Survival after initial diagnosis of Alzheimer disease -...

Survival after initial diagnosis of Alzheimer disease

Background: Alzheimer disease is an increasingly common con-
dition in older people. Knowledge of life expectancy after the
diagnosis of Alzheimer disease and of associations of patient
characteristics with survival may help planning for future care.

Objective: To investigate the course of Alzheimer disease after
initial diagnosis and examine associations hypothesized to corre-
late with survival among community-dwelling patients with Alz-
heimer disease.

Design: Prospective observational study.

Setting: An Alzheimer disease patient registry from a base pop-
ulation of 23 000 persons age 60 years and older in the Group
Health Cooperative, Seattle, Washington.

Patients: 521 newly recognized persons with Alzheimer disease
enrolled from 1987 to 1996 in an Alzheimer disease patient reg-
istry.

Measurements: Baseline measurements included patient demo-
graphic features, Mini-Mental State Examination score, Blessed
Dementia Rating Scale score, duration since reported onset of
symptoms, associated symptoms, comorbid conditions, and
selected signs. Survival was the outcome of interest.

Results: The median survival from initial diagnosis was 4.2 years
for men and 5.7 years for women with Alzheimer disease. Men
had poorer survival across all age groups compared with females.
Survival was decreased in all age groups compared with the life
expectancy of the U.S. population. Predictors of mortality based
on proportional hazards models included a baseline Mini-Mental
State Examination score of 17 or less, baseline Blessed Dementia
Rating Scale score of 5.0 or greater, presence of frontal lobe
release signs, presence of extrapyramidal signs, gait disturbance,
history of falls, congestive heart failure, ischemic heart disease,
and diabetes at baseline.

Limitations: The base population, although typical of the sur-
rounding Seattle community, may not be representative of other,
more diverse populations.

Conclusions: In this sample of community-dwelling elderly per-
sons who received a diagnosis of Alzheimer disease, survival
duration was shorter than predicted on the basis of U.S. popula-
tion data, especially for persons with onset at relatively younger
ages. Features significantly associated with reduced survival at
diagnosis were increased severity of cognitive impairment,
decreased functional level, history of falls, physical examination
findings of frontal release signs, and abnormal gait. The variables
most strongly associated with survival were measures of disease
severity at the time of diagnosis. These results should be useful to
patients and families experiencing Alzheimer disease, other care-
givers, clinicians, and policymakers when planning for future care
needs.


Larson EB, Shadlen MF, Wang L, McCormick WC, Bowen JD, Teri L, Kukull WA. Survival after initial diagnosis of Alzheimer disease. Annals of Internal Medicine. 2004 Apr 6;140(7):501-9.


Alzheimer disease is one of the leading causes of death
in older people (1). One recent study suggests that
7.1% of all deaths in 1995 were attributable to Alzheimer
disease, placing it on a par with cerebrovascular disease as
the third leading cause of death (2). Estimates of predicted
survival of persons who have received a diagnosis of Alz-
heimer disease should be useful for patients, caregivers,
clinicians, and policy planners. Previous studies that relied
on epidemiologic surveys had little opportunity to analyze
clinical factors and also may not be generalizable to every-
day clinical settings. Results from other studies that were
based on convenience samples of persons from specialized
Alzheimer disease centers are probably subject to referral
bias. Our study design allowed us to estimate the magni-
tude of the reduction in age-adjusted life expectancy attrib-
utable to Alzheimer disease in a cohort of patients similar
to those encountered in the clinical setting where Alzhei-
mer disease is initially recognized.

It is not surprising that patients with Alzheimer disease
probably have reduced survival compared with older per-
sons without dementia (3–6). Certain characteristics, par-
ticularly male sex (7–11), initial dementia severity (5, 9,
12–17), presence of behavioral disturbances, wandering
and falling (18), comorbid conditions (19), and presence
of extrapyramidal signs (5, 20), are reported to be associ-
ated with decreased survival among patients with Alzhei-
mer disease.

In 1987, we began a prospective observational study to
investigate the natural history of persons with newly diag-
nosed Alzheimer disease in a community-dwelling popula-
tion (21). This report describes overall survival and exam-
ines the association between factors hypothesized to affect
survival in 521 patients with Alzheimer disease newly rec-
ognized between 1987 and 1996 and followed until time
of death or 2001.


