of death were considered, coded in DRCD by an international standard for coding, automated classification of medical entities [23]. The causes of death of relevance were based on a categorization previously defined for Danish centenarians [25], i.e., cardiovascular disease (ICD10 codes I10-I15, I20-I25, I27, I28, I30-I52, I60-I79), respiratory disease (J00-J06, J10-J11, J13-J18, J20-J22, J30- J47, J60-J99), cancer (C00-C34, C37-C97, D00-D09), digestive disease (K00-K31, K70-K83, K35-67, K85- K93), urinary and genital disease (N00-N08, N20-N23, N10-N19, N25-N99), endocrine disorders (E00-E07, E10-E90), dementia (F00.0-F03.9; G30.0-G30.9), ill-defined conditions (R00-R99), accidents (V01-V99, W00-X59, Y40-Y86, Y87.1-87.2), and other.

Covariates

Marital status (unmarried, married, divorced, widowed) was considered at death and identified using the DCRS.

The Charlson Comorbidity Index (CCI) [26], a weighted (from 1 to 6) index considering 19 conditions (e.g., dementia, ulcer disease, tumors, AIDS, etc.), was constructed from hospital data at each year prior to death and was based on primary and secondary disease diagnoses recorded in the NPR and using ICD10 classification.

Statistical analyses

The study population was described in terms of age at death, sex, marital status and CCI.

Differences between long-lived siblings and sporadic long-livers in categorial outcomes (dementia, hospitalizations, cause of death, and place of death) were studied by considering each outcome bivariate (except for place of death, where dying at home was the reference) and performing unadjusted conditional logistic regression models based on matching data. Continuous outcomes (no. of hospital stays, days at hospital, and ER visits) were studied by performing clustered Poisson regression models.

Multiple testing was corrected using the False Discovery Rate method [27, 28].

Ethics and data protection

The study has been approved by The Regional Scientific Ethical Committees for Southern Denmark (S-VF-20030227), The Danish Data Protection Agency (J.nr. 2008-41-1753), and University of Southern Denmark, Research & Innovation Organisation (J.nr. 10 635).

RESULTS

Study population (Table 1)

Table 1

Population description (n = 5,262)

	Study population (n = 5,262)		Siblings (n = 1,754)		Controls (n = 3,508)
	n	(%)	n	(%)	n	(%)	p
Sex							NA
Female	3,324	(63.2)	1,108	(63.2)	2,216	(63.2)
Male	1,938	(36.8)	646	(36.8)	1,292	(36.8)
Marital status							<0.001
Unmarried	337	(6.4)	134	(7.6)	203	(5.8)
Married	649	(12.3)	233	(13.3)	416	(11.9)
Divorced	263	(5.0)	54	(3.1)	209	(5.9)
Widowed	4,013	(76.3)	1,333	(76.0)	2,680	(76.4)
	Median	(Range)	Median	(Range)	Median	(Range)	p
Age at death	96.1	(75.2–110.7)	96.1	(75.2–110.7)	96.1	(75.5–110.4)	NA
Charlson Comorbidity Index at death	1	(0–9)	1	(0–8)	1	(0–9)	0.325

NA, not applicable.

A total of 5,262 long-lived individuals (1,754 long-lived siblings, 3,508 controls) were included. Among them, 3,324 were women (1,108 long-lived siblings, 2,216 controls) and 1,938 were men (646 long-lived siblings, 1,292 controls). The median age at death was 96.1 (range = 75.2–110.7); 96.7 (range 76.7–110.7) in women and 95.0 (range 75.2–105.3) in men. There was no difference regarding age at death between siblings and controls as individuals were matched on this criterion. At death, individuals were mainly widowed (76.3%). Marital status was significantly different between long-lived siblings and controls, with a larger proportion of divorced individuals in controls than in siblings (5.9% % in controls, 3.1% in siblings). Marital status also differed between women and men, with nearly 85% of women widowed at death, compared to 62% of men.

The median CCI score at death was 1 (range 0–9), with no difference between siblings and controls (p = 0.325 at death).

