Accepted Manuscript

Title: Multimorbidity and health-related quality of life among

the community-dwelling elderly: a longitudinal study

Authors: Jiayi Gu, Jianqian Chao, Wenji Chen, Hui Xu,

Ruizhi Zhang, Tingting He, Lin Deng

PII: S0167-4943(17)30309-6
DOI: https://doi.org/10.1016/j.archger.2017.10.019
Reference: AGG 3575

To appear in: Archives of Gerontology and Geriatrics

Received date: 24-6-2017

Revised date: 21-10-2017
Accepted date: 23-10-2017

Please cite this article as: Gu, Jiayi, Chao, Jianqian, Chen, Wenji, Xu, Hui, Zhang,
Ruizhi, He, Tingting, Deng, Lin, Multimorbidity and health-related quality of life among
the community-dwelling elderly: a longitudinal study.Archives of Gerontology and
Geriatrics https://doi.org/10.1016/j.archger.2017.10.019

This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.
Multimorbidity and health-related quality of life among the

community-dwelling elderly: a longitudinal study

Jiayi Gu 1, Jianqian Chao 1*, Wenji Chen 2, Hui Xu 3, Ruizhi Zhang 1, Tingting He 1,

Lin Deng 1

1
Key Laboratory of Environmental Medicine Engineering of Ministry of Education,

Department of Medical Insurance, School of Public Health, Southeast University,

Nanjing, Jiangsu, China

2
Department of General Practice, Zhongda Hospital, Affiliated to Southeast

University, Nanjing, Jiangsu, China

3
Hospital Office, Hospital of Qinhuai, Nanjing, Jiangsu, China

*
Corresponding author: Jianqian Chao E-mail: chaoseu@163.com

Email address: Jiayi Gu: 843899267@qq.com

Jianqian Chao: chaoseu@163.com

Wenji Chen: chwenji@163.com

Hui Xu: qhyyxh@yahoo.com

Ruizhi Zhang: 1814571929@qq.com

Tingting He: 121147869@qq.com

Lin Deng: 1194898665@qq.com

1
Highlights
 Multimorbidity was found to be negatively associated with health-related quality

of life of the community-dwelling elderly.

 Three multimorbidity patterns were identified and mainly labelled as

degenerative disorders, digestive/respiratory disorders, cardiovascular/metabolic

disorders, respectively.

 Distinct multimorbidity patterns had various impacts on different dimensions of

health-related quality of life.

Abstract

Objectives: Multimorbidity is a growing public health problem. The objective of this

study was to investigate the impact of multimorbidity on health-related quality of life

(HRQoL) of the elderly.

Methods: A 24-month longitudinal study was conducted on the community-dwelling

elderly. There were 411 elderly persons with complete follow-up. Information on

thirteen chronic conditions was collected at baseline. Via a multi-dimensional scale,

HRQoL was measured at baseline, 18 and 24 months post-baseline, respectively.

Exploratory factor analyses were performed to identify multimorbidity patterns. The

linear mixed effects models were conducted to analyze the associations between all

dimensions of HRQoL and multimorbidity including distinct multimorbidity patterns.

Results: Multimorbidity was found to be negatively associated with HRQoL except

memory function. We identified three multimorbidity patterns, which were mainly

labelled as degenerative disorders, digestive/respiratory disorders,

2
cardiovascular/metabolic disorders, respectively. And three multimorbidity patterns

were associated with lower HRQoL including general health, body function, self-care

ability and social adaptability. Besides, the elderly with the multimorbidity pattern

mainly labelled as digestive/respiratory disorders or cardiovascular/metabolic

disorders had a decline on emotion than those without multimorbidity. According to

the analysis of the longitudinal data of the sample, general health, self-care ability,

emotion and social adaptability of the participants decreased in different degrees

every month.

Conclusions: Multimorbidity was associated with lower HRQoL of the

community-dwelling elderly. Distinct multimorbidity patterns had various impacts on

different dimensions of HRQoL. Further studies should be carried out to investigate

effective measures to improve HRQoL of the elderly with multimorbidity.

Key words Multimorbidity; health-related quality of life (HRQoL); chronic disease;

older people; China

1. Introduction

Multimorbidity, known as the occurrence of two or more chronic conditions, is a

growing problem due to the high prevalence and the large burden of diseases (Fortin

et al., 2007). In addition, there were also relevant studies defining “3+ chronic

conditions simultaneously in one person” as multimorbidity (Harrison et al., 2014). In

the count method, there seems to be no consensus on the definition of multimorbidity.