___
Context
The prognosis in patients with Alzheimer disease may be
associated with specific patient characteristics observed
shortly after diagnosis.
Contribution
Severity of initial cognitive impairment and deterioration in
the Mini-Mental State Examination score during the first
year after diagnosis were strongly associated with de-
creased survival. Men had shorter survival than women
across all age groups. Frontal lobe release signs, gait dis-
turbances, falling, congestive heart failure, and diabetes
were all associated with decreased survival.
Implications
Early appraisal of patients with Alzheimer disease may help
set expectations and priorities for planning patient care.
–The Editors
___


METHODS


Cases

From 1987 to 1996, an Alzheimer’s disease patient
registry enrolled persons receiving care in the Seattle and
western King County clinics of the Group Health Coop-
erative (a well-established staff-model health maintenance
organization with an enrollment population base age 60
years and older of 23 000) in Washington. The research
protocol for this study was reviewed and approved by the
health maintenance organization and the University of
Washington’s institutional review boards.

The registry’s objective was to identify all persons en-
rolled in the base population with newly recognized symp-
toms of possible dementia (for example, memory loss, con-
fusion, or wandering). The goal of enrollment was to
assemble a series of incident cases from a defined popula-
tion, that is, possible patients with dementia who had pre-
sented for medical care within the health maintenance or-
ganization 1 year or less before enrollment in the registry.
Study methods have been reported previously (4, 21). The
following strategy was used to detect possible incident
cases. Research associates systematically searched for pa-
tients with symptoms of memory loss, confusion, or wan-
dering; for tests or treatments potentially indicative of de-
mentia (by review of computed tomography and magnetic
resonance imaging logbooks); for discharge diagnoses from
the hospital information system; for logbook records of
visits to neurology, geriatrics, and mental health clinics; for
emergency department logs; and for computerized clinic
treatment record files. In addition, primary care physicians
were invited to refer suspected new cases for evaluation
through brochures and mailings describing the registry.
These primary care physicians also received a monthly
newsletter describing clinical research advances in aging
and dementia and reminding them of our study.

The process of patient identification through consen-
sus diagnosis took an average of 12 weeks. A research as-
sistant screened the medical charts of persons identified as
potential cases. If the person’s record contained any infor-
mation indicating memory loss or other cognitive changes
indicating that the person might have Alzheimer disease,
the primary care physician was contacted to obtain permis-
sion to contact that person.

Persons who were judged to possibly have newly rec-
ognized dementia were contacted to request that they par-
ticipate in the program. After we obtained informed con-
sent, the research nurse interviewed the person and began
collecting data, usually in the home. A psychometrist re-
search assistant administered standard neuropsychological
testing. Persons who agreed to participate in the registry
underwent a complete, standardized physical and neuro-
logic examination, administered by one of the study phy-
sicians, and the usual laboratory work-up (including neu-
roimaging) to rule out other causes of dementia and
complicating illnesses. Chronology of symptoms was deter-
mined by study clinicians from interviews of the patient
and informants and by review of medical records. The ex-
amining physician from the registry estimated the duration
of symptoms before evaluation to the nearest year after
resolving any discrepant information through further ques-
tioning and by relating time frames to the patient’s life
events. All eligible persons met criteria of the revised third
edition of the Diagnostic and Statistical Manual of Mental
Disorders (DSM-III-R) for dementia and criteria of the Na-
tional Institute of Neurological and Communicative Dis-
orders and Stroke–Alzheimer’s Disease and Related Disor-
ders Association (22) for probable or possible Alzheimer
disease at a consensus diagnosis conference that included at
least 2 physicians (including the examining physician), a
neuropsychologist, an epidemiologist, and study staff. On
the basis of chart review, persons whose symptoms led to a
diagnosis of dementia or Alzheimer disease more than 1
year before the time the registry detected them were ex-
cluded as not being incident cases. Of 1028 patients who
consented to participate and who potentially could have
been incident cases after chart review, 58 were excluded
from the study before the standardized evaluation because
of intervening death or acute illness.

