Seven-Year Survival Rate After Age 85 Years: Title and subTitle BreakRelation to Alzheimer Disease and Vascular Dementia FREE

SURVIVAL BEYOND the age of 80 years has increased rapidly in developed countries. The mortality rate at younger ages leaves a select group of survivors at advanced ages. Therefore, the pattern and predictors of survival in this age group may differ from those in younger ages.¹ Dementia is a major health care problem in the elderly and by age 85 years, the prevalence of dementia is 30%.² Both hospital³ andcommunity-based studies^{4 - 6} report a reduced survival rate in individuals with dementia. Although excess mortality in individuals with dementia is reduced with advanced age,^{7 - 9} the influence of dementia on survival in advanced ages may be substantial because of its high prevalence. However, to our knowledge, no population survey has studied how different types of dementia influence survival in relation to other mental and physical disorders in extreme old age.^{9 - 10}

The aim of this study was to use the material from the Longitudinal Gerontological and Geriatric Population Studies in Gothenburg, Sweden,^{2 ,11 - 14} to examine how Alzheimer disease (AD) and vascular dementia influence a 7-year survival rate at the age of 85 years. The sample has repeatedly been shown to be representative for the total population. The comprehensive nature of the study also allowed calculations of the influence on 7-year survival of mental illnesses other than dementia and of the most common physical disorders.

In 1986 and 1987, all individuals aged 85 years born between July 1, 1901, and June 30, 1902, registered for census purposes in Gothenburg, were invited to take part in a health survey. Both people living in the community and those in institutions were included. A systematic subsample was examined with a neuropsychiatric examination (of 494, 143 were men and 351 were women). This sample was described in detail previously² and found to be representative for the total population with regard to sex, marital status, psychiatric registration, 3-year mortality rate, and institutionalization. The mean age at the neuropsychiatric examination was 85 years and 5 months (range, 85 years and 3 months to 86 years and 1 month).

The study was approved by the Ethics Committee for Medical Research at Göteborg University. Informed consent was obtained from the subjects, their nearest relatives, or both.

The detailed examinations of manifestations of aging and somatic and psychiatric disorders included a physical examination by a geriatrician, neuropsychological examination by a psychologist, and laboratory tests, including electrocardiography, chest radiography, computed tomography (CT) of the brain, and an extensive biochemical evaluation including vitamin B₁₂, thyroid function tests, and cerebrospinal fluid analyses.^{11 - 12}

The neuropsychiatric examination was semistructured and performed by a trained psychiatrist in the subject's home or at institutions, and included ratings of symptoms and signs common in dementia and tests of mental functioning.² Psychiatric symptoms and signs were rated with the Comprehensive Psychopathological Rating Scale.¹⁵

A semistructured telephone interview with a close informant was performed by the psychiatrist for 451 subjects (91%) and included questions about changes in behavior and intellectual function and, in the case of dementia, questions about age at onset and course.

Information on date of death was available from the census register in Gothenburg, which accounts for all deaths in the region. Death certificates were available for all subjects who died during the first 3 years (n=161).

The diagnosis of dementia and its severity was based on the neuropsychiatric examination and the close informant interview² using the criteria listed in the Diagnostic and Statistical Manual of Mental Disorders, Revised Third Edition.¹⁶

Mild dementia was accompanied by significant impairments in work and social activities but the capacity for independent living remained, with adequate personal hygiene and relatively intact judgment. In moderate dementia, the individual needed some degree of daily supervision and in severe dementia he/she needed continuous supervision. The classification procedure was based on a detailed and structured assessment of social functioning (eg, the subjects' ability to use a telephone and public transportation, to manage their finances and daily hygiene, dress themselves, and prepare meals and do their own shopping).

Subjects with dementia were classified into etiologic subgroups: AD according to the criteria of the National Institute of Neurological and Communicative Disorders and Stroke–Alzheimer's Disease and Related Disorders Association,¹⁷ vascular dementia, and dementia attributable to other causes as proposed by Erkinjuntti et al.¹⁸ The etiologic diagnosis was based on all available information as described previously.^{2 ,19}