3
The previous study reported that about 25% of adults suffered from multimorbidity

and its prevalence among the elderly was higher (Boyd et al., 2007). A study in China

reported that the prevalence of multimorbidity among the rural elderly had even

reached 90% (Wang et al., 2015). The elderly have been regarded as a high risk group

for multimorbidity (Joseph et al., 2015). To understand the impact of multimorbidity

on the elderly, it is not enough to investigate its prevalence. Further studies are

required to determine the impact of multimorbidity on health outcomes, including

health-related quality of life (HRQoL) (Kirchberger et al., 2012). As a holistic concept,

HRQoL reflects various aspects of health status by capturing a series of health

indicators (Wang et al., 2017).

So far studies that focused on the impact of multimorbidity on HRQoL have been

based on two following categories: 1) the number of chronic conditions and 2) the

pattern of chronic conditions (Le Reste et al., 2013; Marengoni et al., 2011). And

there is evidence that HRQoL decreases with the increase in the number of

co-occurring chronic conditions (Fortin et al., 2004). However, it is unlikely to

capture the full impact of multimorbidity on HRQoL by the number-based count of

chronic conditions (Wei et al., 2016). Specific chronic conditions may be associated

with disorders in different functions, and specific groups of chronic conditions may

have diverse impacts on different dimensions of HRQoL (Kolappa et al., 2013). Thus,

it is necessary to understand the impact of multimorbidity on HRQoL across its

different patterns.

The previous study suggested that methodology should be carefully considered in the

4
study of HRQoL of persons with chronic conditions (Tyack et al., 2016). Repeated

measurements over a certain period are required, since HRQoL may change over time

and an error may be introduced in the single measurement in a cross-sectional design

(Fortin et al., 2004; Tyack et al., 2016). Therefore, a longitudinal design is logical, as

this design permits the changes of HRQoL to be examined separately from differences

between individuals at baseline (Donald and Gibbons, 2006). So far, most studies on

the factors including multimorbidity that affected HRQoL of persons with chronic

conditions have been conducted with the cross-sectional designs (Fortin et al., 2004).

Thus, we conducted a longitudinal study on the community-dwelling elderly in China.

The objectives of this study were to investigate the impact of multimorbidity on

HRQoL of the elderly and to relate multimorbidity patterns to HRQoL on the basis of

identifying multimorbidity patterns.

2. Methods

2.1. Study design and participants

A 24-month longitudinal study was carried out in the communities governed by a

Community Health Service Center in Nanjing, Jiangsu province, China. The sample

was taken from a previous cross-sectional study on the elderly (Gu et al., 2017). By a

simple random sampling, we selected 450 participants to join this present study. The

sample was taken from the community-dwelling elderly (≥ 60 years). Besides, the

participants should have lived for a long time (≥ 2 years) in the community. Exclusion

criteria were as follows: cognitive deficits; severe communication disabilities.

5
Participants were contacted by telephone and 97% of the participants (N = 437)

consented to take part in this study. At the start of this study, every participant was

informed of the study content and signed an informed form. All participants were

interviewed by trained interviewers in the Community Health Service Center. Follow

up data were collected for three times, at baseline, 18 and 24 months post-baseline,

respectively. And this span was believed sufficient to observe longitudinal changes in

HRQoL as well as the impact of multimorbidity. The overall response rate was 91%

(N = 411). Due to the missing rate with less than 10%, complete data were used for

data analyses. The study was approved by the Ethics Committee of Clinical Research

of Zhongda Hospital, Affiliated to Southeast University.

2.2. Multimorbidity and chronic conditions

Multimorbidity was defined as “two or more” chronic conditions occurring at the

same time. At baseline, thirteen chronic conditions were assessed, including

hypertension, diabetes, joint disease (e.g. arthritis), cataract, hearing disorder,

dyslipidaemia, coronary heart disease, stroke, kidney diseases, gastrointestinal

diseases, lung diseases, liver diseases and cancer. In this study, gastrointestinal

diseases mainly included ulcer disease peptic, chronic gastritis, cholecystitis, colitis;

acute gastrointestinal diseases were excluded. Lung diseases included emphysema,

chronic bronchitis, chronic obstructive pulmonary disease, asthma; pulmonary

tuberculosis was excluded. The detailed identification of chronic conditions has been

described in that previous study (Gu et al., 2017).

2.3. Health-related quality of life

6
HRQoL was measured via a multi-dimensional scale at baseline, 18 and 24 months

post-baseline. This scale designed in our previous study, was used as an instrument for

measuring HRQoL and has been validated for reliability and validity on the basis of a

large sample of older adults (Chao et al., 2013a). The indicators for reliability and

validity are detailed as follows. The Cronbach's coefficient of this scale was 0.789,

and the Cronbach's coefficient of each dimension ranged from 0.684 to 0.903; the

reliability of the entire scale was 0.873 for equal-length Spearman and 0.865 for

