BACKGROUND

Depression is a major global health problem with a lifetime prevalence ranging from 1.5% to 19%.1 It is one of the main disease burdens and globally is the third leading cause of years lived with disability.2 Multimorbidity is generally defined as the co-occurrence of two or more chronic diseases.3 Accelerated ageing of the global population has largely contributed to the increased prevalence of multimorbidity.4 Among people over 50 years old in European countries, the prevalence of multimorbidity is 37%.5 A study of the prevalence of multimorbidity in Australia found that 75% of those aged 65–74 years had multimorbidity, and this proportion rose to 80% among people 75 years and older.6 Additionally, the prevalence of multimorbidity ranges from 24.1% to 83% among the population over 60 years old in South Asia7 and from 6.4% to 76.5% among the population over 60 years old in China.8

A large number of studies have proven that multimorbidity is associated with various poor health outcomes, such as poor health-related quality of life (HRQOL)9 and higher healthcare expenditures and resource utilisation.5, 10 Compared with patients who have a single disease, management of the medical needs of patients with multimorbidity increases complexity and a series of impacts on social and emotional functions should be considered.11

In the context of global ageing, people have begun to pay greater attention to the relationship between multimorbidity and depression. Previous research has documented that people with one or more chronic physical diseases have a higher proportion of depression than those with no chronic illness.12 Several meta-analyses have found that the risk of depression is twice as high among people with multimorbidity than in those without multimorbidity and three times greater for people with multimorbidity than those without any chronic physical conditions.13 In further research, it was found that the quality of life in patients with both multiple diseases and depression was significantly lower than in those with chronic diseases but who were not depressed.12 Multimorbidity together with depression is a huge public health challenge for low- and middle-income countries with limited health resources.14

China's older population is expected to reach 400 million by 2050.15 To be able to address problems caused by a large ageing population, China began to explore a long-term care insurance system in 2016.16 Previous research shows that older people with chronic diseases are at higher risk of depression,17, 18 especially those living in long-term care facilities.3 Better understanding and greater attention paid to the relationship between multimorbidity and depression are needed, to promote more targeted measures for reducing the burden of disease and to improve the health of the population.12 Therefore, research on the relationship between multimorbidity and depression is critically important. For China, such research will help improve its long-term care insurance system, to help promote a healthy ageing population. However, there is little research in this area and related studies have mostly been carried out in Western countries. In the Chinese older population, some studies on chronic diseases and depression have emerged in recent years, but most of these have focused on the relationship between a single disease or the number of diseases and depression. Few studies have examined the relationship between multimorbidity patterns and depression. Moreover, there is nearly no research among participants in long-term care insurance. The purpose of this study was to explore multimorbidity patterns among people covered by long-term care insurance in Shanghai and the relationship with depression.

METHODS Study population and data collection

This was a population-based cross-sectional study conducted among participants in a study of older people covered by long-term care insurance in Shanghai, China. Our study adopted stratified cluster sampling. The city of Shanghai was divided into 17 districts in 2015, and further divided into levels according to the city centre, suburbs, and outer suburbs. Random sampling was carried out at each level; three, two, and one sample districts were selected from the city centre, suburbs, and outer suburbs, respectively. In the six sample districts selected, one street was selected as the survey street, based on the Chinese character stroke order of the street name. We randomly selected one to three neighbourhood committees from the corresponding survey streets according to their age composition. We excluded individuals with cognitive impairment and those who were unable to participate owing to hearing, language, or mental impairment. Finally, a total of 1871 older individuals were included in the present study. This research mainly used survey questionnaires to collect information. The questionnaires were completed by specially trained investigators in face-to-face interviews. This study was approved by the Fudan University Research Ethics Committee and performed in accordance with the current institutional and state guidelines and regulations. Written informed consent was obtained from all participants.

Sociodemographic and health-related variables

Social demographic variables include gender, age (60–69, 70–79, 80–89 and ≥90 years), education level (illiterate, primary school, middle school or technical school, college or above), marital status (divorced, widowed, never married, or married), children (with or without), monthly income in Chinese yuan (CNY) (<1800, 1800–3499, 3500–3999 and ≥4000),19 and care mode (home or facility).

In this study, we also used the Perceived Social Support Scale (PSSS) to investigate social support among participants. PSSS is a validated 12-item instrument used to evaluate perceived support from three groups, namely, family, friends, and significant others. The PSSS score ranges from 12 to 84; the higher the score, the higher the perceived level of social support. The Chinese version of the PSSS has been verified and has good internal reliability.20 In this study, social support was divided into three categories; the first category comprised scores of 12–31 points, the second category was 32–50 points, and the third category was 51–84 points.21-23