As reported previously,² 17 men and 47 women had AD, 13 men and 56 women had vascular dementia, 9 men and 5 women had other dementias, and 104 men and 243 women were without dementia. Of those with dementia, 11 men and 30 women had mild, 13 men and 38 women had moderate, and 15 men and 40 women had severe dementia. Mental syndromes were diagnosed according to the symptom criteria of the Diagnostic and Statistical Manual of Mental Disorders, Revised Third Edition¹⁶ based on information from the psychiatric examination and symptoms noted during the last month.^{20 - 21} Depressive syndromes (29 men and 78 women) included major depressive syndrome, dysthymia, and depression not otherwise specified. Psychotic syndromes (13 men and 31 women) included schizophrenic or schizophreniform syndrome, delusional syndrome, and psychosis not otherwise specified. Anxiety syndromes (18 men and 95 women) included phobia, obsessive-compulsive syndrome, and generalized anxiety syndrome. The diagnostic procedures have previously been described in more detail.^{20 - 21}

Physical disorders are listed in Table 1, and were diagnosed using information from the physical and laboratory examinations. Cerebrovascular disorders (stroke, transient ischemic attacks, and brain infarcts) were diagnosed using information from the physical examinations, CT scans, the neuropsychiatric examination, and the key informant interview. Information on cancer was obtained from the Swedish Cancer Registry.

Survival was determined from the time of examination to death. The cause of death was determined by information from death certificates and classified into 6 main categories: cardiovascular disorders (pulmonary embolism, congestive heart failure, myocardial infarction, heart or aortic rupture, and asystole); cerebrovascular disorders; cancer; infections (pneumonia, meningitis, renal infections, septicema, and cholecystitis); dementia; and other causes (gastric bleeding, diabetes mellitus, and trauma).

The interobserver reliability for symptoms and signs and regarding causes of dementia has been reported previously and was found to be satisfactory.²

Differences in proportions were tested for significance using the Fisher exact test with a 2-tailed level of significance.²² Differences in the 7-year survival rate between groups were analyzed with a nonparametric log rank test.²³ Predictors of death were analyzed with the Cox proportional hazards model.²³

Population attributable risk (PAR), which takes into consideration both the relative risk (RR) for death in individuals with the disorder and the prevalence of the disorder in the population, was calculated according to the formula:

PAR=p(r−1)∕[pr+(1−p)]

where p is the proportion of the population with the risk factor and r is the RR.⁹

There were 344 deaths (114 men and 230 women) during the 7-year follow-up. The 7-year survival rate was higher in women (121 [34.5%] of 351) than in men (29 [20.3%] of 143) (P<.001). The 50% survival time was 5.4 years in women (95% confidence interval [CI], 4.6-6.0 years) and 3.7 years in men (95% CI, 3.0-4.4 years).

The analysis according to the Cox proportional hazards model²³ (including dementia and all the mental and physical disorders studied) is shown in Table 2. A 7-year mortality rate in men was predicted by the presence of chronic obstructive lung disease (RR, 2.5), AD (RR, 2.6), vascular dementia (RR, 2.9), cancer of the gastrointestinal tract (RR, 2.8), and skin cancer (RR, 2.9), and in women by vascular dementia (RR, 3.6), AD (RR, 2.9), myocardial infarction (RR, 2.1), congestive heart failure (RR, 1.5), cerebrovascular disorder (RR, 1.7), cancer of the gastrointestinal tract (RR, 2.2), and hypertension (RR, 1.4). The PAR is also shown in Table 2. It is possible to sum PARs for AD and vascular dementia since these variables are disjoint (the paradigm did not allow the same individual to have both diagnoses). Alzheimer disease and vascular dementia together predicted 30.7% of deaths in men and 49.7% of deaths in women.

Multiple physical disorders were common. No physical disorder was diagnosed in 36 men and 46 women; 1 disorder was diagnosed in 44 men and 101 women; 2 disorders in 34 men and 87 women; 3 disorders in 19 men and 65 women; 4 disorders in 8 men and 39 women; 5 disorders in 1 man and 9 women; 6 disorders in 1 man and 3 women; and 7 disorders in 1 woman. An analysis according to the Cox proportional hazards model²³ (including dementia disorders and the number of physical disorders in each individual) showed that the 7-year mortality rate in men was increased by AD (RR, 2.5; 95% CI, 1.5-4.3; P <.001), vascular dementia (RR, 2.4; 95% CI, 1.3-4.4; P =.006), and number of physical disorders (RR, 1.2 per disorder; 95% CI, 1.1-1.4; P =.003), and in women with dementia by AD (RR, 2.5; 95% CI, 1.7-3.5; P <.001), vascular dementia (RR, 3.9; 95% CI, 2.8-5.4; P <.001), and number of physical disorders (RR, 1.2 per disorder; 95% CI, 1.1-1.3; P<.001).

A 7-year survival rate in men was 5.1%(2/39) in those with dementia and 26.0% (27/104) in those without dementia (P<.001); in women, 11.1% (12/108) in those with dementia and 44.9% (109/243) in those without dementia (P<.001) (sex difference: with dementia, P=.36; without dementia, P = .001).

