Anda di halaman 1dari 6

Patterns of Chronic Multimorbidity in the Elderly Population

Alessandra Marengoni, MD, PhD, wz Debora Rizzuto, MS, w Hui-Xin Wang, PhD, w
Bengt Winblad, MD, PhD, w and Laura Fratiglioni, MD, PhD w

OBJECTIVES: To describe patterns of comorbidity and to identify possible preventative strategies. J Am Geriatr Soc
multimorbidity in elderly people. 57:225–230, 2009.
DESIGN: A community-based survey.
SETTING: Data were gathered from the Kungsholmen Pro- Key words: chronic diseases; comorbidity; multimorbidity;
ject, a urban, community-based prospective cohort in Sweden. old age
PARTICIPANTS: Adults aged 77 and older living in the
community and in institutions of the geographically defined
Kungsholmen area of Stockholm (N 5 1,099).
MEASUREMENTS: Diagnoses based on physicians’ exam-
inations and supported by hospital records, drug use, and
blood samples. Patterns of comorbidity and multimorbidity
were evaluated using four analytical approaches: prevalence
figures, conditional count, logistic regression models, and
D espite the increasing prevalence of chronic conditions
with age, knowledge about how diseases distribute or
co-occur in the same individual is limited. Our knowledge is
cluster analysis. incomplete because few studies have attempted to describe
RESULTS: Visual impairments and heart failure were the overall pattern of diseases. In these, a range of different
the diseases with the highest comorbidity (mean 2.9 and approaches have been used, with only one approach used in
2.6 co-occurring conditions, respectively), whereas demen- any one study and no agreement in the field about the best
tia had the lowest (mean 1.4 comorbidities). Heart failure approach to use.1 This lack of consistency makes it difficult
occurred rarely without any comorbidity (0.4%). The to meaningfully compare the findings of different studies. A
observed prevalence of comorbid pairs of conditions fundamental problem in the evaluation of combinations
exceeded the expected prevalence for several circulatory of clinical conditions is the lack of consistent definitions
diseases and for dementia and depression. Logistic regres- of comorbidity and multimorbidity. The term comorbidity
sion analyses detected similar comorbid pairs. The cluster should refer to the combination of additional diseases be-
analysis revealed five clusters. Two clusters included vascular yond an index disorder.2 This definition implies that the
conditions (circulatory and cardiopulmonary clusters), and main interest is in the index condition and the possible
another included mental diseases along with musculoskeletal effects of the other disorders on the index condition, for
disorders. The last two clusters included only one major instance, on the prognosis of the index condition. In
disease each (diabetes mellitus and malignancy) together with contrast, multimorbidity is defined as any co-occurrence
their most common consequences (visual impairment and of two or more chronic or acute diseases and medical con-
anemia, respectively). ditions within one person, whether coincidental or not,3
CONCLUSION: In persons with multimorbidity, there indicating a shift of interest from a given index condition to
exists co-occurrence of diseases beyond chance, which individuals who suffer from multiple disorders.4 In clinical
clinicians need to take into account in their daily practice. practice, the term multimorbidity helps to shift from a dis-
Some pathological mechanisms behind the identified ease-based perspective to an individual-based perspective.
clusters are well known; others need further clarification A second problem is the use of different methods of explor-
ing the co-occurrence of diseases,5 such as conditional
From the Aging Research Center, Department of Neurobiology, Care Sci- count, proportion of people with pairs of comorbid dis-
ences and Society, Karolinska Institutet, Stockholm, Sweden; wStockholm eases, relative association measured using odds ratios, and
Gerontology Research Center, Stockholm, Sweden; and zGeriatric Unit, Civili
Hospital, Department of Medical and Surgery Sciences, University of Brescia, cluster analysis. Conditional count is the number of chronic
Brescia, Italy. diseases occurring with a particular index disease. This
Address correspondence to Alessandra Marengoni, I Medicina, Spedali Civili. approach is useful when studying one particular condition
Piazzale Spedali Civili 1. 25123 Brescia, Italy. E-mail: Alessandra.Marengoni (e.g., arthritis) and its comorbid conditions.6 The results depend strongly on the number of conditions evaluated in
DOI: 10.1111/j.1532-5415.2008.02109.x the study. The Women’s Health and Aging Study7 found