EoL events (Table 2)

Table 2

End-of-life events in long-lived siblings compared to sporadic long-livers (n = 5,262)

	Siblings	Controls	OR	95% CI	p
	n	n
Dementia	155	379	0.80	(0.66–0.97)	0.027
Hospitalizations
5th year prior to death	471	917	1.04	(0.91–1.18)	0.577
4th year prior to death	522	991	1.08	(0.95–1.22)	0.252
3rd year prior to death	538	1,098	0.97	(0.86–1.10)	0.644
2nd year prior to death	589	1,095	1.11	(0.99–1.26)	0.082
Year prior to death	1,129	2,151	1.15	(1.02–1.30)	0.027 ^a,b
	Mean (SD)	Mean (SD)	Coefficient	95% CI	p
Medication
5 years	4.9 (3.6)	5.0 (3.5)	–0.02	(–0.06–0.02)	0.295
4 years	5.2 (3.7)	5.3 (3.6)	–0.02	(–0.06–0.02)	0.329
3 years	5.6 (3.7)	5.7 (3.7)	–0.02	(–0.06–0.02)	0.268
2 years	6.0 (3.9)	6.2 (3.0)	–0.03	(–0.06–0.01)	0.148
1 year	6.6 (3.9)	6.6 (3.8)	0.00	(–0.03–0.04)	0.785
At death	9.2 (4.4)	8.8 (4.5)	0.05	(0.02–0.08)	0.001 ^a,b
Hospital stays
5th year prior to death	0.4 (0.9)	0.4 (0.8)	0.05	(–0.07–0.17)	0.410
4th year prior to death	0.5 (0.9)	0.4 (0.9)	0.04	(–0.08–0.16)	0.503
3rd year prior to death	0.5 (0.9)	0.5 (0.9)	0.00	(–0.11–0.11)	0.966
2nd year prior to death	0.6 (1.0)	0.5 (1.0)	0.07	(–0.04–0.17)	0.232
Year prior to death	1.3 (1.5)	1.2 (1.5)	0.06	(–0.01–0.13)	0.090
Days in hospital
5th year prior to death	2.6 (7.3)	2.6 (7.8)	0.00	(–0.16–0.16)	0.976
4th year prior to death	2.9 (7.8)	2.9 (8.1)	0.00	(–0.16–0.15)	0.955
3rd year prior to death	2.8 (7.2)	3.1 (8.5)	–0.11	(–0.26–0.04)	0.150
2nd year prior to death	3.3 (9.1)	3.1 (8.3)	0.06	(–0.10–0.21)	0.476
Year prior to death	8.0 (12.5)	7.8 (12.9)	0.02	(–0.06–0.11)	0.588
Emergency room visits
5th year prior to death	0.3 (0.8)	0.3 (0.8)	0.00	(–0.16–0.15)	0.981
4th year prior to death	0.3 (0.8)	0.3 (0.8)	–0.06	(–0.21–0.09)	0.430
3rd year prior to death	0.3 (0.8)	0.3 (0.8)	–0.05	(–0.19–0.09)	0.491
2nd year prior to death	0.4 (0.8)	0.4 (0.9)	–0.03	(–0.17–0.10)	0.602
Year prior to death	0.6 (1.2)	0.7 (1.2)	–0.06	(–0.17–0.04)	0.213

^aNot significant after correcting for multiple testing; ^bNot significant after adjusting for Charlson Comorbidity Index at death.

In the last 5 years prior to death, 155 long-lived siblings (8.8%) and 379 controls (10.8%) were diagnosed with dementia. Long-lived siblings had a significantly lower risk of being diagnosed with dementia in the 5 last years of life (OR = 0.80 95% CI = 0.66–0.97).

The average number of prescribed drugs increased in both long-lived siblings and controls with proximity to death (Fig. 1), from 4.9 (SD 3.6) at 5 years prior to death to 9.2 (SD 4.4) at death in long-lived siblings and from 5.0 (SD 3.5) to 8.8 (SD 4.5) in controls. At death, long-lived siblings presented a slightly higher number of prescribed drugs than controls (estimate = 0.05 95% CI = 0.02–0.08), but the association did not remain significant after correcting for multiple testing.

Fig. 1

Average number of prescribed drugs at death, 1 year, 2 years, 3 years, 4 years, and 5 years prior to death in long-lived siblings and controls.