A 50% survival time in men was 1.8 years (95% CI, 1.5-3.3 years) in those with dementia and 4.4 years (95% CI, 3.5-5.8 years) in those without dementia, and in women, 2.8 years in those with dementia (95% CI, 2.5-3.5 years) and 6.5 years (95% CI, 6.0-6.9 years) in those without dementia.

Dementia at age 85 years was associated with a reduced survival rate during the first 3 years (P<.001), during years 3 to 5 (P=.001), and during years 5 to 7 (P=.006) in women. In men the survival rate was reduced during the first 3 years (P=.001) and there was a trend in the same direction during years 3 to 5 (P=.26) and years 5 to 7 (P=.14). However, the number of men during the last 2 periods was small.

The mean duration of dementia from onset was 9.6 years in men (95% CI, 7.6-11.6 years) and 9.7 years in women (95% CI, 8.7-10.8 years) (P=.26). Duration was similar in the different types of dementia.

Seven-year survival curves in relation to severity of dementia at age 85 years are shown in Figure 1. Men with moderate and severe dementia had a shorter survival rate than men without dementia, while there was no difference between no dementia and mild dementia. Women with mild dementia had a trend toward a shorter survival rate and those with moderate or severe dementia had a significantly shorter survival rate than women without dementia. The difference between women with mild dementia and those with moderate and severe dementia was also significant.

The 50% survival time in men was 4.3 years (95% CI, 2.4-6.8 years) in mild dementia, 2.8 years (95% CI, 1.5-3.5 years) in moderate dementia, and 1.4 years (95% CI, 0.7-1.8 years) in severe dementia, and in women, 5.0 years (95% CI, 4.5-6.3 years) in mild dementia, 2.8 years (95% CI, 1.8-3.8 years) in moderate dementia, and 2.4 years (95% CI, 2.1-2.8 years) in severe dementia.

The survival rate was reduced in both men (P=.003) and women (P<.001) with AD and in men (P=.007) and women (P<.001) with vascular dementia, compared with individuals without dementia. In women, the survival rate was shorter in those with vascular dementia than in those with AD (P=.007), while it was similar in men (P=.77). When the types of dementia were divided by dementia severity, the number of men in the different groups was too small for meaningful analyses. Seven-year survival curves in relation to type and severity of dementia at age 85 years are therefore only given for women (Figure 2), but the curves were similar in men. A survival rate in 85-year-old women was reduced in mild, moderate, and severe vascular dementia, and in moderate and severe AD. There were no differences in the survival rate between women without dementia and those with mild AD during the 7-year follow-up, although there was a tendency for those with mild AD to have a steeper decline in their survival curve during the end of the study.

The 50% survival rate in those with AD was 1.8 years in men (95% CI, 0.7-4.3 years) and 4.5 years in women (95% CI, 2.8-5.4 years). In those with vascular dementia the survival rate was 1.8 years (95% CI, 0.8-2.8 years) in men and 2.5 years (95% CI, 2.1-2.8 years) in women.

The analysis according to the Cox proportional hazards model²³ (including all mental and physical disorders studied) showed that the 7-year mortality rate in women with dementia was predicted by the presence of cerebrovascular disease (n= 39; RR, 1.8; 95% CI, 1.1-3.0; P =.02) and urinary tract infection (n=41; RR, 1.7; 95% CI, 1.0-2.7; P=.04). Among men with dementia, the number of individuals in each of the subgroups of physical and mental disorders was too small for meaningful analyses.

No single disorder met the 0.10 level for entry into the model of AD. In women with vascular dementia, a 7-year mortality rate was predicted by the presence of urinary tract infection (n=21; RR, 3.0; 95% CI, 1.4-6.6; P=.007) and depression (n=15; RR, 2.4; 95% CI, 1.1-5.1; P =.03).

In men without dementia, a 7-year mortality rate was predicted by the presence of chronic obstructive lung disease (n=17; RR, 2.3; 95% CI, 1.3-4.1; P=.005), cancer of the gastrointestinal tract (n=7; RR, 2.4; 95% CI, 1.1-5.3; P=.04), and urinary tract infection (n=24; RR, 1.7; 95% CI, 1.0-2.8; P =.05), and in women without dementia by myocardial infarction (n=28; RR, 2.5; 95% CI, 1.6-4.0; P <.001), congestive heart failure (n=60; RR, 1.7; 95% CI, 1.2-2.5; P =.004), cerebrovascular disease (n=30; RR, 1.9; 95% CI, 1.2-3.0; P=.006), cancer of the gastrointestinal tract (n=10; RR, 2.9; 95% CI, 1.4-6.0; P=.004), breast cancer (n = 14; RR, 2.2; 95% CI, 1.1-4.2; P =.004), hip fracture (n=33; RR, 1.7; 95% CI, 1.1-2.7; P=.02), and hypertension (n=64; RR, 1.4; 95% CI, 1.0-2.1; P =.05).