JAGS 57:225–230, 2009

r 2009, Copyright the Authors
Journal compilation r 2009, The American Geriatrics Society 0002-8614/09/$15.00

that the most common comorbid pair was arthritis and chronic disorder when diagnosed by the examining physi-
visual impairment (44%). Another study6 used the same cian or when detected in the computerized Stockholm In-
approach in community-dwelling individuals aged 55 and patient Register, as described below. Physicians asked the
older and found that the combination of arthritis and high participants to show prescription forms or drug containers
blood pressure was the most commonly occurring comor- of the drugs used. Drugs were registered according to the
bid pair (21.1%). This approach, as well as the estimation Anatomical Therapeutic Chemical classification system.12
of the odds ratios, is useful in assessing the degree to which At the end, the physician coded all diagnoses of current
comorbid diseases exceed a level expected by chance alone.6 diseases following the International Classification of Dis-
Conditions that co-occur can have a noncausal statistical eases, Ninth Revision.13 There were a few exceptions to this
association or can share a common pathophysiological procedure. Deafness was defined as being unable to hear the
cause.8 Finally, the cluster analysis is a descriptive technique interviewer’s voice, and visual impairment was defined as
that considers how variables tend to occur in conjunction being blind or almost blind (being unable to see the phy-
with each other. With this method, it is possible to go be- sician at a close distance). A psychiatrist made the diagnosis
yond simple comorbid pairs and to obtain a general overall of major depression according to Diagnostic and Statistical
picture of the broad pattern of how diseases are associated Manual of Mental Disorders (DSM), Fourth Edition, cri-
in a particular population and where a particular disease of teria14 using data from the Comprehensive Psychopatho-
interest appears in the pattern. One study9 used cluster logical Rating Scale.15 Two different physicians made the
analysis to describe the distribution of diseases in a sample diagnosis of dementia and different dementia types; in case
of elderly Native Americans. It found that diseases aggre- of disagreement, a third physician made a final diagnosis
gated in two major clusters: cardiopulmonary and sensory using the DSM, Third Edition, Revised, diagnostic crite-
motor. ria.16 Anemia was defined as a hemoglobin level less than
The aims of the current study were to evaluate patterns 13 g/dL in men and less than 12 g/dL in women, according
of comorbidity and multimorbidity in a Swedish population to the World Health Organization definition.17 Further in-
aged 77 and older using four different analytical approaches formation on health status for all participants was derived
(prevalence figures, conditional count, logistic regression from the computerized Stockholm Inpatient Register
models, and cluster analysis) and to compare the results System, which encompasses all hospitals in the Stockholm
attained by these different approaches. area since 1969 and records up to six diagnoses at dis-
charge. As previously described, 30 chronic diseases were
detected in the population.18 For the purpose of this study,
METHODS the analysis focused on the 15 most common chronic con-
Study Population ditions (prevalence 43%). Multimorbidity was defined as
The Kungsholmen Project is a community-based prospective the co-occurrence of two or more chronic diseases within
study on aging and dementia that included all inhabitants in one person.
the Kunghsolmen district of Stockholm, Sweden, aged 75 and
older in October 1987.10 Informed consent was requested
Statistical Analysis
from all subjects. Of the 1,700 elderly subjects who agreed to
participate in the first examination (1987–1989), 429 died For each chronic disease, the prevalence (per 100 persons)
before the first follow-up, and 172 moved or refused the first and 95% confidence intervals (CIs) of the disease occurring
follow-up (1991–1993). Thus, 1,099 subjects participated at with and without comorbidity and the mean number and
the first follow-up, and they constituted the population of the standard deviation of co-occurring conditions were esti-
present study. At the first follow-up, each participant received mated. Expected prevalence of disease pairs was computed
a clinical examination performed by a physician. as (prevalence of disease A)  (prevalence of disease B) and
compared with the observed co-prevalence. If disorders
are completely independent of one another, they can be ex-
Data Collection pected to co-occur at a rate that equals the product of the
All participants were examined according to a standardized prevalence of the separate conditions (coincidental comor-
protocol. The examination lasted approximately 3 hours bidity).3 When the prevalence of comorbidity exceeds the
and included a social interview, a neuropsychological bat- expected level and biases can be excluded, there must be an
tery, and a clinical examination. The clinical examination association that is not due by chance alone or a proven
covered general, neurological, and psychiatric status and pathophysiological correlation between diseases. To confirm
contained questions about medication use. Blood samples the results of the method described above and to control for
were obtained. Trained nurses interviewed the elderly sub- possible confounders, several logistic regression models
jects and their next of kin using a structured questionnaire were conducted to analyze the crude and adjusted associ-
on living conditions and social status. ation between each pair of co-occurring diseases. The ad-
justment accounted for age, sex, education, and all of the
other diseases. Finally, to analyze different patterns of as-
Chronic Diseases sociative multimorbidity, without any a priori hypothesis, a
A disease was classified as chronic if one or more of the cluster analysis was performed. The cluster analysis makes
following characteristics were fulfilled. The condition was it possible to go beyond comorbid pairs of conditions and
permanent, was caused by a nonreversible pathological considers how diseases tend to occur in conjunction with
alteration, or required rehabilitation or a long period each other. With the use of this technique, it is possible to
of care.11 A subject was classified as being affected by a obtain an overall picture of how diseases are associated in a