A total of 970 participants identified with suspected
dementia (from 1987 to 1996) were evaluated and fol-
lowed annually. When discrepant findings or clinical
course indicated a possible change from the initial diagno-
sis, we reevaluated the person according to the standard-
ized protocol, which included an evaluation by a study
physician to verify or modify the initial diagnoses. Persons
initially classified with possible or probable Alzheimer dis-
ease but who were subsequently given a different diagnosis
were excluded from these analyses. Persons with more than
one cause of dementia (so-called mixed dementia) were ex-
cluded from these analyses. Of the individuals with sus-
pected dementia (n=970), 521 received a diagnosis of
probable or possible Alzheimer disease (431 probable, 90
possible), 174 received diagnoses of other causes of demen-
tia, and 237 did not meet DSM-III-R criteria for dementia
within 18 months from their initial intake. The remaining
38 met diagnostic criteria of Alzheimer disease after at least
18 months of follow-up. Of the 174 persons with other
causes of dementia, 74 were classified as having vascular
dementia, 12 were classified as having alcohol-related
causes, 54 were classified as having mixed and other causes
(such as Parkinson disease or progressive supranuclear
palsy), and 34 were classified as having unknown causes by
DSM-III-R criteria. Duration of survival was assessed an-
nually. The dates of death were available for all study par-
ticipants who died during the study period. Mean (SD)
follow-up time was 5.2±3.1 years (range, 0.2 to 14 years).

The ethnic composition of the persons in the registry
was 88% white, 8.7% black, and 3.3% Asian or Hispanic.
The ethnic composition from the King County 1990 cen-
sus was 84.8% white, 5.1% black, and 10.1% Asian or
Hispanic. The overall ethnic composition of adults in the
Group Health Cooperative was 90.2% white, 3.3% black,
and 6.5% other.

Study Variables

The research nurse interviewed the patient and care-
giver before scheduling an examination. The baseline in-
formation determined by questionnaire and physician in-
terview included age (categorized as ≤75, 76 to 80, 81 to
85, and >85 years for analyses), sex (male vs. female), eth-
nicity (white, black, other), baseline systolic blood pressure
(<110 mm Hg, 110 to 160 mm Hg, and ≥160 mm Hg),
and education (<12 years, 12 years, >12 years, or un-
known). As reported previously (4, 21), the severity of
symptoms of Alzheimer disease was determined on the
basis of the baseline Mini-Mental State Examination
(MMSE) as continuous and categorical (≤17, 18 to 21, 22
to 24, or ≥25 points) variables (23) and baseline Dementia
Rating Scale (DRS) as continuous and categorical (≤2, 2.5
to 3, 3.5 to 5, >5 points) variables (24).

Symptoms of dementia (for example, agitation, para-
noia, and wandering) were assessed with the Revised Mem-
ory and Behavior Problems Checklist (25, 26). Physical
signs of frontal release reflexes and other neurologic
changes associated with dementia (for example, glabellar
sign and grasp reflex) were categorized as present or not
present by the examining physician. Study physicians as-
sessed the presence or absence of rigidity on the basis of
passive range of motion of the right and left elbow and
wrist joints. The presence of a gait disorder was also deter-
mined through observation by the study physicians. All
persons completed a general medical history and a review
of systems questionnaire administered by a trained inter-
viewer (4, 21). Comorbid conditions included hyper-
tension, diabetes, ischemic heart disease, congestive heart
failure, and history of strokes. Onset of symptoms, as esti-
mated at the time of diagnosis, and the age at enrollment
into the study were used to calculate the reported duration
of disease symptoms (defined as <1, 1 to 3, or >3 years).

Statistical Analysis

Our general hypothesis was that the severity of Alzhei-
mer disease at the time of diagnosis, as measured by cog-
nitive and functional measures as well as symptoms of de-
mentia, would be associated with survival. The event time
was the time of death. Persons who were still alive at the
end of the study on 12 September 2001 were censored at
that time. Survival time was defined as the time from the
initial date of diagnosis of Alzheimer disease until the date
of death or study conclusion. Patients who did not receive
in-person follow-up were tracked for vital status through
Group Health Cooperative records, informant reports, or
local newspaper obituaries.

Variables investigated were 1) baseline MMSE score;
2) baseline DRS score; 3) psychiatric symptoms of para-
noia or hallucinations at the time of evaluation; 4) behav-
ioral disturbances, including agitation, irritability, or emo-
tional lability; 5) frontal release signs, including glabellar,
snout, or grasp; 6) extrapyramidal signs, including rigidity
or tremor; 7) wandering; 8) falls; 9) gait disorder; 10) uri-
nary incontinence; and 11) depression. Variables adjusted
as potential confounders included 1) age, 2) sex, 3) ethnic-
ity, 4) education, 5) ischemic heart disease, 6) congestive
heart failure, 7) diabetes, 8) hypertension, and 9) stroke.