Table 3 shows the causes of death reported in death certificates. Cardiovascular disorders were the most frequent cause of death (106 [66]) in both individuals with dementia (AD, 17 [63%] and vascular dementia, 32 [70%]) and without dementia (51 [64%]). Cerebrovascular disorder was the cause of death in 5% of subjects with vascular dementias and in 3 (11%) of 27 subjects with AD. Dementia was reported in death certificates as the cause of death in 31 (19%) of 161 subjects and in 29 (36%) of 81 subjects with dementia.

We found that among 20 mental and physical disorders studied, AD and vascular dementia were the most important predictors of 7-year mortality in a representative population of individuals aged 85 years, predicting 31% of all deaths in men and 50% of all deaths in women. Furthermore, together with cancer of the gastrointestinal tract, they were the only conditions that predicted death in both men and women. The other most important disorders predicting mortality were chronic obstructive lung disease (predicting 19%) and cancer of the gastrointestinal tract (predicting 10%) in men, and cerebrovascular disorders (predicting 12%) and congestive heart failure (predicting 9%) in women. Our findings are consistent with the results of a recent Chinese study⁹ that reported, in a mainly younger sample of elderly individuals, a PAR for death related to AD and vascular dementia of 5% in the group aged 65 to 74 years and 21% in those older than 75 years. These data suggest that the PAR of death due to the dementias increases with advancing age, despite a decreased RR of death with age in these disorders.⁹ This is explained by the increasing prevalence of dementia with age.

Our study thus indicates that dementia, so common at these ages, is also a major killer in advanced ages, illustrated by the finding that almost half of women without dementia aged 85 years survived up to age 92 years compared with one tenth of those with dementia, and that dementia shortened the 50% survival time by 3.7 years in women and by 2.6 years in men.

Our results are in line with those of previous reports^{24 - 28} from mainly younger samples that life expectancy decreases with severity of dementia. However, the survival rate was similar in those with mild AD and in individuals without dementia during the 7-year follow-up, while those with mild vascular dementia had a reduced survival rate. There was a tendency for a decreased survival rate in those with mild AD during the end of the period. This suggests that when AD is diagnosed at a very early stage, survival will be similar to that in individuals without dementia during a rather long period. It has to be elucidated whether the survival curves will start to differ if the subjects are followed up longer.

Although both AD and vascular dementia are associated with a decreased survival rate,^{9 - 10 ,29 - 36} there is conflicting evidence as to whether the survival rate differs between these disorders. In consistency with most other studies on mainly younger samples,^{9 ,29 - 32} we found a poorer survival rate in those with vascular dementia than in those with AD when all types of severity were considered, in contrast to 1 study³³ that reported the opposite and some reports of no difference.^{10 ,34 - 36} The disparate results may be due to differences in severity of dementia or different definitions of vascular dementia between studies. With the criteria used in this study, the agreement between clinical and neuropathologic diagnosis of AD has been reported to be 80% to 90%.^{37 - 38} The agreement between the clinical and pathologic diagnosis of multi-infarct dementia, a more pure form of vascular dementia, has generally been lower, between 50% and 60%.^{37 - 38} However, it is often difficult to differentiate between vascular dementia and AD both on clinical grounds and at autopsy.³⁹ Although most patients with dementia show pathologic features of AD at autopsy, mixed cases are common, and it is often difficult to decide which lesions caused the dementia.^{2 ,40 - 41} If multi-infarct dementia and mixed dementias are grouped together, the agreement between clinical and neuropathologic diagnosis increases to 80% to 95% for vascular dementia,^{18 ,39} and with the criteria used in this study, in which mixed cases are included among the vascular dementias, the diagnostic accuracy was 90% for vascular dementia.¹⁸ However, clinicopathologic correlation studies have not been performed in samples from the general population. No diagnostic approach to date has satisfactorily addressed the imprecision in the discrimination between pure and mixed forms of vascular dementia,⁴² and different criteria may result in considerable differences in the proportion diagnosed as having AD or vascular dementia.² With our criteria, more individuals may be diagnosed as having vascular dementia than if other criteria had been used. On the other hand, if we had included individuals with cerebrovascular disease in the group with AD, we might have overestimated the contribution of AD on the mortality rate. Inclusion of individuals with cerebrovascular pathologic conditions in the group with AD may be one reason why some studies did not find a difference in the mortality rate between those with AD and vascular dementia. Despite our strict criteria for AD, we cannot rule out the possibility that silent cerebrovascular disease may contribute to the increased mortality rate in individuals with AD.⁴³