particular population. The distribution of diseases seen in Table 1. Persons Affected by Each Chronic Disease
the cluster should be significantly different from random Independently of Comorbidities (All Cases) and by Each
distribution. A correlation matrix was computed of all the Chronic Disease Without Any Comorbidity
conditions using the Yule’s Q measure of association and
average linkage as a combination method. To further an- Cases Co-Occurring
alyze the distribution of diseases, a cluster analysis was run All Without Conditions,
after stratification according to age (77–84 and 85) and sex. Cases, Comorbidity, Mean  Standard
Chronic Diseases n % Deviation

Visual impairment 160 15.6 2.9  1.4

Heart failure 194 2.1 2.6  1.2
The majority of the 1,099 examined subjects were women Deafness 70 10.0 2.3  1.5
(77%). Women were older (84.8  4.6 vs 83.7  4.2,
Diabetes mellitus 56 14.3 2.2  1.6
Po.01) and less educated (55.0% vs 39.2% with 2–7 years
Hip fracture 42 4.8 2.2  1.4
of education, Po.001) than men. The prevalence figures of
Atrial fibrillation 113 10.6 2.1  1.3
each chronic disease with and without comorbidity are
shown in Figure 1. Hypertension, dementia, and heart failure Cerebrovascular 81 8.6 2.1  1.4
were the most frequent disorders, with prevalences of 38%,
21%, and 18%, respectively. Coronary heart disease (CHD), Depression 89 11.2 2.1  1.5
anemia, and visual impairments reached a prevalence ranging Coronary heart disease 165 16.4 2.0  1.4
from 11% to 15%, whereas all the other disorders were less Hypertension 417 17.0 1.9  1.3
frequent (10%). Dementia and hypertension, although Malignancy 57 16.4 1.9  1.5
frequently associated with other diseases, had the highest Anemia 145 15.2 1.8  1.3
prevalence when cases with comorbidity were excluded Thyroid dysfunction 75 20.0 1.8  1.5
(7.2% and 6.5%, respectively). Hip fracture and heart fail- Chronic obstructive 59 25.4 1.7  1.5
ure were the diseases occurring most rarely without any pulmonary disease
comorbidity (0.2% and 0.4%, respectively). All the other Dementia 234 33.8 1.4  1.4
chronic conditions occurred without any comorbidity, at a
prevalence ranging from 0.6% to 2.5% (Figure 1).
Table 1 shows the number of persons affected by each
chronic condition independently of comorbidities (all cases), often occurred in pairs, mostly associated with heart failure
as well as the proportion of persons affected by each chronic and hypertension. Of the other pairs of disorders, the asso-
disease without any comorbidity and the mean number ciation between dementia and depression, hip fracture, or
of co-occurring conditions for each disease. Dementia and cerebrovascular disease (CVD) was higher than expected, as
hypertension cases were the ones occurring more frequently well as the prevalence of visual impairments and deafness. On
