Multimorbidity Analysis According to Sex and Age towards Cardiovascular Diseases of Adults in Northeast China

Non-communicable diseases (NCDs) are great challenges in public health, where cardiovascular diseases (CVD) accounted for the large part of mortality that caused by NCDs. Multimorbidity is very common in NCDs especially in CVD, thus multimorbidity could make NCDs worse and bring heavy economic burden. This study aimed to explore the multimorbidity among adults, especially the important role of CVD that played in the entire multimorbidity networks. A total of 21435 participants aged 18–79 years old were recruited in Jilin province in 2012. Weighted networks were adopted to present the complex relationships of multimorbidity, and Charlson Comorbidity Index (CCI) was used to evaluate the burden of multimorbidity. The prevalence of CVD was 14.97%, where the prevalence in females was higher than that in males (P < 0.001), and the prevalences of CVD increased with age (from 2.22% to 38.38%). The prevalence of multimorbidity with CVD was 96.17%, and CVD could worsen the burden of multimorbidity. Multimorbidity and multimorbidity with CVD were more marked in females than those in males. And the prevalence of multimorbidity was the highest in the middle-age, while the prevalence of multimorbidity with CVD was the highest in the old population.


Results
Sex-specific and age-specific distributions of top 20 NCDs of participants. Table 1 presents the top 20 NCDs with the highest prevalences, and the ranks of these diseases were a little different by sex and age. The prevalences of majority diseases were different by sex and age as well (P < 0.05). The prevalence of CVD was 14.49% (ranked the third), where in females it was higher than that in males, and the prevalences of CVD increased with age (from 2.22% to 38.38%).
Sex-specific and age-specific top 5 patterns of multimorbidity. Table 2 shows the sex-specific and age-specific prevalences of top 5 patterns of multimorbidity, with the pair hyperlipidemia & hypertension as the highest one. Generally, the prevalences of CVD & hyperlipidemia and CVD & hypertension were also very high, especially in the old population.
Sex-specific and age-specific top 5 patterns of multimorbidity with CVD. Multimorbidity was extremely common among CVD patients, where there were 96.17% (3086/3209) CVD patients suffered from at least one other NCDs. Further, the prevalence of multimorbidity with CVD in females (97.29%) was more marked than that in males (94.31%), and it was the worst among old (97.54%) CVD patients (P < 0.001, 85.81% for the young and 96.42% for the middle-age). The top 5 patterns of sex-specific and age-specific multimorbidity with CVD were shown in Table 3. In general, the ranks for multimorbidity with CVD were similar, where hyperlipidemia and hypertension were the most frequent occurrences of multimorbidity among CVD patients, thus the "CVD-Hyperlipidemia-Hypertension" (CVD-H-H) triangle was inclined to play an important role in the multimorbidity networks.
Evaluation of multimorbidity and multimorbidity networks.  showed the networks of the multimorbidity in the whole population, as well as the sex-specific and age-specific populations, and Table 4 list the indices which could measure the features of the networks. The network density and the average degree of females were larger than those of males, thus the network of females was much denser than that of males, i.e., the NCDs in females tended to co-occur more frequently than those in males. Meanwhile, the network density (as well as the average degree) reached the largest in the middle-age, and smallest in the young. In Table 4, each average degree of the CVD-H-H triangle was extremely higher than that of its network, where the average degree

Rank Disease Total
Gender of the triangle in old population was 6.9 (22.67/3.27) times of its own network ( Fig. 3(c)), which was the largest. Meanwhile, the proportion of CVD that contributed to the CVD-H-H triangle in old population was also the largest (11/(11 + 12 + 11) = 32.35%). The average degree of the CVD-H-H triangle in females was more marked than that in males, but compared with the corresponding female/male network, the CVD-H-H triangle in males was more important, which was 6.6 (24.67/3.74) times of its network ( Fig. 2(b)). However, the proportion of CVD that contributed to the CVD-H-H triangle in males (8/(8 + 14 + 15) = 21.62%) was smaller than that in females (14/(14 + 21 + 13) = 29.17%). Finally, the severity of the multimorbidity using Charlson Comorbidity Index (CCI) was also shown in Table 4. It was no surprising that the CCI in the population with CVD was extremely larger than that without CVD in all groups (P < 0.001), which indicated that CVD would bring extra burden to multimorbidity. Further, the CCIs in males were smaller than those in females (P = 0.013 for CVD and P = 0.002 for non-CVD), and CCIs were the highest among the elderly, and the lowest among the young (all P < 0.001).