While nearly a quarter (26.9% in siblings, 26.1% in controls) of long-lived siblings and controls were hospitalized during the fifth year prior to death, there were near two thirds (64.4% in siblings, 61,3% in controls) during the last year prior to death, including the hospitalizations ending with death (38.6% in siblings, 40.9% in controls). Siblings were more frequently hospitalized than controls during the last year of life (OR = 1.15 95% CI = 1.02–1.30) however, this association did not remain significant after adjusting for CCI nor after controlling for multiple testing. The number of hospital stays and days in hospital increased during the last year of life in both siblings and controls (Fig. 2). It was not significantly different between siblings and controls (hospital stays the year prior to death estimate = 0.06 95% CI = –0.01–0.13, days in hospital the year prior to death estimate = 0.02 95% CI = –0.06–0.110). Similar results were found regarding ER visits, with no significant difference between siblings and controls at any time.

Fig. 2

Average numbers of A) hospital stays (inpatients only) and B) days in hospital during the year, 2nd year, 3rd year, 4th year, and 5th year prior death in long-lived siblings and controls.

Cause and location of death (Table 3)

Table 3

Risk of dying from different causes between long-lived siblings compared to sporadic Danish long-livers (n = 5,262)

	Siblings n	Controls n	OR	95% CI	p
Cause of death
Cardiovascular disease	664	1,234	1.12	(0.99–1.26)	0.07
Respiratory disease	212	410	1.04	(0.87–1.24)	0.67
Cancer	188	296	1.30	(1.07–1.58)	0.01 ^*
Digestive diseases	57	124	0.84	(0.61–1.16)	0.29
Urinary &genital disease	35	95	0.73	(0.49–1.08)	0.12
Endocrine disorders	47	111	0.84	(0.59–1.19)	0.33
Dementia	118	300	0.76	(0.62–0.97)	0.02 ^*
Ill–defined conditions	222	521	0.83	(0.70–0.98)	0.03 ^*
Accidents	56	111	1.01	(0.73–1.40)	0.96
Place of death
Home	298	561	1	Reference
Hospital or hospice	441	897	0.93	(0.77–1.11)	0.394
Nursing home	689	1,348	0.96	(0.82–1.13)	0.646
Other	83	150	1.04	(0.77–1.41)	0.792

^*Non-significant after correcting for multiple testing.

Cardiovascular diseases were the main cause of death in both long-lived siblings and sporadic long-livers, accounting for 37.9% of deaths in siblings and 35.2% of deaths in controls, and the risk of dying from cardiovascular disease was higher, but non-significant, among long-lived siblings (OR = 1.12 95% CI = 0.99–1.26). In addition, long-lived siblings had a higher risk of dying from cancer (OR = 1.30 95% CI = 1.07–1.58) than controls, but this risk did not remain significant after correcting for multiple testing (corrected p-value=0.075). Compared to controls, long-lived siblings presented a lower risk of dying from dementia (OR = 0.76 95% CI = 0.62–0.97) and ill-defined conditions (OR = 0.83 95% CI = 0.70–0.98). However, after correction for multiple tests, none of these associations remained significant (corrected p-values=0.075).

In both long-lived siblings and sporadic long-livers, nursing home was the major place of death (39.3% in siblings, 38,4% in controls), followed by hospital or hospice (25.1% in siblings, 25.6% in controls, where deaths in hospice represents less than 1 percentage point in siblings and controls), followed by home (25.1% in siblings, 25.6% in controls, where deaths in hospice represents less than 1 point in siblings and controls) and home (17.0% in siblings, 16.0% in controls). There was no significant difference regarding place of death between siblings and controls.