in the absence of other chronic conditions (33.8% and 17% the contrary, the expected prevalence of dementia and
of all cases, respectively, including cases with or without hypertension was higher than that observed. These results
comorbidities). The number of comorbid conditions varied were confirmed according to logistic regression analysis
from an average of 1.4 to 2.9. Visual impairments and heart (Table 2). Hypertension and heart failure emerged as the
failure had the highest comorbidity in terms of number of strongest associated pair of diseases. Beyond circulatory con-
co-occurring chronic conditions, whereas dementia had the ditions, dementia and depression were significantly correlated
lowest (Table 1). not only with each other but also with other diseases such
Observed and expected prevalence of the most frequently as hip fracture and CVD. Two sensorial disorders, visual im-
co-occurring pairs of conditions and odds ratios testing their pairments and deafness, were also significantly correlated.
association are reported in Table 2. Circulatory disorders Dementia and hypertension constituted the only pair with an
inverse correlation. Most associations remained significant
Hypertension after adjustment for age, sex, education, and all of the other
Dementia chronic conditions (Table 2).
Heart Failure
Coronary Heart Disease A five-cluster structure was derived from the cluster
Visual impairments analysis (Figure 2). One cluster consisted of four conditions:
Anemia hypertension, heart failure, chronic atrial fibrillation, and
Atrial Fibrillation
CVD. Three clusters consisted of three chronic conditions
Cerebrovascular disease each. The first one included thyroid dysfunction, chronic
Thyroid dysfunctions obstructive pulmonary disease (COPD), and CHD. The
Chronic Obstructive Pulmonary Disease
second one included diabetes mellitus, visual impairments,
Malignancy and deafness. The third one comprised dementia, depres-
Diabetes Mellitus sion, and hip fracture. The last of the five clusters consisted
Hip Fracture
of two diseases: malignancy and anemia. The structure of
0 5 10 15 20 25 30 35 40 the clusters for men and women was similar. There was
a sex difference in the distribution of depression, which
Figure 1. Prevalence per 100 of most frequent chronic diseases clustered with malignancy in women and with hip fracture
occurring independently of comorbidity (gray1black) or with- in men. There was also a difference between older- and
out any comorbidity (black). younger-old persons, with CVD clustering with depression

Table 2. Persons Affected by the Most Frequently Co-Occuring Pairs of Chronic Conditions and Their Observed (O)
and Expected (E) Prevalence per 100 Population
Prevalence/100 OR (95% CI)