To describe associations of baseline factors with mor-
tality, we calculated both proportions of patients who died
and the time until death by baseline variables. Survival
time quartiles were used to describe the time until death
when 25%, 50%, and 75% of patients died. Log-rank tests
were used for testing the equality of survival among groups.
We compared the estimated life expectancy, expressed in
quartiles, of patients with Alzheimer disease in the study
sample with the reported life expectancy of the U.S. pop-
ulation by sex and age strata (27). Cox proportional haz-
ards regression models were used to examine how baseline
characteristics affected risk for death during the study pe-
riod (28). The associations of predictors and mortality are
estimated, with adjustments by age, sex, and ethnicity. We
also adjusted for medical conditions that were found to be
associated with survival and repeated analyses on the pre-
dictors of interest. Schoenfeld residual tests (29) were used
to evaluate the proportional hazard assumption. If the as-
sumption failed, time-varying covariates were introduced.
To help illustrate the findings, Kaplan–Meier curves were
estimated. All data were analyzed by using Stata software,
version 7 (Stata Corp., College Station, Texas).

A secondary investigation of potential interest to clini-
cians was determining whether patients who had greater
cognitive decline during the first year after diagnosis had a
higher risk for dying after the first year. Greater decline
during the first year was defined as a decrease of 5 or more
points in MMSE score (that is, MMSE score at diagnosis
[baseline MMSE score] minus MMSE score at the first
year). This value was chosen a priori to reflect clinically
meaningful cognitive decline (30). The Cox proportional
hazards model was used to estimate the risk for dying by
adjusting for baseline MMSE scores, age at diagnosis, sex,
ethnicity, and comorbid conditions. The Schoenfeld resid-
ual test was used to evaluate the proportional hazards
assumptions. Sensitivity analyses (Appendix, available at
www.annals.org) were performed to evaluate potential bi-
ases that could be introduced because of missing data on
follow-up MMSE scores and floor effect of low baseline
MMSE scores. Logistic regression analyses were performed
on the patients who had follow-up MMSE scores to ex-
plore which baseline factors predict greater decline during
the first year (Appendix Table, available at www.annals
.org). On the basis of the risk factors for decline identified
by logistic regression, we divided those with missing data
on follow-up MMSEs into declined and nondeclined
groups. Then, we repeated the Cox model to see whether
any change in estimates resulted from adding those with-
out follow-up MMSEs into the analytic sample. The Ap-
pendix Figure (available at www.annals.org) is a Kaplan–
Meier survival graph of these results.

Role of the Funding Sources

The funding sources had no role in the design, con-
duct, or reporting of this study or in the decision to submit
the manuscript for publication.


RESULTS


Table 1 shows quartile estimates of survival for pa-
tients with Alzheimer disease according to baseline charac-
teristics. Median survival was longer for younger persons
and women but did not vary by education level. Patients
age 85 years and older, with gait disturbance, wandering,
and comorbid diabetes and congestive heart failure had the
poorest survival (median survival times were 3.2 years, 3.5
years, 4.1 years, 3.8 years, and 3.0 years, respectively; P <
0.01 for all comparisons [log-rank test]). Other predictors
of decreased survival were male sex, lower MMSE score,
higher (worse) DRS score, presence of frontal release signs,
presence of extrapyramidal signs, history of falls, presence
of urinary incontinence, history of ischemic heart disease,
and history of stroke. Duration of survival among patients
with Alzheimer disease did not differ by ethnicity, presence
of hypertension, presence of psychiatric symptoms, pres-
ence of behavioral disturbances, presence of depression
symptoms, or duration of dementia symptoms at the time
of diagnosis.

Table 2 compares the life expectancy of patients with
Alzheimer disease in this study with that of the U.S. pop-
ulation (27). Patients with Alzheimer disease at 70 years of
age had a significantly decreased survival compared with
the life expectancy of the U.S. population. Median survival
time for women with Alzheimer disease at 70 years of age
was 8.0 years compared with 15.7 years for the U.S. pop-
ulation. Survival time for the 50% quartile (median) for
men at 70 years of age was 4.4 years compared with 9.3
years for the U.S. population. Across all age groups, men
had poorer survival than women. At progressively older
ages, the absolute survival difference between patients with
Alzheimer disease and the U.S. population diminished. For
example, survival time for the 50% quartile (median) for
women at 85 years of age was 3.9 years compared with 5.9
years for the U.S. population. Survival time for the 50%
quartile (median) for men at 85 years of age was 3.3 years
compared with 4.7 years for the U.S. population. As ex-
pected, survival declined with age, but the decline for men
with Alzheimer disease with increasing age was less than for
women. Median life expectancy for men with Alzheimer
disease was 4.4 years at 70 years of age versus 3.3 years at
85 years of age, and median life expectancy for women was
8.0 years at 70 years of age versus 3.9 years at 85 years of
age.