DISCUSSION

This study aimed to investigate EoL events, causes and location of death in long-lived siblings compared to sporadic long-livers. Mainly, we showed that long-lived siblings end their life in a similar way to sporadic long-livers measured by medication, hospitalizations, cause and location of death. The only exception was a lower occurrence of dementia in long-lived siblings than in sporadic long-livers during the last 5 years prior to death. This result aligns with a previous study reporting a lower prevalence of Alzheimer’s disease and related disorders in long-lived siblings compared to controls [7]. Some hypotheses may explain this result, and the association between depression and dementia is one of them. Indeed, some studies suggested that depressed individuals have a higher risk of dementia [29, 30], and previous studies reported a lower depression in long-lived siblings compared to controls [5, 7]. Thus, better mental health in long-lived siblings may be related to lower risk of dementia in late life, which is strengthened by our findings showing a lower risk of dying from mental diseases in long-lived siblings (non-significant after controlling for multiple testing). Another hypothesis may be that long-lived siblings experience better aging, by avoiding common age-related diseases, even in the last years of life. Studies reported better cognitive and physical functioning [31] as well as delayed aging-related chronic conditions in long-lived siblings than in sporadic long-livers [5, 7], supporting this hypothesis. In addition, in the present study, we reported a lower risk of ill-defined conditions in long-lived siblings (non-significant after controlling for multiple testing). Thus, the lower incidence of dementia in the last years of life suggests a compression of mental morbidity [32]. Another hypothesis lies on the relationship between marital status and dementia. Several studies have been reported an association between marital status and the risk of dementia; unmarried individuals having elevated risk of dementia compared to married individuals [33, 34]. In the same way, a recent paper highlighted that having more severe neuropsychiatric behavioral symptoms are associated with a higher likelihood of divorce or separation [35]. In the present work, marital status was significantly different between siblings and controls, with a higher proportion of married individuals in siblings and divorced individuals in controls. However, adjusting for marital status did not change the result (OR = 0.80, 95% CI = 0.66–0.97).

It is well known that dementia is highly underdiagnosed, with a meta-analysis estimating that nearly 62% of dementia cases are undetected [36]. However, this bias is likely not to impact long-lived siblings more than controls. Actually, the underreporting may be larger in the control group: We previously reported a lower cumulative incidence of depression in female long-lived siblings [5] and a recent study reported a higher proportion of depression in individuals with undetected dementia than in detected dementia cases [37]. These results suggested that dementia may more often be undiagnosed in patients presenting depression as depression symptoms may mask dementia symptomatology [37, 38]. However, other studies have not been able to confirm this bias [39, 40].

Regarding the other EoL outcomes, we showed that long-lived siblings experienced EoL events similar to sporadic long-livers regarding medication and hospitalization patterns during the last years of life.

In our study, the average number of prescribed drugs increased in both long-lived siblings and controls with proximity to death and was near 9 prescribed drugs at death. These results are in line with a previous Danish study reporting an increased number of prescribed drugs with proximity to death and between 6 and 10 prescribed drugs at death, depending on the year and the birth cohort [41]. At death, the number of drugs was slightly higher in long-lived siblings than in controls, but the association did not remain significant after correcting for multiple testing.

In our study population, the number of hospital stays, days at hospital and ER visits increased with proximity to death and mainly during the year prior to death in both long-lived siblings and sporadic long-livers. These results are concordant with previous studies reporting an increase of planned and unplanned hospitalizations during the last year of life and mainly during the last months of life [10 , 43]. While we reported a higher odds of being hospitalized during the last year of life in long-lived siblings compared to controls before adjusting for CCI and controlling for multiple testing, hospitalization patterns were similar between siblings and controls regarding number of hospital stays, days at hospital, and ER visits.

Our results also suggest that long-lived siblings, in many aspects, end their lives similar to sporadic long-livers. Cardiovascular diseases, which are the leading cause of death worldwide in older adults aged 70+, were also the main cause of death in our study population [44]. We reported a lower risk of dying from dementia and ill-defined conditions in long-lived siblings compared to controls (non-significant after correcting for multiple testing). Regarding dementia, these results are in line with previous findings highlighting a lower odd of dementia in long-lived siblings compared to controls [7], as well as our present results reporting lower dementia diagnoses during the last years of life. In the same way, the lower risk of dying from ill-defined conditions strengthen the hypothesis of better aging in long-lived siblings, even during the last years of life. We also reported a higher risk to die from cancer (non-significant after correcting for multiple testing), which is concordant with our previous study on long-lived female siblings where we found a higher but non-significant risk of cancer at extreme ages compared to controls [5]. While cancer incidence increases up to age 80–89, it then declines [45 –47]. Thus, dying from cancer at extreme ages may indicate delayed age-related diseases with a later onset of cancer. However, cancer may also be less diagnosed in the oldest, mainly among those frail or presenting with dementia [45, 48].