Guttman's split-half. According to factor analyses for structural validity, the

Kaiser-Meyer-Olkin was 0.761, and the Bartlett statistic was χ2 = 14231.9, P<0.001;

the extracted factors could interpret 67.37% of the total variance. The

above-mentioned indicators suggested high reliability and validity for the

measurement of the health status of older Chinese adults (Chao et al., 2013b). This

scale comprised the following five dimensions: body function, self-care ability,

emotion, memory function and social adaptability. There were 26 positive and 13

negative indicators in the scale. The scoring method for each indicator was as follows:

for the positive indicators, the indicator with five categories was scored the integer

from 1 to 5, the indicator with three categories was scored 1, 3 and 5, the indicator

with two categories was scored 1 and 5; for the negative indicators, the indicator with

five categories was scored the integer from 5 to 1, the indicator with three categories

was scored 5, 3 and 1, the indicator with two categories was scored 5 and 1. The

weights of every indicator and dimension defined by the Analytic Hierarchy Process

are shown in supplementary file 1. The score of every dimension was calculated via

7
the weighted average method. General health that reflected the composite score of

HRQoL could be calculated as follows:

GH=0.376*BF+0.215*SA+0.215*EP+0.074*MF+0.121*SA

GH, BF, SA, EP, MF and SA represented general health, body function, self-care

ability, emotion, memory function and social adaptability, respectively.

2.4. Covariates

Covariates in this study included gender, age (60~69, 70~79, ≥80 years), education

level (primary school or lower, middle school, college or higher), living condition

(living with spouse, living alone, living with children), smoking status (≥ 1 cigarette

per day for consecutive or accumulative half a year) (Dai et al., 2015), sufficient

physical exercise (≥ 4 times per week) (Dai et al., 2015). Covariate information,

including living condition, smoking status and sufficient physical exercise, was

collected at baseline, 18 and 24 months post-baseline.

2.5. Statistical methods

Descriptive statistics were used to describe the characteristics of the sample. Baseline

data were reported as frequencies and percentages for count data. Exploratory factor

analyses were performed to identify multimorbidity patterns. In view of dichotomous

variables, the principal factor method was performed and based on a tetrachoric

correlation matrix. The number of the patterns was determined if the eigenvalue was

greater than 1. The sampling adequacy for adopting factor analyses was estimated by

calculating the Kaiser-Meyer-Olkin (KMO). The chronic condition included in any

pattern was determined if its loading was greater than 0.25 in the corresponding

8
pattern (Wang et al., 2015). An oblique rotation (Oblimin) was applied to facilitate the

interpretation of the factor. The frequencies of the patterns were estimated by

assigning individuals to a pattern if individuals suffered from at least two conditions

with the loading of at least 0.25 on the corresponding pattern (Kirchberger et al.,

2012). The linear mixed effects models were conducted to analyze the associations

between all dimensions of HRQoL and multimorbidity including distinct

multimorbidity patterns. The scores on every dimension and general health were

regarded as the dependent variable. Chronic condition (different categories in the

different models), covariates and time variable (on the basis of months) were entered

into the models as fixed effects. The slope against time variable and the

subject-specific effect for the intercept were entered as random effects. All hypothesis

tests used two-sided tests and set alpha at 0.05. Statistics analyses were performed

using SPSS 17.0 (SPSS, Inc., Chicago, Illinois). The figures were generated using R

software (version3.2.5).

3. Results

3.1. General characteristics of the participants at baseline

Table 1 shows general characteristics of the participants. The average age of the

participants was 70.93 years (standard deviation: 6.33) and females accounted for

54%. Hypertension was the chronic condition with the highest prevalence (48.2%).

Two or more chronic conditions were reported by 56.5% of the elderly. Among the

participants with multimorbidity, two chronic conditions were the most common and

9
accounted for one fifth of the sample.

3.2. Analyses of the association between multimorbidity and HRQoL

Fixed effects in the linear mixed effects models were shown in Table 2. In the models,

chronic condition was entered into the model as a categorical variable (multimorbidity,

no multimorbidity). No multimorbidity was regarded as the combined group of 0 and

1 chronic condition. Compared with no multimorbidity, multimorbidity was

negatively associated with HRQoL, which included general health (β: -0.121), body

function (β: -0.168), self-care ability (β: -0.066), emotion (β: -0.124) and social

adaptability (β: -0.072). There were gender, age, smoking and sufficient physical

exercise as the significant covariates, which affected different dimensions of HRQoL.

And the results presented that there was a monthly decline on general health (-0.004),

self-care ability (-0.002), emotion (-0.010) and social adaptability (-0.004) among the

participants.

3.3. Results of multimorbidity patterns identified by factor analyses

Three multimorbidity patterns identified by factor analyses were shown in Table 3.