Table 3 shows the hazard ratios of death according to
the severity of symptoms and typical symptoms of demen-
tia. Compared with the presence or absence of dementia
signs and symptoms at baseline, measures of the severity of
symptoms at the time of diagnosis were the strongest pre-
dictors of survival. The independent risk factors in propor-
tional hazards models adjusted by age, sex, and ethnicity
were low MMSE score, poor DRS score, presence of base-
line frontal lobe release signs, presence of baseline gait dis-
turbance, history of falls, congestive heart failure, history of
diabetes, and history of ischemic heart disease. After adjust-
ment for age, sex, ethnicity, ischemic heart disease, conges-
tive heart failure, and diabetes, the presence of extrapyra-
midal signs and symptoms of wandering became more
significant as independent risk factors for death. The haz-
ard ratio of death for older age increased over time by an
exponential factor. Comorbid conditions further reduced
survival, with hazard ratios for death of 1.2 for ischemic heart
disease, 1.3 for congestive heart failure, and 1.7 for diabetes.

A 5-point or greater decline in the MMSE score after
the first year of follow-up was also associated with a signif-
icantly increased risk for death. The estimated hazard ratio
for death in patients with a 5-point or greater decline in
MMSE score was between 1.60 and 1.66 in 4 models con-
trolling for age, baseline MMSE score, sex, ethnicity, isch-
emic heart disease, congestive heart failure, and diabetes. In
sensitivity analyses, we found that the floor effect of low
baseline MMSE scores led to slight underestimates of the
hazard ratio of death for the declined group, but missing
data on follow-up MMSE scores did not.

The Figure contains graphic illustrations of Kaplan–
Meier survival curves by MMSE score category, DRS cat-
egory, presence or absence of gait disorder, and presence or
absence of wandering at baseline.


DISCUSSION


In all age groups, patients with Alzheimer disease in
our study had decreased survival compared with survival in
the general U.S. population (27). Men had a median sur-
vival of 4.2 years from their initial diagnosis, and women
had a median survival of 5.7 years. Our findings differ
from estimates of median survival from the onset of symp-
toms of dementia in the Canadian Study of Aging (3.17
years for men and 3.36 years for women) (3). The shorter
survival estimates in the Canadian study could relate to 2
issues. First, the patients in that study were older (average
age, 83.8 years); second, because the study dealt with a
population-based prevalence sample, it included a relatively
greater proportion of persons with severe dementia and
nursing home residents who were at later stages in the
course of the natural history of this progressive disease (3).
Also, sex was not associated with Alzheimer disease survival
in the Canadian study (3). By contrast, we found the me-
dian life expectancy for a 70-year-old man with Alzheimer
disease to be 4.4 years compared with 8.0 years for a
woman of the same age with Alzheimer disease. This sex
gap narrowed at more advanced ages. Our findings are
consistent with previous reports that men with Alzheimer
disease have approximately half the life expectancy esti-
mated for women with Alzheimer disease (5, 7, 9–11, 31).
Our sample and study design are arguably more typical of
the persons clinicians see in everyday medical practice and
who may be recognized as having typical symptoms and
signs of dementia. Our results are also consistent with our
previously reported average survival from study entry (5.3
years) in a consecutive outpatient (convenience) sample of
patients with existing Alzheimer disease in King County,
Washington (18).

As reported previously, the association between mea-
sures of the severity of Alzheimer disease symptoms and
survival was strong (5, 9, 12–17). We found that the effect
of cognitive performance scores and dementia severity
scores on survival remained significant even after adjust-
ment for age, sex, ethnicity, and presence of vascular co-
morbid conditions. Comorbid conditions, such as a history
of diabetes, congestive heart failure, and ischemic heart
disease, reduced survival of patients with Alzheimer disease
in this cohort, with hazard ratios of 1.7, 1.3, and 1.2,
respectively. In addition, survival was reduced by the
symptoms and signs of the disease. Few studies of similar
design have taken into account the influence of possible
confounders, such as vascular disease comorbid conditions,
on estimates of survival in Alzheimer disease.