In our study, more than two thirds of both siblings and controls died in nursing home, hospital or hospice. Although home is usually the most preferred place of death among older adults, the reality is often different [8, 49]. A US study highlighted that older adults aged 85 + have a higher odds of dying in nursing home or in hospice facility than their younger counterparts [50]. At any rate, there was no difference regarding place of death between siblings and controls in our population, suggesting that they ended life in similar settings.

The main strength of this work lies in the study design consisting of matching long-lived siblings on sex, age at death and birth year with controls from the Danish population. Thus, the registry-based study design permitted avoiding selection bias, e.g., different ages at death leading to biased results due to different birth cohorts. Long-lived siblings and controls were all identified in high-quality data from National Danish registers, also leading to a large sample size.

This study also presented some limitations. First, we cannot exclude that some controls were from non-identified long-lived families, but if so, they would be rare [15]. Second, interviewed siblings may present better cognitive function than non-interviewed siblings and controls. Considering that a large part of long-lived siblings included in this study were interviewed, the association between membership and dementia in long-lived families may be overestimated. However, a sensitivity analysis on non-interviewed long-lived siblings compared to controls also showed a lower risk of being diagnosed with dementia in the 5 last years of life (Supplementary Material). Third, socioeconomic status (including education, income, etc.) was not available, and we cannot rule out that the long-lived siblings had higher socioeconomic status than sporadic long-lived controls. However, in a previous study comparing the siblings to control samples using the Danish 1916 Census, we did not find significant differences in income, wealth, or taxes between the long-lived families and the controls, indicating that the long-lived siblings were not generally socioeconomically privileged in their childhood [51]. Then, hospitalization at EoL may relate to two different situations: individuals may be hospitalized because of severe disease, or, they may be hospitalized because they are robust enough to get treated and have a chance of recovery. Thus, hospitalization differences, particularly at old ages, are difficult to interpret. Finally, quality of ending life may also be studied regarding cognitive or functional limitations, including falls, pain, and sensory organ deficits during the last years of life [12 , 52]. Social network and the presence of caregivers are also very relevant for EoL. However, these data are not available in the National Danish registers.

To conclude, we showed that long-lived siblings end their lives in a similar way to sporadic long-livers measured by medication, hospitalizations, cause and location of death. The only exception was a lower occurrence of dementia in long-lived siblings than in sporadic long-livers during the last 5 years prior to death. These results suggest that long-lived families are excellent candidates to identify environmental and genetic protective factors of dementia.

AUTHOR CONTRIBUTIONS

Angéline Galvin (Conceptualization; Data curation; Formal analysis; Methodology; Project administration; Validation; Writing – original draft; Writing – review & editing); Jacob Krabbe Pedersen (Conceptualization; Methodology; Validation; Writing – review & editing); Konstantin G. Arbeev (Conceptualization; Validation; Writing – review & editing); Mary F. Feitosa (Conceptualization; Validation; Writing – review & editing); Svetlana Ukraintseva (Conceptualization; Validation; Writing – review & editing); Shanshan Yao (Conceptualization; Validation; Writing – review & editing); Anne B Newman (Conceptualization; Validation; Writing – review & editing); Kaare Christensen (Conceptualization; Formal analysis; Funding acquisition; Methodology; Project administration; Validation; Writing – original draft; Writing – review & editing).

Footnotes

ACKNOWLEDGMENTS

The authors have no acknowledgments to report.

FUNDING

Research reported in this publication was supported by the National Institute on Aging of the National Institutes of Health (NIA/NIH) under Award Number U19AG063893. The work of KA, SU was partly supported by the NIA/NIH grant R01AG062623.

CONFLICT OF INTEREST

The authors have no conflict of interest to report.

DATA AVAILABILITY

Data sharing is not applicable to this article due to GDPR rules and the restricted access of national registers.

The supplementary material is available in the electronic version of this article: .

References

Aburto

, Villavicencio

, Basellini

, Kjærgaard

, Vaupel

(2020) Dynamics of life expectancy and life span equality. Proc Natl Acad Sci U S A 117, 5250–5259.

Van Den Berg

(2020) Family matters in unraveling human longevity. Aging 28, 22354–22355.