That KMO was 0.630 indicated a moderate sampling adequacy. And the percentage of

the cumulative variance was 41.603%. The first pattern (pattern 1), mainly labelled as

degenerative disorders, comprised cataract, joint disease, hearing disorder, cancer and

its prevalence was 14.599%. The second pattern (pattern 2), labelled as digestive and

respiratory disorders, was characterized by lung diseases, gastrointestinal diseases,

10
liver diseases and its prevalence was 4.136%. The third pattern (pattern 3), mainly

labelled as cardiovascular and metabolic disorders, comprised hypertension, diabetes,

coronary heart disease, stroke, kidney diseases, dyslipidaemia and its prevalence was

33.090%. There were 2 (0.487%) participants with pattern 1 and 2, 28 (6.813%)

participants with pattern 1 and 3, 4 (0.973%) participants with pattern 2 and 3.

3.4. Analyses of the associations between multimorbidity patterns and HRQoL

Analyses of the associations between multimorbidity patterns and HRQoL were

shown in Table 4. Participants with two multimorbidity patterns were excluded from

the analyses due to possible interactions between the patterns. Compared with no

multimorbidity, three patterns were negatively associated with HRQoL; and their

significant dimensions and coefficients were as follows: general health (βpattern1:

-0.129, βpattern2: -0.362, βpattern3: -0.124), body function (βpattern1: -0.157, βpattern2: -0.482,

βpattern3: -0.149), self-care ability (βpattern1: -0.114, βpattern2: -0.301, βpattern3: -0.062),

emotion (βpattern2: -0.331, βpattern3: -0.160) and social adaptability (βpattern1: -0.122,

βpattern2: -0.183, βpattern3: -0.058).

The changes of mean scores on every dimension and general health were shown in

Fig. 1. In the characteristics of the sample, the mean scores of various groups on

general health and every dimension were fluctuant over the observation period.

Compared with the participants without multimorbidity, the participants with any

multimorbidity pattern have lower mean scores on general health and every

11
dimension except memory function at baseline. At other testing time points,

participants without multimorbidity were at a higher level on general health, self-care

ability and social adaptability.

4. Discussion

Based on thirteen chronic conditions and the count method (two or more conditions),

56.5% of the participants were identified with multimorbidity in our study. Another

study the United Kingdom reported that the prevalence of multimorbidity among

older adults registered in general practices was 81% (Kadam et al., 2007). The

difference in the prevalence between the two studies might be partly due to our

sample from the community-dwelling elderly, which were relatively healthy elderly

population. Our analyses suggested that multimorbidity was significantly associated

with the decline of HRQoL, which covered general health, body function, self-care

ability, emotion and social adaptability. Previous studies regarding the association

between multimorbidity and HRQoL have been carried out on different cut-off values

of the number-based count of multimorbidity (Ramond-Roquin et al., 2016; Wang et

al., 2017). Their results uniformly suggested a negative association between the

number of chronic conditions and HRQoL. For example, stratified analyses for the

general population in Australia indicated that the elderly (≥65 years) with 3 or more

chronic conditions had lower HRQoL than those with 2 or more chronic conditions

(Wang et al., 2017).

12
Our study made a further contribution by investigating the impacts of distinct

multimorbidity patterns on HRQoL of the elderly. First, we identified three

multimorbidity patterns, which were mainly labelled as degenerative disorders,

digestive/respiratory disorders, cardiovascular/metabolic disorders, respectively. A

study on the elderly population (≥65 years) in Germany identified four multimorbidity

patterns: cardiovascular/metabolic disorders, liver/lung/joint/eye disorders,

mental/neurologic disorders, gastrointestinal disorders and cancer (Kirchberger et al.,

2012). Generally speaking, it is difficult to compare results across different studies on

multimorbidity, since there are remarkable differences in the respects regarding the

sample, the number and types of chronic conditions. However, there are some

similarities with our findings, for example, both studies found the pattern labelled as

cardiovascular/metabolic disorder. And another study on the rural elderly (≥60 years)

in China also identified a multimorbidity pattern characterized by stroke, hypertension,

diabetes, dyslipidemia (Wang et al., 2015). Thus, the identified multimorbidity

patterns substantially reflect the common antecedents and pathways of disease to

some extent (Jackson et al., 2015). Further analyses in our study indicated that three

multimorbidity patterns were associated with lower HRQoL including general health,

body function, self-care ability, social adaptability. Besides, we found the elderly with

the pattern labelled as digestive/respiratory disorders or cardiovascular/metabolic

disorders had a decline on emotion than those without multimorbidity. This indicates

that identification of specific multimorbidity pattern may contribute to the prediction

of HRQoL and help to understand the impact of multimorbidity on different

13
dimensions of HRQoL. What's more, it is significant for health service planning and

targeted improvements of HRQoL of the elderly with multimorbidity, which may ease

the burden of multimorbidity. A previous study on older women in Australia identified

three multimorbidity patterns labelled as musculoskeletal/somatic,

neurological/mental health, cardiovascular, respectively and indicated the significant

associations between poorer functional ability and three multimorbidity patterns

(Jackson et al., 2015). Despite the inconsistence in inclusion criteria, both two studies

indicated that multimorbidity based on pattern identification could be associated with

the decline of physical health. The difference was that our study confirmed the

association between distinct multimorbidity patterns and mental health (emotion,

social adaptability) after controlling the effects of gender. It is worthy of note that

there were no significant associations between memory function and multimorbidity

including three multimorbidity patterns in our study. In a study on the elderly in