The data in Table 2 can be used as a reference to
estimate age- and sex-adjusted life expectancy for patients
with Alzheimer disease. The information in Table 3 can be
used to estimate additional reductions in survival on the
basis of the presence or absence of other signs and symp-
toms of dementia. A recent report also showed that the
presence of a gait disorder was associated with increased
risk for death in older persons (32). The presence of extra-
pyramidal signs can be a manifestation of Lewy body dis-
ease. Without neuropathologic data, we cannot rule out
the possibility that some of our patients might have had
concomitant Lewy body disease. Although symptoms of
agitation, irritability, emotional lability, paranoia, or hallu-
cinations affect the quality of life of patients with Alzhei-
mer disease, survival did not differ among patients with or
without these symptoms in our study. The relationship of
psychiatric or behavioral symptoms and Alzheimer disease
mortality has been difficult to study because of the variabil-
ity of presentation of these symptoms during the natural
history of the disease. Patients with early dementia can
present with behavioral symptoms, such as wandering, that
subside at later stages of the disease when mobility is di-
minished. Conversely, patients without behavioral symp-
toms at initial presentation can develop paranoid ideations
and delusions at a later course of their illness. By only
examining the influence of baseline psychiatric symptoms
on the outcome of interest, we may not have captured the
dynamic interactions of change in the frequency and inten-
sity of behavioral symptoms on survival.

We believe that our observations, if confirmed, would
be of great interest to patients with Alzheimer disease and
family members making plans for future care needs.

The strengths of this study are the ability to examine
demographic factors, signs and symptoms of dementia,
measures of the severity of symptoms, and the presence of
comorbid conditions in a community-based cohort. The
study had a large sample size and relatively long duration
of follow-up. We must caution, however, that this health
maintenance organization–based sample may not be rep-
resentative of patients with Alzheimer disease in the U.S.
population as a whole, although the demographic features
of the sample do resemble those of the surrounding Seattle
community. We think that the lack of association between
duration of dementia symptoms and survival reflects the
variability of self-reports and informant reports of duration
for an illness that is typically insidious in onset and often
difficult to distinguish from age-related decline. These
measurement errors would bias the results toward the null.
It would be useful to analyze the interrelationships of co-
morbid conditions and the actual cause of death for our
participants. However, we do not have reliable data on
causes of death. Because of the study design, we could not
evaluate the potential confounding effect of use of drugs
with potential psychiatric effects in this cohort.

The most powerful predictors of reduced survival were
a poor score on the MMSE and increased Alzheimer dis-
ease–related functional impairment, as measured by the DRS.
Baseline psychiatric symptoms or behavioral disturbances
were not strongly associated with survival. The conventional
wisdom that persons with Alzheimer disease who experi-
ence a rapid decline in cognitive performance in the first
year will have a more malignant course is confirmed by our
findings that an MMSE score decline of 5 points or more
at 1 year of follow-up was associated with a 60% increased
risk for death. Our findings suggest that a straightforward
clinical evaluation consisting of the patient’s history, a cog-
nitive screening test, functional assessment from an infor-
mant, and a careful neurologic examination can detect fea-
tures associated with prognosis for survival of patients with
Alzheimer disease typically seen in clinical practice.


From University of Washington and the Group Health Cooperative,
Seattle, Washington.

Grant Support: By National Institute on Aging grant AG 06781 and by
a Minority Supplement to National Institute of Aging grant AG 06781-
13S1 to the University of Washington Alzheimer’s Disease Patient Registry.

Potential Financial Conflicts of Interest: None disclosed.

Requests for Single Reprints: Eric B. Larson, MD, MPH, Center for
Health Studies, 1730 Minor Avenue, Suite 1600, Seattle, WA 98101-
1448; e-mail, larson.e@ghc.org.

Current author addresses and author contributions are available at www
.annals.org.