Perls

, Shea-Drinkwater

, Bowen-Flynn

, Ridge

, Kang

, Joyce

, Daly

, Brewster

, Kunkel

, Puca

(2000) Exceptional familial clustering for extreme longevity in humans. J Am Geriatr Soc 48, 1483–1485.

Ukraintseva

, Yashin

, Arbeev

(2016) Resilience versus robustness in aging. J Gerontol A Biol Sci Med Sci 71, 1533–1534.

Galvin

, Ukraintseva

, Arbeev

, Feitosa

, Christensen

(2020) Physical robustness and resilience among long-lived female siblings: A comparison with sporadic long-livers. Aging 12, 15157–15168.

Westendorp

RGJ

, Van Heemst

, Rozing

, Frölich

, Mooijaart

, Blauw

, Beekman

, Heijmans

, de Craen

AJM

, Slagboom

; Leiden Longevity Study Group (2009) Nonagenarian siblings and their offspring display lower risk of mortality and morbidity than sporadic nonagenarians: The Leiden Longevity Study. J Am Geriatr Soc 57, 1634–1637.

Ash

, Kroll-Desrosiers

, Hoaglin

, Christensen

, Fang

, Perls

(2015) Are members of long-lived families healthier than their equally long-lived peers? Evidence from the Long Life Family Study. J Gerontol A Biol Sci Med Sci 70, 971–976.

Higginson

, Daveson

, Morrison

, Yi

, Meier

, Smith

, Ryan

, McQuillan

, Johnston

, Normand

; BuildCARE (2017) Social and clinical determinants of preferences and their achievement at the end of life: Prospective cohort study of older adults receiving palliative care in three countries. BMC Geriatr 17,, 271.

Zhang

, Nilsson

, Prigerson

(2012) Factors important to patients’ quality of life at the end of life. Arch Intern Med 172, 1133–1142.

10.

Van den Block

, Deschepper

, Drieskens

, Bauwens

, Bilsen

, Bossuyt

, Deliens

(2007) Hospitalisations at the end of life: Using a sentinel surveillance network to study hospital use and associated patient, disease and healthcare factors. BMC Health Serv Res 7, 69.

11.

Martinsson

, Lundström

, Sundelöf

(2018) Quality of end-of-life care in patients with dementia compared to patients with cancer: A population-based register study. PLoS One 13, e0201051.

12.

GBD 2019 Dementia Forecasting Collaborators (2022) Estimation of the global prevalence of dementia in 2019 and forecasted prevalence in 2050: An analysis for the Global Burden of Disease Study 2019. Lancet Public Health 7, e105–125.

13.

Browne

, Kupeli

, Moore

, Sampson

, Davies

(2021) Defining end of life in dementia: A systematic review. Palliat Med 35, 1733–1746.

14.

Skytthe

, Valensin

, Jeune

, Cevenini

, Balard

, Beekman

, Bezrukov

, Blanche

, Bolund

, Broczek

, Carru

, Christensen

, Christiansen

, Collerton

, Cotichini

, de Craen

AJM

, Dato

, Davies

, De Benedictis

, Deiana

, Flachsbart

, Gampe

, Gilbault

, Gonos

, Haimes

, Hervonen

, Hurme

, Janiszewska

, Jylhä

, Kirkwood

TBL

, Kristensen

, Laiho

, Leon

, Marchisio

, Masciulli

, Nebel

, Passarino

, Pelicci

, Peltonen

, Perola

, Poulain

, Rea

, Remacle

, Robine

, Schreiber

, Scurti

, Sevini

, Sikora

, Skouteri

, Slagboom

, Spazzafumo

, Stazi

, Toccaceli

, Toussaint

, Törnwall

, Vaupel

, Voutetakis

, Franceschi

; GEHA consortium (2011) Design, recruitment, logistics, and data management of the GEHA (Genetics of Healthy Ageing) project. Exp Gerontol 46, 934–945.

15.

Wojczynski

, Jiuan Lin

, Sebastiani

, Perls

, Lee

, Kulminski

, Newman

, Zmuda

, Christensen

, Province

(2022) NIA Long Life Family Study: Objectives, design, and heritability of cross-sectional and longitudinal phenotypes. J Gerontol A Biol Sci Med Sci 77, 717–727.