Europe, the result that multimorbidity was related to subjective memory complaints

indicated the association between multimorbidity and poorer objective memory

performance (Pedro et al., 2016). We speculated that the differences between these

results were caused by different types of chronic conditions included in the two

studies, for example, chronic constipation, which was included in that study, was

associated with memory complaints as a specific disease. And chronic constipation

probably mediated the association between memory function and multimorbidity.

In our study, factor analyses were applied to the identification of multimorbidity

patterns. The strengths of factor analysis were that it did not depend on the prior

14
hypothesis of grouping including the number of groups, permitted chronic conditions

cross-loading and promoted grouping naturally (Jackson et al., 2015). But there were

some limitations in our study. Firstly, most chronic conditions were identified by

self-reported information so that errors might be introduced into our study. Secondly,

the effects of treatments on HRQoL have not been analyzed in our study. Further

studies on multimorbidity and HRQoL are needed to analyze or control the effects of

a variety of therapeutic regimens for different chronic conditions and their interactions.

Thirdly, similar to another study (Jackson et al., 2015), our study identified

multimorbidity and chronic conditions at a certain time (baseline). However, the

elderly might acquire new chronic conditions during a two-year span. The impacts of

new chronic conditions on HRQoL were ignored, because our study concerned the

long-term impacts of multimorbidity on HRQoL and ensured a uniform period (2

years) of observation. There might be a lack of precision in our study. Last but not

least, the list of chronic conditions was inexhaustive in our study, and systematic

reviews suggested that the number of chronic conditions ranged from 4 to 102 in

related studies (Diederichs et al., 2011; Fortin et al., 2012). The common mental

conditions such as depression and anxiety were not included in our study. According

to previous studies in China (Liu and Meng, 2004; Tang et al., 2001; Xiaolu et al.,

2013), the cognitive level of mental health of Chinese elderly was low; due to the

influence of the oriental culture, Chinese elderly often described themselves as

physical discomfort instead of expressing themselves as depression or anxiety;

compared with other chronic conditions, the rate of active visit due to mental

15
conditions was relatively lower, so most elderly patients might lack definitive

diagnoses. In our study, information on chronic conditions were mainly collected

through self-reporting. In the absence of definitive diagnoses, the collection of

information on mental conditions might bring a bias into our study. Thus, the list of

chronic conditions in our study did not cover common mental conditions. However,

our study is still helpful to understand the association between HRQoL and

multimorbidity among the community-dwelling elderly in China.

5. Conclusions

Our study indicated that multimorbidity and distinct multimorbidity patterns were

associated with lower HRQoL including lower scores on different dimensions among

the community-dwelling elderly in China. The findings could help us to understand

and improve HRQoL of elderly persons with multimorbidity. Further studies on

multimorbidity are required to include more chronic conditions and analyze

interactions of multiple chronic conditions.

Funding

This work was supported by the National Natural Science Foundation of China (Grant

Number 81273189).

Conflict of interest

None.

16
Acknowledgements

The authors would like to thank the participants. And the authors would also like to

thank Hospital of Qinhuai, Nanjing, Jiangsu, China for providing help for this study.