Discussion
NCDs are believed to bring great challenges to and have important impacts on public health nowadays, which accounted for 63% of deaths worldwide in 2008, and CVD is one of the most important main causes of deaths 24 . Meanwhile, multimorbidity was extremely common among the CVD patients 7,25 . In this study, we investigated the multimorbidity of 57 kinds of NCDs based on 21435 adults in Jilin province in 2012, especially the multimorbidity with CVD. Hyperlipidemia, hypertension and CVD were top 3 NCDs with the highest prevalences in Jilin province. Multimorbidity and the multimorbidity with CVD were more marked in females than those in males. The prevalence of multimorbidity was the highest in middle-age, whereas the prevalence of multimorbidity with  CVD was the highest in the old population. 96.17% CVD patients suffered from multimorbidity, where the prevalence of multimorbidity increased with age, and CVD would worsen the burden of multimorbidity. Hyperlipidemia, hypertension and CVD were top 3 NCDs with the highest prevalences in Jilin province, regardless of sex and age. The prevalence of CVD was 14.97%, which was lower than other studies in literature 5,26 , due to that hypertension was not included in CVD in this study. Although CVD ranked the third, the analysis of CCI suggested that CVD would bring extra burden to multimorbidity and increase the 10-year mortality 27 , thus the lethality and the burden of CVD with multimorbidity was much higher. Besides, among the top 5 patterns of multimorbidity there were 2 patterns of multimorbidity with CVD, which suggested that multimorbidity with CVD were very common. And 96.17% CVD patients suffered from multimorbidity, which was higher than other studies 22,23 , one possible reason might be that hyperlipidemia was investigated in our study.
Further, the CVD-H-H triangle in males was more marked than that in females, relative to their own network, but CVD in males contributed less proportions to the triangle than that in females. It was suggested that hyperlipidemia and hypertension in males played more important roles in multimorbidity, while CVD and its multimorbidity were more common in females, and would bring more risk to females 28,29 . Therefore, different strategies should be developed to prevent NCDs and their multimorbidity in males and females separately.
Finally, the prevalences of majority multimorbidity were the highest in the eldly, and the lowest in the young, which were consistent with other studies 30,31 . The possible reason might be that body immunity and function declined with age, so that the old people were more vulnerable to NCDs and their multimorbidity 32 . Although the middle-age had a denser multimorbidity network, the CVD-H-H triangle in the old population played a more important role, relative to their own network, where there CVD occupied large percentage compared with that of the young and middle-age. Thus it suggested different key prevention towards different age groups: multimorbidity with CVD were tended to cluster in the old population, while nutritional or metabolic diseases were common for young people 33 .
Some limitations should be noted here. Firstly, the participants in the study were selected in Jilin province, which could not represent the (CVD) multimorbidity in other places. Secondly, the disease situations were mainly based on self-report, which might cause bias. Thirdly, only cerebrovascular disorders, angina pectoris, coronary disease and myocardial infarction were involved in CVD, which might underestimate the prevalence of CVD and its multimorbidity. Finally, only sex and age were investigated in the study, but other factors that might have effects on the multimorbidity were worthy of further study.  Data collection and measurement. The data of this study included demographics, anthropometric measurements (e.g., height, weight, blood pressure) and NCDs situations (57 NCDs, including liver cancer, lung cancer, gastric cancer, colorectal cancer, breast cancer, cervical cancer, prostate cancer, thyroid carcinoma, leukemia and other tumor (except the above 9 ones); anemia, rheumatic and other hematologic and immune related   Table 4. Evaluation of multimorbidity and multimorbidity networks. a The people with age less than 40 (age ≤ 40) were viewed as young, and people with 41 ≤ age ≤ 65 and age ≥ 66were middle-age and old, respectively. b CVD-H-H refers to the "CVD-Hyperlipidemia-Hypertension" triangle in the network, and the values in the brackets are degrees of CVD, hyperlipidemia and hypertension.