In line with previous reports,^{10 ,29 ,44} women had a higher survival rate than men both among individuals without dementia and across different severities of dementia. However, since the mortality rate was low in women without dementia, the RR for death in women with dementia was higher than in men with dementia. In our study, the disorders that predicted death differed somewhat between the sexes. Vascular disorders (with the exception of vascular dementia) were predictors of death only in women but not in men. It is possible that women in this age group are more vulnerable to vascular disorders than men are.

Others⁴⁵ have reported a reduced survival rate in those with dementia in the presence of "physical illnesses" and depression, but most studies¹⁰ have not taken comorbidity into consideration. We found that few physical disorders predicted death in the group with dementia, while several such disorders predicted death among those without dementia. It is possible that the factor that reduces the survival rate in individuals with dementia is so strong that it overshadows other possible predictors of mortality in the group with dementia. Furthermore, it is often suggested that the decreased survival rate in individuals with vascular dementia is related to the presence of vascular disorders, such as cerebrovascular disorder and cardiac diseases. We found a decreased survival rate in those with vascular dementia that was independent of the presence of these disorders, in line with the finding of Tatemichi et al⁴⁶ that dementia is the most important predictor of death in patients with stroke. This suggests a common pathway leading to death in individuals with different types of dementia, although it does not explain the mechanism by which survival is reduced in individuals with dementia.

We were most concerned with disorders that predicted death, and not with the immediate causes of death stated in death certificates, which are known to be unreliable sources. It is, however, noteworthy that cardiovascular disorders were the most common causes of death according to death certificates not only in individuals with vascular dementia but also in the population with AD and without dementia. Others^{47 - 49} have reported bronchopneumonia as the most common cause of death in those with AD. The discrepancy may be explained by the advanced age of our sample, because silent vascular disorders are probably also common in individuals with AD in these age groups.^{13 ,39 ,43}

Our sample included people living in the community as well as in institutions.² There were no differences regarding the survival rate between responders and nonresponders at age 85 years. We are, therefore, confident that the sample is representative of its population base. However, our results may be underestimates because we did not take into consideration later development of dementia. This may be an important factor since the incidence of dementia in this age group is high.^{19 ,50}

The main finding in this 7-year follow-up of a representative population sample of individuals aged 85 years was that vascular dementia and AD per se were major predictors of mortality, predicting almost one third of deaths in men and half of the deaths in women. The prevalence of dementia may therefore have a major impact on mortality rates in different countries, which has important public health implications. The findings are also important with regard to counseling and treatment in the very old since a 7-year survival rate in those with mild AD, the stage at which treatment with antidementia drugs (eg, tacrine hydrochloride or donepezil) should be initiated, did not differ from that in those without dementia. Despite the strong impact of dementia on survival, dementia was reported in death certificates as the cause of death in only a minority of cases, which has also been found by others.^{49 ,51 - 52}

Reprints: Ólafur Aevarsson, MD, PhD, Department of Psychiatry, University Hospital Reykjavík, Landspítalinn, PO Box 10, IS-121 Reykjavík, Iceland (e-mail: olaaevar@rsp.is).

Accepted for publication October 29, 1997.

This study was supported by grants K97-27P-11337-02B, and K97-21X-11267-03A from the Swedish Medical Research Council, Göteborg Medical Services and Social Services Administrations, Stiftelsen Söderström-Königska Sjukhemmet, Konung Gustaf V:s och Drottning Victorias Stiftelse, Stiftelsen för Gamla Tjänarinnor, Handlanden Hjalmar Svenssons Forskningsfond, Stiftelsen Professor Bror Gadelius' Minnesfond, Swedish Society of Medicine, Göteborg Medical Society, Alzheimerfonden, Alma och Anna Yhlen's Foundation, and Fredrik och Rosa Malmborgs Stiftelse, Science Foundation of the University Hospital (Landspítalinn) Reykjavík, Iceland.

This study is a project within the gerontological and geriatric population studies in Gothenburg, Sweden. Project leader from 1971 to 1987 was Dr Svanborg and from 1988 Bertil Steen, MD, PhD. We thank Liselott Ågren and Yvonne Sundin for technical assistance.