Definition of chronic conditions and multimorbidity

Chronic diseases are measured based on participants' answers to the following question: ‘Have you had a chronic disease diagnosed by a doctor in the past 6 months?’ This referred to 33 chronic diseases, including hypertension, diabetes, cardiovascular disease (heart attack and other cardiovascular disorders), cerebrovascular disease (CVD), bronchitis, pneumonia, emphysema, asthma or chronic obstructive pulmonary disease, tuberculosis, cataract, glaucoma, cancer, prostatitis/prostatic hypertrophy, Parkinson disease, bedsores, injury or poisoning, rheumatoid arthritis, intervertebral disc disease, chronic low back pain, dyslipidaemia, severe vision loss, psychiatric disease, lower extremity varicose veins, gout, haemorrhoids, hypothyroidism, non-inflammatory gynaecological diseases, psoriasis, osteoporosis, chronic cholecystitis/gallstones, urinary stones, and anaemia. Multimorbidity was defined as having two or more of the 33 chronic diseases at the same time.

Assessments of depression

To assess depression, we used the 30-item Geriatric Depression Scale (GDS-30), which contains 30 questions. Respondents answer ‘yes’ or ‘no’ based on how they have felt in the past week.24 Among the questions, 20 questions receive one point for a response of ‘yes’ and zero points for ‘no’. The other 10 questions are reverse scored; the full score is 30, and an individual score ≥ 11 is considered to indicate depressive symptoms. The GDS-30 is widely used as an effective screening tool for depressive symptoms in older people,25, 26 and it is also applicable to people over age 60 years in China.27, 28

Statistical analyses

Descriptive analysis was used to describe the demographic characteristics of participants. The Chi-square test was used for categorical variables. Exploratory factor analysis is often used to identify multimorbidity patterns.29, 30 We applied the principal factor method based on a tetrachoric correlation matrix with oblimin rotation. The Kaiser–Meyer–Olkin method and Bartlett test of sphericity were performed to estimate the adequacy of the data for our model. Eigenvalues greater than one and a scree plot were used to determine the number of retained factors.31 A condition with factor loading ≥0.40 was regarded as having a strong association,32 and one condition was assigned to the pattern with larger factor loading if its factor loadings were >0.40 in more than one pattern when labelling patterns. To improve robustness, we excluded conditions with a prevalence <5.0%.29

Binary logistic regression was used to estimate the relationship between multimorbidity patterns and depressive symptoms, and to adjust for sociodemographic variables. We used IBM SPSS version 25.0 for all data analyses (IBM Corp., Armonk, NY, USA).

RESULTS General characteristics of participants

Table 1 shows the characteristics of participants with respect to depression. Among the 1871 participants, 1142 (61%) were women; most participants were 80–89 years old, accounting for 53.9%. Overall, the study population reported having a good degree of social support. Among all participants, 64.6% were considered to have depressive symptoms; those with depressive symptoms were more likely to be receiving home-based care and to have a primary school education and low social support scores. Differences concerning gender, marital status, and age between participants with and without depressive symptoms were not significant. Multimorbidity was present in 64.7% of participants.

Table 1. Characteristics of the study participants according to depression Characteristic n (%) No depression, n (%) Depression, n (%) χ2 P-value Total 1871 (100.0) 662 (35.4) 1209 (64.6) Care mode 10.093 0.001 Home 1067 (57.0) 345 (32.3) 722 (67.7) Facility 804 (43.0) 317 (39.4) 487 (60.6) Gender 0.619 0.431 Male 729 (39.0) 250 (34.3) 479 (65.7) Female 1142 (61.0) 412 (36.1) 730 (63.9) Age, years 4.997 0.172 60–69 123 (6.6) 36 (29.3) 87 (70.7) 70–79 345 (18.4) 116 (33.6) 229 (66.4) 80–89 1009 (53.9) 378 (37.5) 631 (62.5) ≥90 394 (21.1) 132 (33.5) 262 (66.5) Education levels 16.712 0.001 Illiteracy 656 (35.0) 243 (37.0) 413 (63.0) Primary school 473 (25.3) 140 (29.6) 333 (70.4) Middle school or technical school 574 (30.7) 201 (35.0) 373 (65.0) Junior college or above 168 (9.0) 78 (46.4) 90 (53.6) Marital status 1.350 0.245 Divorced, widowed, never married 1080 (57.7) 394 (36.5) 686 (63.5) Married 791 (42.3) 268 (33.9) 583 (66.1) Children 1.834 0.176 No 78 (4.2) 22 (28.2) 56 (71.8) Yes 1793 (95.8) 640 (35.7) 1153 (64.3) Income, CNY† 14.012 0.003 <1800 460 (24.6) 177 (38.5) 283 (61.5) 1800–3499 375 (20.0) 121 (32.3) 254 (67.7) 3500–3999 239 (12.8) 63 (26.4) 176 (73.6) ≥4000 797 (42.6) 301 (37.8) 496 (62.2) Social support 101.347 0.000 12–31 58 (3.1) 12 (20.7) 46 (79.3) 32–50 706 (37.7) 156 (22.1) 550 (77.9) 51–84 1107 (59.2) 494 (44.6) 613 (55.4) Multimorbidity 0.430 0.512 Without 660 (35.3) 240 (36.4) 420 (63.6) With 1211 (64.7) 422 (34.8) 789 (65.2) Chronic diseases and depression

Of the 33 somatic conditions included in our study, 10 with a prevalence of >5% were included in the analysis. Table 2 shows the prevalence of chronic disease, depression under different conditions, and the relationship between chronic diseases and depression. Hypertension showed the highest prevalence in this population. Further, only hypertension had an association with the presence of depressive symptoms when occurring without multimorbidity. Most conditions were significantly associated with depression when co-occurring with other conditions, after adjusting for all potential confounding factors.