Methods
diseases (except the above 2); obesity, diabetes, hyperlipidemia, thyrotoxicosis, osteoporosis, gout and other endocrine, nutritional and metabolic diseases (except the above 6 ones); schizophrenia, depression and other mental & behavioral disorders (except the above 2); cognition disorders, epilepsy, Parkinson's disease and other neurological diseases (except the above 3 ones); cataract, glaucoma and other eye diseases (except the above 2);  ≤ 65), and (c) for old (age ≥ 66), "Prostate" represented "Prostate hyperplasia or inflammation", "Gynecological" represented gynecological inflammation, "Gastric ulcer" represented gastric ulcer or duodenal ulcer; "other tumor" represented tumors except 9 cancer like liver cancer, lung cancer, etc., and "other digestive" represented other diseases of digestive system except 7 ones like fatty liver, cirrhosis, etc. hypertension, CVD (including cerebrovascular disorders, angina pectoris, coronary heart disease and myocardial infarction), corpulmonale, varicose veins of lower extremity and other diseases of circulatory system (except the above 4 ones); chronic obstructive pulmonary emphysema, asthma, nasopharyngitis, chronic bronchitis, and other respiratory diseases (except the above 4 ones); gastric ulcer or duodenal ulcer, fatty liver, cirrhosis, cholecystitis, gallstone, gastroenteritis, hernia of abdominal cavity, and other diseases of digestive system (except the above 7 ones); rheumatoid arthritis, disc disease, and other musculoskeletal and connective tissue diseases(except the above 2); nephritis, gynecological inflammation, breast diseases, urinary calculus, prostate hyperplasia or inflammation and other diseases of genitourinary system (except the above 5 ones)). After 12 or more hours of overnight fasting, finger-tip blood samples were taken from the subjects, and the plasma glucose (FPG) level was analyzed; the 2-hour FPG level was also tested. These tests were conducted by a Bayer Bai Ankang fingertip blood glucose monitor machine. The serum lipids, including total cholesterol (TC), triglycerides (TG), high-density lipoprotein (HDL-C) and low-density lipoprotein (LDL-C), were measured before breakfast, using enzymatic methods in a central laboratory with standardized testing. Weight and height were performed after removing shoes and heavy outer clothing. Weights were measured to the nearest 0.1 kg using a calibrated scale with the subjects standing in an upright position, and heights were measured to the nearest 0.1 cm using a standard anthropometer. Body mass index (BMI) was calculated as weight/height 2 (kg/m 2 ). Blood pressure was measured using mercury sphygmomanometer in the sitting position after a 5-min rest period by trained professionals. Two readings each of systolic blood pressure (SBP) and diastolic blood pressure (DBP) were recorded, and the average of each measurement was used for data analysis. If the first two measurements differed by more than 5 mmHg, additional readings were taken 34-36 . Assessment criteria of disease. Hypertension was referred to those with SBP ≥ 140 mm Hg, and/or DBP ≥ 90 mm Hg, and/or normotensives treated with antihypertensive medications, and/or a self-reported history of hypertension 37 . Hyperlipidemia was defined as TC ≥ 5.18 mmol/L, and/or LDL-C ≥ 3.37 mmol/L, and/or HDL-C < 1.04 mmol/L, and/or TG ≥ 1.70 mmol/L, and/or normolipidemic subjects treated with antihyperlipidemia medications, and/or with history of hyperlipidemia diseases 38 . Obesity was defined as the BMI ≥ 28 kg/ m 2 39 . Diabetes was defined as FPG ≥ 7.0 mmol/l (126 mg/dl) and/or a self-reported history of diabetes. CVD was defined as a participant carried at least one of the following disease: cerebrovascular disorders, angina pectoris, coronary heart disease and myocardial infarction. Other NCDs were judged only by self-reported history of diseases which diagnosed in hospitals on county level and above.
Statistical analysis. The continuous variables were expressed as means ± standard deviations (SD) and compared using the t test or Wilcoxon rank-sum test. The categorical variables were expressed as counts or percentages and compared using the Rao-Scott-χ 2 test. All statistical analyses were performed with R version 3.4.1 (University of Auckland, Oakland, New Zealand). Statistical significance was set at a P value < 0.05.
In this study, weighted networks were applied to study the relationships among multimorbidity. The nodes of the network represented the diseases, and the height of each node was proportional to the prevalence of each disease. The edge in the network represented the co-occurrence of a multimorbidity pair, and the weight of the edge was proportional to the prevalence of each multimorbidity pair. When a participant carried more than 2 diseases, the count of every multimorbidity pair would have an increment of 1 (e.g., when a participant carried CVD, hypertension and hyperlipidemia, then the multimorbidity pair CVD & hypertension, hypertension & hyperlipidemia and hyperlipidemia & CVD would have an increment of 1). The prevalence of disease or multimorbidity pair were calculated as the total counts of participants which carried the disease or multimorbidity pair divided by the corresponding sample size. The relationships of the multimorbidity with prevalence higher than 1% were list in the networks in our study.
Degree was adopted to measure the centrality of a disease (e.g. CVD), where degree was the number of nodes that a focal node was connected to, which measured the involvement of the node in the network. Network density and average degree were used to evaluate the sparsity of a network. The network density of an undirected graph with N nodes and M edges was defined as 2 M/N(N − 1), which described the portion of the potential connections (N(N − 1)/2) in a network that were actual connections (M). The average degree was defined as the average of degrees of all nodes. The larger the network density (or average degree), the denser the network 40,41 . CCI was used to measure the burden of multimorbidity or comorbidity, which had been validated in many clinical settings to describe the condition of comorbidity and multimorbidity 42,43 . The larger the CCI, the worse the condition of multimorbidity (the larger 10-year mortality).