Chronic Conditions Cases, n Observed Expected Ratio O/E Crude Adjusted

Heart failure and CHD 61 5.6 2.6 2.2 3.5 (2.5–5.0) 3.5 (2.2–5.6)
Heart failure and atrial fibrillation 42 3.8 1.8 2.1 3.2 (2.1–4.9) 3.1 (1.8–5.3)
Heart failure and diabetes mellitus 20 1.8 0.9 2.0 2.8 (1.6–4.9) 3.2 (1.6–6.6)w
Hypertension and heart failure 166 15.1 6.7 2.3 15.4 (10.1–23.6) 16.0 (10.1–25.4)
Hypertension and CHD 84 7.6 5.7 1.3 1.9 (1.3–2.6) 1.1 (0.7–1.6)
Hypertension and atrial fibrillation 60 5.5 3.9 1.4 1.9 (1.3–2.9) 1.3 (0.8–2.1)
Hypertension and CVD 38 3.5 2.8 1.3 1.5 (0.9–2.3) 1.7 (1.0–2.9)
Hypertension and dementia 60 5.5 8.1 0.7 0.5 (0.4–0.7) 0.5 (0.3–0.7)
Hypertension and diabetes mellitus 28 2.5 1.9 1.3 1.7 (2.0–1.9) 1.1 (0.6–2.2)
Dementia and depression 33 3.0 1.7 1.8 2.4 (1.5–3.7) 1.9 (1.2–3.4)w
Dementia and hip fracture 19 1.7 0.8 2.1 3.2 (1.7–6.0) 2.3 (1.1–4.8)w
Dementia and CVD 30 2.7 1.6 1.7 2.3 (1.5–2.8) 3.1 (1.7–5.4)
Depression and CVD 12 1.1 0.6 1.8 2.1 (1.1–4.1) 2.0 (1.0–4.1)
Depression and hip fracture 7 0.6 0.3 2.0 2.4 (1.02–5.5) 1.3 (0.5–3.3)
Deafness and visual impairments 24 2.2 0.9 2.4 3.4 (2.0–5.8) 3.2 (1.8–5.6)

Results from logistic regression models testing the association between pairs of chronic conditions: odds ratios (ORs; crude and adjusted for age, sex, education,
and all the other diseases) and 95% confidence intervals (CI) are reported.
Po  .001, w .01.
CHD 5 coronary heart disease; CVD 5 cerebrovascular disease.

in the younger-old persons but with dementia in the older- few cases of heart failure and hip fracture occurred without
old subjects. any comorbidity. In a recent report from the National
Health and Nutrition Examination Survey (NHANES),
the authors showed that the majority of older Americans
DISCUSSION experiencing one chronic disease had at least one other
This study used several approaches to describe comorbidity co-occurring disease,20 although the prevalence of subjects
and multimorbidity in the elderly population. Consistent affected by one specific condition without any other co-
findings were achieved using the different methods. All morbidity was higher than that found in the population of
chronic diseases were more likely to occur with comorbid the current study. This difference could be because, in the
conditions than alone, confirming previous reports NHANES study, only five chronic conditions were taken
that most elderly people have two or more disorders.8,19 into account.
Hypertension and dementia were the most frequent diseases Heart failure and visual impairment were associated
occurring with and without a comorbid disorder, whereas with the highest number of comorbid diseases and dementia
with the lowest. The finding that dementia had the lowest
comorbidity is not surprising. In fact, the health status of
–1 persons affected by dementia is still a debated issue. Previous
studies have shown controversial results about the preva-
lence of comorbidities in older adults with dementia.21,22
–.5 We are more prone to interpret this finding as an under-
Yule similarity measure

estimation of comorbidity because of the underreporting of

symptoms or difficulties in performing diagnostic procedures
rather than a real lower comorbidity in dementia.
When the most common co-occurring pairs of condi-
tions were evaluated, it was found that they were mainly
circulatory diseases, several of which exceeded the expected
prevalence and were found to have a significantly greater
relative odds of co-occurrence. The strongest association
was found between hypertension and heart failure, and
consistent with these findings, the cluster analysis identified
two groups related to vascular diseases. First, there was a
‘‘circulatory cluster’’ that was one of the main aggregations
Figure 2. Dendogram resulting from clusters analysis testing the and included the majority of circulatory diseases. The cir-
distribution and aggregation of chronic diseases in the population. culatory cluster divided into a subgroup of cardiovascular