17
References
Boyd, C.M., Boult, C., Shadmi, E., Leff, B., Brager, R., Dunbar, L., Wolff, J.L., Wegener, S., 2007.
Guided care for multimorbid older adults. The Gerontologist 47, 697-704.
Chao, J., Li, Y., Xu, H., Yu, Q., Wang, Y., Liu, P., 2013a. Health status and associated factors among the
community-dwelling elderly in China. Archives of gerontology and geriatrics 56, 199-204.
Chao, J.Q., Li, Y.Y., Xu, H., Yu, Q., Wang, Y.M., Liu, P., 2013b. Health status and associated factors
among the community-dwelling elderly in China. Archives of gerontology and geriatrics 56, 199-204.
Dai, H., Jia, G., Liu, K., 2015. Health-related quality of life and related factors among elderly people in
Jinzhou, China: a cross-sectional study. Public health 129, 667-673.
Diederichs, C., Berger, K., Bartels, D.B., 2011. The Measurement of Multiple Chronic Diseases-A
Systematic Review on Existing Multimorbidity Indices. J Gerontol a-Biol 66, 301-311.
Donald, H., Gibbons, R.D., 2006. Chapter 4. Mixed-Effects Regression Models for Continuous
Outcomes. John Wiley & Sons, Inc.
Fortin, M., Lapointe, L., Hudon, C., Vanasse, A., Ntetu, A.L., Maltais, D., 2004. Multimorbidity and
quality of life in primary care: a systematic review. Health Qual Life Outcomes 2, 51.
Fortin, M., Soubhi, H., Hudon, C., Bayliss, E.A., van den Akker, M., 2007. Multimorbidity's many
challenges. Brit Med J 334, 1016-1017.
Fortin, M., Stewart, M., Poitras, M.E., Almirall, J., Maddocks, H., 2012. A Systematic Review of
Prevalence Studies on Multimorbidity: Toward a More Uniform Methodology. Annals of Family
Medicine 10, 142-151.
Gu, J., Chao, J., Chen, W., Xu, H., Wu, Z., Chen, H., He, T., Deng, L., Zhang, R., 2017. Multimorbidity
in the community-dwelling elderly in urban China. Archives of gerontology and geriatrics 68, 62-67.
Harrison, C., Britt, H., Miller, G., Henderson, J., 2014. Examining different measures of multimorbidity,
using a large prospective cross-sectional study in Australian general practice. Bmj Open 4.
Jackson, C.A., Jones, M., Tooth, L., Mishra, G.D., Byles, J., Dobson, A., 2015. Multimorbidity patterns
are differentially associated with functional ability and decline in a longitudinal cohort of older women.
Age and ageing 44, 810-816.
Joseph, N., Nelliyanil, M., Nayak, S.R., Agarwal, V., Kumar, A., Yadav, H., Ramuka, G., Mohapatra,
K.T., 2015. Assessment of morbidity pattern, quality of life and awareness of government facilities
among elderly population in South India. Journal of Family Medicine & Primary Care 4, 405-410.
Kadam, U.T., Croft, P.R., Gp, N.S., 2007. Clinical multimorbidity and physical function in older adults:
a record and health status linkage study in general practice. Fam Pract 24, 412-419.
Kirchberger, I., Meisinger, C., Heier, M., Zimmermann, A.K., Thorand, B., Autenrieth, C.S., Peters, A.,
Ladwig, K.H., Doring, A., 2012. Patterns of multimorbidity in the aged population. Results from the
KORA-Age study. PLoS One 7, e30556.
Kolappa, K., Henderson, D.C., Kishore, S.P., 2013. No physical health without mental health: lessons
unlearned? Bulletin of the World Health Organisation 91, 3-3A.
Le Reste, J.Y., Nabbe, P., Manceau, B., Lygidakis, C., Doerr, C., Lingner, H., Czachowski, S., Munoz,
M., Argyriadou, S., Claveria, A., Le Floch, B., Barais, M., Bower, P., Van Marwijk, H., Van Royen, P.,
Lietard, C., 2013. The European General Practice Research Network presents a comprehensive
definition of multimorbidity in family medicine and long term care, following a systematic review of

18
relevant literature. Journal of the American Medical Directors Association 14, 319-325.
Liu, H., Meng, C., 2004. Survey on Depression of Beijing Residents over 55 Year-old. Chinese Mental
Health Journal 18, 794-795.
Marengoni, A., Angleman, S., Melis, R., Mangialasche, F., Karp, A., Garmen, A., Meinow, B.,
Fratiglioni, L., 2011. Aging with multimorbidity: a systematic review of the literature. Ageing research
reviews 10, 430-439.
Pedro, M.C., Mercedes, M.P., Ramon, L.H., Borja, M.R., 2016. Subjective memory complaints in
elderly: relationship with health status, multimorbidity, medications, and use of services in a
population-based study. International psychogeriatrics 28, 1903-1916.
Ramond-Roquin, A., Haggerty, J., Lambert, M., Almirall, J., Fortin, M., 2016. Different Multimorbidity
Measures Result in Varying Estimated Levels of Physical Quality of Life in Individuals with
Multimorbidity: A Cross-Sectional Study in the General Population. Biomed Res Int 2016, 7845438.
Tang, M., Liu, X.A., Han, H., 2001. Prevalence of Depressive Disorders among Residents aged 55 or
above in Chengdu Area. Chinese Mental Health Journal.
Tyack, Z., Frakes, K.A., Barnett, A., Cornwell, P., Kuys, S., McPhail, S., 2016. Predictors of
health-related quality of life in people with a complex chronic disease including multimorbidity: a
longitudinal cohort study. Quality of life research : an international journal of quality of life aspects of
treatment, care and rehabilitation 25, 2579-2592.
Wang, L., Palmer, A.J., Cocker, F., Sanderson, K., 2017. Multimorbidity and health-related quality of
life (HRQoL) in a nationally representative population sample: implications of count versus cluster
method for defining multimorbidity on HRQoL. Health Qual Life Outcomes 15, 7.
Wang, R., Yan, Z., Liang, Y., Tan, E.C., Cai, C., Jiang, H., Song, A., Qiu, C., 2015. Prevalence and
Patterns of Chronic Disease Pairs and Multimorbidity among Older Chinese Adults Living in a Rural
Area. PLoS One 10, e0138521.
Wei, M.Y., Kawachi, I., Okereke, O.I., Mukamal, K.J., 2016. Diverse Cumulative Impact of Chronic
Diseases on Physical Health-Related Quality of Life: Implications for a Measure of Multimorbidity.
American journal of epidemiology 184, 357-365.
Xiaolu, N., Wang, H., Sun, F., Yang, Z., Tang, S., Tao, Q., Wang, H., Lv, X., Yu, X., Siyan, Z., Xiaolu,
N., 2013. Detection rate of depression among community-dwelling older adults in China:A systematic
review and updated meta-analysis of studies in 2000-2012. Chinese Mental Health Journal 27,
805-814.