16.

Sebastiani

, Hadley

, Province

, Christensen

, Rossi

, Perls

, Ash

(2009) A family longevity selection score: Ranking sibships by their longevity, size, and availability for study. Am J Epidemiol 170, 1555–1562.

17.

Schmidt

, Pedersen

, Sørensen

(2014) The Danish Civil Registration System as a tool in epidemiology. Eur J Epidemiol 29, 541–549.

18.

Pedersen

(2011) The Danish Civil Registration System. Scand J Public Health 39(7_suppl), 22–25.

19.

Schmidt

, Schmidt

SAJ

, Sandegaard

, Ehrenstein

, Pedersen

, Sørensen

(2015) The Danish National Patient Registry: A review of content, data quality, and research potential. Clin Epidemiol 7, 449–490.

20.

Lynge

, Sandegaard

, Rebolj

(2011) The Danish National Patient Register. Scand J Public Health 39(7 Suppl), 30–33.

21.

Pottegård

, Schmidt

SAJ

, Wallach-Kildemoes

, Sørensen

, Hallas

, Schmidt

(2017) Data Resource Profile: The Danish National Prescription Registry. Int J Epidemiol 46, 798–798f.

22.

Wallach Kildemoes

, Toft Sørensen

, Hallas

(2011) The Danish National Prescription Registry. Scand J Public Health 39(7 Suppl), 38–41.

23.

Helweg-Larsen

(2011) The Danish Register of Causes of Death. Scand J Public Health 39(7 Suppl), 26–29.

24.

Osler

, Christensen

, Garde

, Mortensen

, Christensen

(2017) Cognitive ability in young adulthood and risk of dementia in a cohort of Danish men, brothers, and twins. Dementia 13, 1355–1363.

25.

Kaalby

, Skytthe

, Andersen-Ranberg

, Jeune

(2021) Causes of death among 9000 Danish centenarians and semisuper-centenarians in the 1970-2012. In Exceptional Lifespans, Maier

, Vaupel

, Jeune

, eds. Springer, pp. 85–102.

26.

Charlson

, Pompei

, Ales

, MacKenzie

(1987) A new method of classifying prognostic comorbidity in longitudinal studies: Development and validation. J Chronic Dis 40, 373–383.

27.

Benjamini

, Hochberg

(2000) On the adaptive control of the false discovery rate in multiple testing with independent statistics. J Educ Behav Stat 25, 60–83.

28.

Benjamini

, Hochberg

(1995) Controlling the false discovery rate: A practical and powerful approach to multiple testing. J Royal Stat Soc Ser B 57, 289–300.

29.

Almeida

, Hankey

, Yeap

, Golledge

, Flicker

(2017) Depression as a modifiable factor to decrease the risk of dementia. Transl Psychiatry 7, e1117.

30.

Diniz

, Butters

, Albert

, Dew

, Reynolds

(2013) Late-life depression and risk of vascular dementia and Alzheimer’s disease: Systematic review and meta-analysis of community-based cohort studies. Br J Psychiatry 202, 329–335.

31.

Newman

, Glynn

, Taylor

, Sebastiani

, Perls

, Mayeux

, Christensen

, Zmuda

, Barral

, Lee

, Simonsick

, Walston

, Yashin

, Hadley

(2011) Health and function of participants in the Long Life Family Study: A comparison with other cohorts. Aging 3, 63–76.

32.

Cheng

(2014) Double compression: A vision for compressing morbidity and caregiving in dementia. Gerontologist 54, 901–908.

33.

Liu

, Zhang

, Choi S

won

, Langa

(2020) Marital status and dementia: Evidence from the Health and Retirement Study. J Gerontol B Psychol Sci Soc Sci 75, 1783–1795.

34.

Sommerlad

, Ruegger

, Singh-Manoux

, Lewis

, Livingston

(2018) Marriage and risk of dementia: Systematic review and meta-analysis of observational studies. J Neurol Neurosurg Psychiatry 89, 231–238.

35.

Monin

, McAvay

, Zang

, Vander Wyk

, Carrión

, Allore

(2023) Associations between dementia staging, neuropsychiatric behavioral symptoms, and divorce or separation in late life: A case control study. Sarangi A, editor. PLoS One 18, e0289311.