References


1. Ganguli M, Rodriguez EG. Reporting of dementia on death certificates: a
community study. J Am Geriatr Soc. 1999;47:842-9. [PMID: 10404929]
2. Ewbank DC. Deaths attributable to Alzheimer’s disease in the United States.
Am J Public Health. 1999;89:90-2. [PMID: 9987474]
3. Wolfson C, Wolfson DB, Asgharian M, M’Lan CE, Ostbye T, Rockwood
K, et al. A reevaluation of the duration of survival after the onset of dementia.
N Engl J Med. 2001;344:1111-6. [PMID: 11297701]
4. Bowen JD, Malter AD, Sheppard L, Kukull WA, McCormick WC, Teri L,
et al. Predictors of mortality in patients diagnosed with probable Alzheimer’s
disease. Neurology. 1996;47:433-9. [PMID: 8757016]
5. Molsa PK, Marttila RJ, Rinne UK. Long-term survival and predictors of
mortality in Alzheimer’s disease and multi-infarct dementia. Acta Neurol Scand.
1995;91:159-64. [PMID: 7793228]
6. Belloni-Sonzogni A, Tissot A, Tettamanti M, Frattura L, Spagnoli A. Mor-
tality of demented patients in a geriatric institution. Arch Gerontol Geriatr. 1989;
9:193-7. [PMID: 2589919]
7. Jagger C, Clarke M, Stone A. Predictors of survival with Alzheimer’s disease:
a community-based study. Psychol Med. 1995;25:171-7. [PMID: 7792352]
8. Ostbye T, Hill G, Steenhuis R. Mortality in elderly Canadians with and with-
out dementia: a 5-year follow-up. Neurology. 1999;53:521-6. [PMID: 10449114]
9. Heyman A, Peterson B, Fillenbaum G, Pieper C. The consortium to establish
a registry for Alzheimer’s disease (CERAD). Part XIV: Demographic and clinical
predictors of survival in patients with Alzheimer’s disease. Neurology. 1996;46:
656-60. [PMID: 8618662]
10. Lapane KL, Gambassi G, Landi F, Sgadari A, Mor V, Bernabei R. Gender
differences in predictors of mortality in nursing home residents with AD. Neu-
rology. 2001;56:650-4. [PMID: 11245718]
11. Gambassi G, Lapane KL, Landi F, Sgadari A, Mor V, Bernabie R. Gender
differences in the relation between comorbidity and mortality of patients with
Alzheimer’s disease. Systematic Assessment of Geriatric drug use via Epidemiol-
ogy (SAGE) Study Group. Neurology. 1999;53:508-16. [PMID: 10449112]
12. Molsa PK, Marttila RJ, Rinne UK. Survival and cause of death in Alzhei-
mer’s disease and multi-infarct dementia. Acta Neurol Scand. 1986;74:103-7.
[PMID: 3776457]
13. Hier DB, Warach JD, Gorelick PB, Thomas J. Predictors of survival in
clinically diagnosed Alzheimer’s disease and multi-infarct dementia. Arch Neurol.
1989;46:1213-6. [PMID: 2818256]
14. Claus JJ, van Gool WA, Teunisse S, Walstra GJ, Kwa VI, Hijdra A, et al.
Predicting survival in patients with early Alzheimer’s disease. Dement Geriatr
Cogn Disord. 1998;9:284-93. [PMID: 9701680]
15. Bianchetti A, Scuratti A, Zanetti O, Binetti G, Frisoni GB, Magni E, et al.
Predictors of mortality and institutionalization in Alzheimer disease patients 1
year after discharge from an Alzheimer dementia unit. Dementia. 1995;6:108-12.
[PMID: 7606278]
16. Aguero-Torres H, Fratiglioni L, Winblad B. Natural history of Alzheimer’s
disease and other dementias: review of the literature in the light of the findings
from the Kungsholmen Project. Int J Geriatr Psychiatry. 1998;13:766-66.
[PMID: 9850872]
17. Knopman DS, Kitto J, Deinard S, Heiring J. Longitudinal study of death
and institutionalization in patients with primary degenerative dementia. J Am
Geriatr Soc. 1988;36:108-12. [PMID: 3339214]
18. Walsh JS, Welch HG, Larson EB. Survival of outpatients with Alzheimer-
type dementia. Ann Intern Med. 1990;113:429-34. [PMID: 2386336]
19. Nielsen H, Lolk A, Pedersen I, Autzen M, Sennef C, Kragh-Sorensen P.
The accuracy of early diagnosis and predictors of death in Alzheimer’s disease and
vascular dementia—a follow-up study. Acta Psychiatr Scand. 1991;84:277-82.
[PMID: 1950629]
20. Stern Y, Albert M, Brandt J, Jacobs DM, Tang MX, Marder K, et al.
Utility of extrapyramidal signs and psychosis as predictors of cognitive and
functional decline, nursing home admission, and death in Alzheimer’s dis-
ease: prospective analyses from the Predictors Study. Neurology. 1994;44:
2300-7. [PMID: 7991116]
21. Larson EB, Kukull WA, Teri L, McCormick W, Pfanschmidt M, van Belle
G, et al. University of Washington Alzheimer’s Disease Patient Registry (ADPR):
1987-1988. Aging (Milano). 1990;2:404-8. [PMID: 2094380]
22. McKhann G, Drachman D, Folstein M, Katzman R, Price D, Stadlan EM.
Clinical diagnosis of Alzheimer’s disease: report of the NINCDS-ADRDA Work
Group under the auspices of Department of Health and Human Services Task
Force on Alzheimer’s Disease. Neurology. 1984;34:939-44. [PMID: 6610841]
23. Folstein MF, Folstein SE, McHugh PR. “Mini-mental state”. A practical
method for grading the cognitive state of patients for the clinician. J Psychiatr
Res. 1975;12:189-98. [PMID: 1202204]
24. Blessed G, Tomlinson BE, Roth M. The association between quantitative
measures of dementia and of senile change in the cerebral grey matter of elderly
subjects. Br J Psychiatry. 1968;114:797-811. [PMID: 5662937]
25. Piccininni M, Baldereschi M, Teri L, Campani D, Piccini C, Amaducci L,
et al. Revised Memory and Behavior Problems Checklist (RMBPC): An inter-
observer agreement study. Alzheimer’s Reports. 1998;1:321-5.
26. Teri L, Truax P, Logsdon R, Uomoto J, Zarit S, Vitaliano PP. Assessment
of behavioral problems in dementia: the revised memory and behavior problems
checklist. Psychol Aging. 1992;7:622-31. [PMID: 1466831]
27. Walter LC, Covinsky KE. Cancer screening in elderly patients: a frame-
work for individualized decision making. JAMA. 2001;285:2750-6. [PMID:
11386931]
28. Cox DR. Regression models in life tables (with discussion). Journal of the
Royal Statistical Society. Series B. 1972;34:187-220.
29. Grambsch PM, Therneau TM. Proportional hazard tests and diagnostics
based on weighted residuals. Biometrika. 1994;81:515-26.
30. Doody RS, Massman P, Dunn JK. A method for estimating progression
rates in Alzheimer disease. Arch Neurol. 2001;58:449-54. [PMID: 11255449]
31. Geerlings MI, Deeg DJ, Schmand B, Lindeboom J, Jonker C. Increased risk
of mortality in Alzheimer’s disease patients with higher education? A replication
study. Neurology. 1997;49:798-802. [PMID: 9305343]
32. Wilson RS, Schneider JA, Beckett LA, Evans DA, Bennett DA. Progression
of gait disorder and rigidity and risk of death in older persons. Neurology. 2002;
58:1815-9. [PMID: 12084882]