Table 2. Associations between chronic diseases and depression (N = 1871) Chronic diseases Prevalence, n (%) Depression prevalence, n (%) Adjusted† OR (95%CI) Without multimorbidity With multimorbidity Without multimorbidity With multimorbidity No condition Without multimorbidity With multimorbidity Hypertension 106 (16.1) 911 (75.2) 57 (3.0) 614 (32.8) Reference 0.601 (0.377–0.957)* 1.516 (1.132–2.030)* Diabetes 16 (2.4) 396 (32.7) 7 (0.3) 275 (14.7) 0.513 (0.180–1.461) 1.274 (0.965–1.682) Cardiovascular disease 49 (7.4) 605 (50.0) 34 (1.8) 420 (22.4) 1.178 (0.600–2.311) 1.467 (1.135–1.897)* CVD 26 (3.9) 237 (19.6) 20 (1.1) 170 (9.1) 1.678 (0.627–4.493) 1.666 (1.191–2.331)* Cataract 6 (0.9) 151 (12.5) 4 (0.2) 79 (4.2) 1.167 (0.186–7.311) 0.451 (0.307–0.662)* Severe vision loss 0 (0.0) 105 (8.7) 0 (0.0) 63 (3.4) -‡ 0.902 (0.564-1.442) Rheumatoid arthritis 6 (0.9) 151 (12.5) 0 (0.0) 118 (6.3) -‡ 0.935 (0.659-1.326) Intervertebral disc disease 3 (0.5) 114 (9.4) 2 (0.1) 76 (4.1) 1.772 (0.143–21.962) 1.486 (0.909–2.426) Chronic low back pain 0 (0.0) 134 (11.1) 0 (0.0) 78 (4.2) -‡ 0.708 (0.455-1.102) Osteoporosis 7 (1.1) 164 (13.5) 4 (0.2) 96 (5.1) 0.680 (0.138–3.338) 0.912 (0.623–1.333) * P < 0.05 compared to no condition group. † Participants without the condition were regarded as the reference group in all models. All patterns were adjusted for care mode, age, gender, marital status, education level, personal income, children, and social support. ‡ There were no patients with depression among participants with no multimorbidity. OR, odds ratio; CVD, cerebrovascular disease. Patterns of chronic multimorbidity

Among the 10 diseases analysed, we identified three patterns of chronic multimorbidity, which could explain 45.1% of the total variance (Table 3): a musculoskeletal pattern (including rheumatoid arthritis, intervertebral disc disease, and chronic low back pain), cardiometabolic pattern (including hypertension, diabetes, cardiovascular diseases, and CVD), and degenerative diseases pattern (including cataract, severe vision loss, and osteoporosis).

Table 3. Factor loadings of the three multimorbidity patterns for each disease Chronic diseases and parameters Factor† Factor 1 Factor 2 Factor 3 Rheumatoid arthritis 0.52 0.25 0.16 Intervertebral disc disease 0.78 0.09 −0.07 Chronic low back pain 0.76 0.06 −0.15 Hypertension 0.01 0.71 −0.11 Diabetes −0.07 0.62 −0.04 Cardiovascular disease −0.01 0.55 −0.22 Cerebrovascular disease 0.14 0.45 0.26 Cataract −0.09 0.18 −0.65 Severe vision loss 0.08 0.07 −0.69 Osteoporosis 0.32 0.08 −0.52 Eigenvalue 2.12 1.33 1.06 Cumulative percentage 21.2% 34.5% 45.1% † Factor loadings indicate the strength of association between each variable and each factor, with a factor loading of ≥0.4 considered to be strong in this study (indicated in bold). Different multimorbidity patterns and depression

Table 4 shows the results of binary logistic regression between multimorbidity patterns and depression. After adjusting for potential confounding variables, the increase in factor scores in cardiometabolic and degenerative disease patterns was associated with a high risk of depressive symptoms.

Table 4. Associations between multimorbidity patterns with the presence of depressive symptoms (N = 1871) Multimorbidity patterns Unadjusted Adjusted† OR 95% CI OR 95% CI Musculoskeletal 0.964 0.876–1.061 0.992 0.896–1.099 Cardiometabolic 1.190* 1.079–1.313 1.223* 1.102–1.357 Degenerative diseases 1.157