chronic conditions such as hypertension and heart failure pairs of diseases that showed a significant association with a
and a second group including chronic atrial fibrillation and P-value o.01. Finally, the generalizability of the study is
one of its consequences: CVD. Second, there was a ‘‘cardio- limited to relatively highly educated elderly people living in
pulmonary cluster’’ that included CHD and COPD and urban areas in Western countries; the Kungsholmen Project
unexpectedly linked these two conditions with thyroid dys- cohort consisted of older individuals living in a geograph-
function.23,24 Beyond circulatory diseases, dementia and ically defined central area of Stockholm. One special feature
depression were significantly correlated not only with each of this population is that it included only two persons who
other, but also with other diseases such as hip fracture and had been farmers as their main occupation. The present
CVD and were grouped together according to the cluster study also has some advantages. First, different sources of
analysis, along with hip fracture, confirming the public medical diagnoses were employed, including direct clinical
health relevance of this disorder. The association between examination, which reduced potential ascertainment biases
dementia and depression is well established, although there that often affect the accuracy of the assessment of multi-
remains uncertainty regarding the pathophysiological morbidity in elderly people. This allowed the detection of
mechanisms.25 Because the presence of depressive symptoms usually underreported disorders, such as mental diseases.
could be underestimated in persons affected by moderate to Second, the study population covered persons living at
severe dementia due, for example, to a poorer ability to com- home and in institutions, providing a more complete picture
municate symptoms, the association that was found between of the elderly population.
these two diseases could have been underestimated. Hip frac- In conclusion, multimorbidity is a common feature in
ture can be easily interpreted as a consequence of dementia older persons, and moreover, the co-occurring diseases
because of the link between osteoporosis, less mobility, and often cluster together beyond what would be expected by
greater risk of fall due to balance problems in persons with chance. Five major clusters were identifiedFtwo linked to
dementia.26 Finally, diabetes mellitus and malignancy vascular diseases with hypertension and heart failure playing
emerged as relevant disorders grouped in two clusters to- the main role and the others to dementia, diabetes mellitus,
gether with their major consequences (diabetes mellitus with and malignancy. The results of the cluster analysis could be
visual impairment and malignancy with anemia). A study particularly useful for research and clinical purposes,
evaluating the health status of Native Americans aged 60 and and the evidence that disease A clusters with disease
older found four clusters, two of which were similar to the B could suggest new pathophysiological correlations or
ones that emerged in the current population. It found that disease-related susceptibility. For example, an association
several circulatory conditions clustered together, as well as was found between deafness and visual impairment consis-
two sensorial problems: visual and hearing impairment.9 tently across different methods. Further evaluation and
Stratifying the population according to age groups and clarification of the possible causal pathways linking
sex did not substantially change the results of the cluster these two conditions are needed. Similarly, the association
analysis conducted in the whole sample, although some between hip fracture and dementia may suggest common
interesting differences were found. For example, in younger pathophysiological mechanisms such as estrogen deficiency
persons, CVD clustered together with depression, whereas leading to dementia and osteoporosis.29 For clinical prac-
in the older persons, it clustered with dementia. These find- tice, better understanding of the type of relationships
ings suggest that cerebrovascular pathology may lead to the existing between co-occurring diseases could lead to the
development of depressive syndromes in relatively young detection of specific targets for intervention. The clustering
elderly people, whereas in very old persons it may be more of conditions such as dementia, depression, and hip fracture
often correlated with cognitive impairment. Alternatively, suggests that subjects affected by one of these conditions
depression was in the same cluster as malignancy in women, could be at risk of developing the others. This kind of infor-
whereas in men it occurred in the hip fracture cluster, mation could be particularly relevant for the application of
suggesting that different diseases can have different effects tertiary prevention.
on the affective status of men and women. Finally, multimorbidity requires complex therapy and
Dementia and hypertension emerged as the only disease care, which demands special attention, knowledge, and
pair inversely correlated. The reduction of blood pressure levels skills for clinicians, nurses, and families. It has been shown
in preclinical and clinical stages of dementia could explain this that, in older adults with multiple chronic diseases, disor-
finding.27 A second explanation could be that there is a ders not designated as the ‘‘primary’’ condition are under-
protective effect of antihypertensive medications on the devel- treated.30 Evidence-based medical treatment has usually
opment of dementia, as has been demonstrated previously.28 been directed toward single diseases, yet ignoring concom-
Some limitations need to be discussed. First is the pos- itant diseases may lead to harm.31
sible underestimation of disorders such as musculoskeletal
conditions,18 mainly because the Kungsholmen Project
started in the late 1980s, when patients and physicians often ACKNOWLEDGMENTS
underreported diagnoses of arthritis and osteoporosis. Conflict of Interest: The editor in chief has reviewed the
Another disease that was probably underestimated was conflict of interest checklist provided by the authors and has
renal failure, because the laboratory data necessary for this determined that the authors have no financial or any other
diagnosis were not available. Second is the matter of kind of personal conflicts with this manuscript. This study
multiple comparisons. To minimize this problem, the has been supported by the Swedish Council for Working
analysis were limited to diseases with a prevalence greater Life and Social Research.
than 3% in this population. Moreover, possible common Author Contributions: Dr. Marengoni had full access
pathophysiological pathways were discussed only for those to all of the data in the study and takes responsibility for the