APPENDIX: SENSITIVITY ANALYSIS


Sensitivity analyses were performed to examine potential bi-
ases that may have been introduced because of the inability to
observe a decline in patients with no follow-up MMSE scores
and in patients who had very low baseline MMSE scores. Sixty-
two patients survived for more than 1 year but did not have
follow-up MMSEs. Logistic regression analyses were performed
on the patients who had follow-up MMSE scores to explore
which baseline factors predict significant cognitive decline during
the first year. We found that only the baseline MMSE score was
a significant predictor for the first year decline. Patients with
baseline MMSE scores between 10 and 20 were 2 times more
likely (odds ratio, 2.16 [95% CI, 1.43 to 3.27]; P < 0.001) to
decline during the first year than were the patients with baseline
MMSE scores between 21 and 30. Therefore, on the basis of
their baseline MMSE scores, we classified patients who had sur-
vived for more than 1 year but without follow-up MMSE scores
(62 patients) into declined (22 patients) or nondeclined (40 pa-
tients) groups. Cox regression models, in which analysis time was
truncated at the first year of follow-up, were performed on 4
different analytic samples. Model 1 was performed on declined
(146) and nondeclined (273) patients; model 2 was performed
on declined (144) and nondeclined (267) patients with baseline
MMSE scores greater than 8; model 3 was performed on declined
(146) patients, nondeclined (273) patients, and those without
follow-up MMSE scores who survived for more than 1 year (62);
and model 4 was performed on patients who were included in
model 3 but who had baseline MMSE scores greater than 8 (267
nondeclined patients, 144 declined patients, and 56 patients with
no follow-up). Results are presented in the Appendix Table. The
floor effect (that is, patients who reached the low end of MMSE
scoring at baseline and who were unable or less likely to decline
at follow-up) leads to slightly underestimating the risk for dying
for the declined group. The follow-up MMSE data on 62 pa-
tients that were missing at the 1-year follow-up were not found
to bias the estimated risk for dying for the declined group. We
note that of the 102 patients with no follow-up MMSE scores,
40 (39.2%) died during the first year after receiving a diagnosis
of Alzheimer disease. The 40 patients who died earlier might
have declined faster but were not observed during the first year.
Thus, follow-up MMSE data on these patients could not be used
in this analysis.