19
Fig. 1 Changes of mean score on every dimension and general health

20
Table 1 General characteristics of the participants.
Variable n %
Gender
male 189 46.0
female 222 54.0
Age
60~69 208 50.6
70~79 152 37.0
≥80 51 12.4
Education level
primary school or lower 78 19.0
middle school 276 67.2
college or higher 57 13.9
Living condition
living with spouse 309 75.2
living alone 24 5.8
living with children 78 19.0
Smoking
Yes 77 18.7
No 334 81.3
Sufficient physical exercise
Yes 193 47.0
No 218 53.0
Chronic conditions
Hypertension 198 48.2
Diabetes 123 29.9
Cataract 84 20.4
Joint disease 80 20.0
Hearing disorder 76 18.5
Dyslipidaemia 50 12.2
Coronary heart disease 50 12.2
Gastrointestinal diseases 35 8.5
Stroke 32 7.8
Kidney diseases 28 6.8
Liver diseases 23 5.6
Cancer 22 5.4
Lung diseases 18 4.4
Number of chronic conditions
0 102 24.8
21
1 77 18.7
2 86 20.9
3 64 15.6
4 44 10.7
5 26 6.3
6 10 2.4
7 2 0.5

22
Table 2 Fixed effects in the linear mixed effects models: regression coefficients (β) and P Value.

General health Body function Self-care ability Emotion Memory function Social adaptability
Variable
β(SE) P β(SE) P β(SE) P β(SE) P β(SE) P β(SE) P
Chronic
condition
multimorbidity
vs no
-0.121(0.026) <0.001 -0.168(0.038) <0.001 -0.066(0.025) 0.010 -0.124(0.043) 0.004 0.018(0.038) 0.643 -0.072(0.025) 0.004
multimorbidity
(ref.)
Gender
male vs female
0.064(0.030) 0.033 0.060(0.043) 0.162 0.039(0.029) 0.178 0.082(0.049) 0.097 0.111(0.044) 0.012 0.021(0.029) 0.463
(ref.)
Age(years) a
≥80 -0.407(0.043) <0.001 -0.353(0.062) <0.001 -0.571(0.041) <0.001 -0.368(0.071) <0.001 -0.463(0.063) <0.001 -0.316(0.041) <0.001
70~79 -0.067(0.029) 0.022 -0.101(0.042) 0.016 -0.017(0.028) 0.553 -0.049(0.047) 0.298 -0.096(0.042) 0.024 -0.008(0.028) 0.762
Education
level b
college or
0.048(0.048) 0.314 0.031(0.069) 0.654 0.036(0.046) 0.433 0.077(0.078) 0.324 0.089(0.070) 0.206 0.035(0.045) 0.442
higher
middle school 0.039(0.035) 0.267 0.034(0.050) 0.495 0.032(0.034) 0.334 0.062(0.057) 0.277 0.076(0.051) 0.137 -0.013(0.033) 0.691
Living
condition c

23
 living with
-0.036(0.034) 0.915 -0.024(0.049) 0.628 0.028(0.033) 0.393 -0.004(0.056) 0.947 0.060(0.050) 0.232 -0.060(0.032) 0.092
spouse
living alone 0.038(0.062) 0.543 0.026(0.089) 0.769 0.056(0.059) 0.343 0.040(0.101) 0.690 0.083(0.090) 0.361 0.009(0.059) 0.873
Smoking
yes vs no (ref.) -0.194(0.038) <0.001 -0.193(0.054) <0.001 -0.166(0.035) <0.001 -0.226(0.062) <0.001 -0.129(0.055) 0.021 -0.173(0.036) <0.001
Sufficient
physical
exercise
yes vs no (ref.) 0.054(0.026) 0.042 0.039(0.038) 0.303 0.019(0.025) 0.446 0.142(0.043) 0.001 0.022(0.038) 0.566 0.046(0.025) 0.079
Time -0.004(0.001) <0.001 -0.002(0.001) 0.176 -0.002(0.001) <0.001 -0.010(0.002) <0.001 0.001(0.002) 0.765 -0.004(0.001) 0.010
a
60~69 years was regarded as the reference. b primary school or lower was regarded as the reference. c living with children was regarded as the reference. P value that was
less than or equal to 0.05 presented in bold.