integrity of the data and the accuracy of the data analysis. 14. American Psychiatric Association. Diagnostic and Statistical Manual of Men-
Study design: Dr. Marengoni and Prof. Fratiglioni. Analysis tal Disorders, 4th Ed. Washington, DC: American Psychiatric Association,
1994, pp 133–158.
and interpretation: Dr. Marengoni, Dr. Rizzuto, Dr. Wang, 15. Åsberg M, Montgomery SA, Perris C et al. A comprehensive psychopatho-
Prof. Winblad, and Prof. Fratiglioni. Manuscript prepara- logical rating scale. Acta Psychiatr Scand 1978;271:5–27.
tion: Dr. Marengoni and Prof. Fratiglioni. All of the authors 16. American Psychiatric Association. Diagnostic and Statistical Manual of
read and approved the manuscript for publication. Mental Disorders, 3rd Ed., Revised. Washington, DC: American Psychiatric
Association, 1987, pp 97–163.
Sponsor’s Role: None. 17. World Health Organization. Nutritional anemias. Report of a WHO scientific
group. Geneva, Switzerland, 1968, Technical Report Series No. 405.
18. Marengoni A, Winblad B, Karp A et al. Prevalence of chronic diseases and
multimorbidity in the old population: Differential distribution by age, gender
and socio-economic status. Am J Public Health 2008;98:1198–1200.
REFERENCES 19. Wolff JL, Starfield B, Anderson G. Prevalence, expenditures, and complica-
1. Gijsen R, Hoeymans N, Schellevis FG et al. Causes and consequences of tions of multiple chronic conditions in the elderly. Arch Intern Med 2002;162:
comorbidity: A review. J Clin Epidemiol 2001;54:661–674. 2269–2276.
2. Feinstein AR. The pretherapeutic classification of comorbidity in chronic 20. Weiss CO, Boyd CM, Yu Q et al. Patterns of prevalent major chronic disease
disease. J Chronic Dis 1970;23:455–468. among older adults in the United States. JAMA 2007;298:1160–1162.
3. van den Akker M, Buntinx F, Knottnerus JA. Comorbidity or multimor- 21. Lyketsos CG, Toone L, Tschanz J et al. Population-based study of medical
bidity: What’s in a name? A review of literature. Eur J Gen Pract 1996;2: comorbidity in early dementia and ‘‘cognitive impairment, no dementia
65–70. (CIND)’’: Association with functional and cognitive impairment: The Cache
4. Batstra L, Bos EH, Neeleman J. Quantifying psychiatric comorbidity: Lessons County Study. Am J Geriatr Psychiatry 2005;13:656–664.
from chronic disease epidemiology. Soc Psychiatry Psychiatr Epidemiol 2002; 22. Sanderson M, Wang J, Davis DR et al. Co-morbidity associated with dementia.
37:105–111. Am J Alzheimers Dis Other Dem 2002;17:73–78.
5. Guralnik JM. Assessing the impact of comorbidity in the older population. 23. Sin DD, Paul Man SF. Why are patients with chronic obstructive pulmonary
Ann Epidemiol 1996;6:376–380. disease at increased risk of cardiovascular diseases? Circulation 2003;107:
6. Verbrugge LM, Lepkowski JM, Imanaka Y. Comorbidity and its impact on 1514–1519.
disability. Milbank Q 1989;67:450–484. 24. Singh S, Duggal J, Molnar J et al. Impact of subclinical thyroid disorders
7. Fried LP, Bandeen-Roche K, Kasper JD et al. Association of comorbidity with on coronary heart disease, cardiovascular and all-cause mortality: A meta-
disability in older women: The Women’s Health and Aging Study. J Clin analysis. Int J Cardiol 2008;125:41–48.
Epidemiol 1999;52:27–37. 25. Berger AK, Fratiglioni L, Forsell Y et al. The occurrence of depressive symp-
8. van den Akker M, Buntinx F, Metsemakers JF et al. Multimorbidity in general toms in the preclinical phase of AD. A population-based study. Neurology
practice: Prevalence, incidence, and determinants of co-occurring chronic and 1999;53:1998–2002.
recurrent diseases. J Clin Epidemiol 1998;51:367–375. 26. Sheridan PL, Hausdorff JM. The role of higher-level cognitive function in gait:
9. John R, Kerby DS, Hagan Hennessy C. Patterns and impact of comorbidity Executive dysfunction contributes to fall risk in Alzheimer’s disease. Dement
and multimorbidity among community-resident American Indian elders. Geriatr Cogn Disord 2007;24:125–137.
Gerontologist 2003;43:649–660. 27. Guo Z, Viitanen M, Winblad B et al. Low blood pressure and incidence of
10. Fratiglioni L, Viitanen M, Backman L et al. Occurrence of dementia in advanced dementia in a very old sample: Dependent on initial cognition. J Am Geriatr
age: The study design of the Kungsholmen Project. Neuroepidemiology 1992; Soc 1999;47:723–726.
11:29–36. 28. Qiu C, Winblad B, Fastbom J et al. Combined effect of APOE genotype, blood
11. Timmreck TC, Cole GE, James G et al. Health education and health promo- pressure, and antihypertensive drug use on incident AD. Neurology 2003;61:
tion: A look at the jungle of supportive fields, philosophies and theoretical 655–660.
foundations. Health Educ 1987;18:23–28. 29. Xu H, Wang R, Zhang YW et al. Estrogen, beta-amyloid metabolism/trafficking,
12. World Health Organization. Guidelines for ATC Classification. WHO Col- and Alzheimer’s disease. Ann NY Acad Sci 2006;1089:324–342.
laborating Centre for Drug Statistics Methodology, Norway and Nordic 30. Redelmeier DA, Tan SH, Booth GL. The treatment of unrelated disorders in
Councils on Medicines, Sweden: 1990. patients with chronic medical diseases. N Engl J Med 1998;338:1516–1520.
13. World Health Organization. International Classification of Diseases, 31. Boyd CM, Darer J, Boult C et al. Clinical practice guidelines and quality of
Injuries, and Causes of Death. Ninth Revision (ICD-9). Geneva, Switzerland: care for older patients with multiple comorbid diseases: Implications for pay
World Health Organization, 1987. for performance. JAMA 2005;294:716–724.