36.

Lang

, Clifford

, Wei

, Zhang

, Leung

, Augustine

, Danat

, Zhou

, Copeland

, Anstey

, Chen

(2017) Prevalence and determinants of undetected dementia in the community: A systematic literature review and a meta-analysis. BMJ Open 7,, e011146.

37.

Gamble

, Matthews

, Jones

, Hillman

, Woods

, Macleod

, Martyr

, Collins

, Pentecost

, Rusted

, Clare

(2022) Characteristics of people living with undiagnosed dementia: Findings from the CFAS Wales study. BMC Geriatr 22, 409.

38.

Tetsuka

(2021) Depression and dementia in older adults: A neuropsychological review. Aging Dis 12, 1920.

39.

Amjad

, Roth

, Sheehan

, Lyketsos

, Wolff

, Samus

(2018) Underdiagnosis of dementia: An observational study of patterns in diagnosis and awareness in US older adults. J Gen Intern Med 33, 1131–1138.

40.

Savva

, Arthur

(2015) Who has undiagnosed dementia? A cross-sectional analysis of participants of the Aging, Demographics and Memory Study. Age Ageing 44, 642–647.

41.

Pedersen

, Jensen

, Waldorff

, Søndergaard

, Christensen

(2020) Use of prescription medication in the last years of life: A population-based comparison of two oldest old Danish birth cohorts born 10 years apart. Age Ageing 49, 1105–1109.

42.

Barbiellini Amidei

, Macciò

, Cantarutti

, Gessoni

, Bardin

, Zanier

, Canova

, Simonato

(2021) Hospitalizations and emergency department visits trends among elderly individuals in proximity to death: A retrospective population-based study. Sci Rep 11, 21472.

43.

Luta

, Diernberger

, Bowden

, Droney

, Howdon

, Schmidlin

, Rodwin

, Hall

, Marti

(2024) Healthcare trajectories and costs in the last year of life: A retrospective primary care and hospital analysis. BMJ Support Palliat Care 14, e807–e815.

44.

GBD 2019 Ageing Collaborators (2019) Global, regional, and national burden of diseases and injuries for adults 70 years and older: Systematic analysis for the Global Burden of Disease 2019 Study. BMJ 376, e068208.

45.

Pedersen

, Engholm

, Skytthe

, Christensen

, on behalf of the Academy of Geriatric Cancer Research (AgeCare) (2016) Cancer and aging: Epidemiology and methodological challenges. Acta Oncol 55 (Supp 1), 7–12.

46.

DeSantis

, Miller

, Dale

, Mohile

, Cohen

, Leach

, Goding Sauer

, Jemal

, Siegel

(2019) Cancer statistics for adults aged 85 years and older, 2019. CA Cancer J Clin 69, 452–467.

47.

Tanskanen

, Seppä

KJM

, Virtanen

, Malila

, Pitkäniemi

(2021) Cancer incidence and mortality in the oldest old: A nationwide study in Finland. Am J Epidemiol 190, 836–842.

48.

Jones

, Di Martino

, Bradley

, Essang

, Hemphill

, Wright

, Renzi

, Surr

, Clegg

, De Wit

, Neal

(2022) Factors affecting the decision to investigate older adults with potential cancer symptoms: A systematic review. Br J Gen Pract 72, e1–10.

49.

van Doorne

, van Rijn

, Dofferhoff

, Willems

, Buurman

(2021) Patients’ preferred place of death: Patients are willing to consider their preferences, but someone has to ask them. Age Ageing 50, 2004–2011.

50.

Cross

, Warraich

(2019) Changes in the place of death in the United States. N Engl J Med 381, 2369–2370.

51.

Christensen

, Wojczynski

, Pedersen

, Larsen

, Kløjgaard

, Skytthe

, McGue

, Vaupel

, Province

(2020) Mechanisms underlying familial aggregation of exceptional health and survival: A three-generation cohort study. Aging Cell 19, e13228.

52.

Phua

, Visaria

, Østbye

, Malhotra

(2022) Association of vision and hearing impairments with quality of life among older adults: Mediation by psychosocial factors. Geriatr Gerontol Int 22, 56–62.