24
Table 3 The results of factor analyses for chronic conditions.
Factor a
Variable
1 2 3
Eigenvalue 2.052 1.976 1.381
Cumulative percent 15.782 30.983 41.603
Prevalence 14.599 4.136 33.090
Chronic condition
Cataract 0.692 0.143 0.201
Joint disease 0.680 0.026 -0.228
Hearing disorder 0.626 -0.165 0.151
Cancer 0.474 0.050 -0.257
Lung diseases 0.008 0.847 0.111
Gastrointestinal diseases 0.086 0.797 -0.045
Liver diseases -0.181 0.624 -0.062
Hypertension -0.207 -0.155 0.672
Diabetes 0.137 -0.091 0.662
Coronary heart disease 0.080 0.201 0.545
Stroke 0.146 0.005 0.295
Kidney diseases 0.240 0.019 0.278
Dyslipidaemia -0.050 -0.033 0.260
a
KMO = 0.630, the factor loading ≥0.25 have been highlighted in bold.

25
Table 4 Fixed effects in the linear mixed effects models based on multimorbidity patterns: regression coefficients (β) and P Value.
General health Body function Self-care ability Emotion Memory function Social adaptability
Variable
β(SE) P β(SE) P β(SE) P β(SE) P β(SE) P β(SE) P
Chronic
condition a
pattern 1 -0.129(0.050) 0.010 -0.157(0.074) 0.034 -0.114(0.048) 0.018 -0.057(0.083) 0.492 0.048(0.072) 0.507 -0.122(0.047) 0.010
pattern 2 -0.362(0.079) <0.001 -0.482(0.116) <0.001 -0.301(0.075) <0.001 -0.331(0.131) 0.012 -0.055(0.113) 0.628 -0.183(0.074) 0.014
pattern 3 -0.124(0.031) <0.001 -0.149(0.046) 0.001 -0.062(0.030) 0.038 -0.160(0.052) 0.002 0.040(0.045) 0.369 -0.058(0.029) 0.046
Gender
male vs
0.059(0.032) 0.098 0.058(0.048) 0.225 0.035(0.031) 0.260 0.069(0.054) 0.203 0.111(0.047) 0.018 0.008(0.031) 0.796
female (ref.)
Age(years) b
≥80 -0.412(0.049) <0.001 -0.381(0.072) <0.001 -0.604(0.046) <0.001 -0.337(0.081) <0.001 -0.481(0.070) <0.001 -0.287(0.046) <0.001
70~79 -0.065(0.031) 0.037 -0.116(0.046) 0.013 -0.006(0.030) 0.845 -0.028(0.052) 0.588 -0.128(0.045) 0.005 0.003(0.029) 0.911
Education
level c
college or
0.032(0.052) 0.541 -0.013(0.077) 0.863 0.024(0.050) 0.625 0.063(0.087) 0.467 0.058(0.075) 0.441 0.056(0.049) 0.253
higher
middle
0.039(0.038) 0.298 0.011(0.056) 0.844 0.030(0.036) 0.402 0.084(0.063) 0.178 0.057(0.054) 0.287 0.003(0.035) 0.935
school
Living
condition d
living with 0.011(0.036) 0.750 0.025(0.053) 0.646 0.014(0.034) 0.692 -0.009(0.060) 0.881 0.060(0.052) 0.249 -0.066(0.034) 0.072

26
 spouse
living alone 0.094(0.070) 0.182 0.142(0.104) 0.173 0.009(0.067) 0.896 0.110(0.117) 0.346 0.135(0.101) 0.181 -0.027(0.066) 0.679
Smoking
yes vs no
-0.210(0.040) <0.001 -0.230(0.060) <0.001 -0.163(0.038) <0.001 -0.263(0.068) <0.001 -0.163(0.059) 0.006 -0.166(0.039) <0.001
(ref.)
Sufficient
physical
exercise
yes vs no
0.070(0.028) 0.015 0.061(0.042) 0.150 0.023(0.027) 0.390 0.182(0.047) <0.001 0.051(0.041) 0.214 0.035(0.027) 0.187
(ref.)
Time -0.004(0.001) <0.001 -0.003(0.002) 0.103 -0.002(0.001) 0.002 -0.010(0.002) <0.001 -0.001(0.002) 0.741 -0.003(0.002) 0.046
a
the group without multimorbidity was regarded as the reference. b 60~69 years was regarded as the reference. c
primary school or lower was regarded as the reference. d

living with children was regarded as the reference. P value that was less than or equal to 0.05 presented in bold.

27