Changes in prevalence and patterns of non-communicable disease multimorbidity across socioeconomic spectra in rural Southwest China Authors: Lan Liu, Jing-jing Huang, Guo-hui Li, Qing-rou Bao, Rui-min Shi, Fan Gu, Allison Rabkin Golden, Yuan Zhao, Le Cai Published: 2026 ABSTRACT BackgroundGlobally, the burden of non-communicable diseases (NCDs) and multimorbidity continues to grow. This study aimed to uncover the changes in the prevalence and patterns of multimorbidity involving eight common NCDs across socioeconomic gradients in rural Southwest China.MethodsData were collected through two waves of community-based, cross-sectional health interviews and physical examinations among individuals aged ≥ 35 years in rural Southwest China. Individual socioeconomic position (SEP) was constructed using principal component analysis. Association rule mining was employed to identify the NCD multimorbidity patterns of eight chronic conditions. Anthropometric measurements, blood pressure, fasting blood glucose, and post-bronchodilator spirometry were recorded for all participants. The survey’s response rates of 98.9% in 2014 and 98.2% in 2024.ResultsFrom 2014 to 2024, the age-standardized prevalence of hypertension (35.6% vs. 52.4%), diabetes (8.9% vs. 10.4%), dyslipidemia (4.4% vs. 6.8%), obesity (7.3% vs. 11.7%), CHD (3.0% vs. 5.0%), stroke (2.3% vs. 3.3%), chronic obstructive pulmonary disease (3.4% vs. 4.2%), and multimorbidity of eight NCDs (14.5% vs. 22.8%) increased significantly (P < 0.01). This increasing rate was also observed across all subgroups by sex, age, level of education, level of income, access to medical services, and SEP (P < 0.05). In both 2014 and 2024, all NCD multimorbidity patterns involved hypertension. The most prevalent NCDs combination was hypertension + obesity in 2014 and hypertension + diabetes in 2024. Over this decade, the number of identified NCDs multimorbidity patterns increased across SEP subgroups, with the most pronounced increase observed in those with high SEP.ConclusionsThe prevalence of NCDs multimorbidity rose substantially across socioeconomic gradients from 2014 to 2024, and significant socioeconomic disparities were observed in temporal trends of NCDs multimorbidity patterns in rural Southwest China. Future interventions to improve prevention and management of NCDs multimorbidity must be tailored to account for socioeconomic factors. Introduction More than 43 million people worldwide died from non-communicable diseases (NCDs) in 2021, with 73% of these deaths occurring in low- and middle-income countries (LMICs) [1]. In China, a developing country with a large population, NCDs constitute the primary disease burden [2], and this burden is on the rise due to factors such as rapid aging, socioeconomic changes, and lifestyle shifts. As the population with chronic diseases continues to grow, multimorbidity — the coexistence of two or more chronic conditions — is becoming increasingly prevalent [3, 4]. Multimorbidity is associated with poorer health outcomes, increased healthcare costs, and a greater socioeconomic burden [3, 4]. Increased multimorbidity thus poses a challenge not only for patients and their families, but also for healthcare systems, especially in resource-poor settings. In China, numerous recent studies have measured multimorbidity, but significant heterogeneity in the prevalence of NCD multimorbidity across studies ranged from 6.4% to 76.5% [5]. Moreover, limited longitudinal evidence exists, particularly in rural China; a study using data from the China Health and Nutrition Survey (CHNS) demonstrated that the prevalence of NCDs multimorbidity rose from 28.3% in 2009 to 38.9% in 2023, with a notably rapid upward trend observed in rural regions [6]. Thus, further research is necessary to provide a better understanding of the epidemiology of NCDs multimorbidity, which in turn can aid development of more targeted policies in response to this escalating health issue. Existing literature has also revealed diverse multimorbidity patterns of chronic conditions, with cardiovascular-metabolic pattern being the most consistent and reproducible [3, 7]. Although common extraction methods, such as factor analysis, latent class analysis, and cluster analysis, have been used to examine broad patterns [7, 8], they are limited to elucidating the associations between individual diseases within these clusters. Association rule mining (ARM), originally proposed by Agrawal et al. as an algorithm for marketing data analysis [9], has recently become a valuable tool for conducting exploratory data analysis in medical research. By applying ARM, the specific patterns of NCDs multimorbidity are examined to identify current priorities for multiple chronic diseases. Socioeconomic status (SES) plays a pivotal role in shaping the distribution of NCDs multimorbidity, regardless of whether SES is measured by education, income, occupation, or other indicators [3, 10]. A systematic review of 97 studies on multimorbidity revealed that individuals with lower SES were more likely to suffer from multimorbidity [7]. Furthermore, SES gradients exert differential influences on NCDs multimorbidity patterns, with lower educational attainment and income levels associated with elevated odds of cardiometabolic, musculoskeletal, and respiratory diseases [7]. Yet, limited research has been conducted to investigate the role of individual SES in temporal trends of both the prevalence and patterns of NCDs multimorbidity, including in China. Thus, this study aimed to determine temporal trends in prevalence and patterns of multimorbidity of eight common chronic conditions — hypertension, diabetes, dyslipidemia, obesity, coronary heart disease (CHD), stroke, chronic obstructive pulmonary disease (COPD), and asthma— in rural Southwest China using the ARM method, and to investigate how the changing prevalence and patterns of NCDs multimorbidity have differed across socioeconomic gradients over the past decade (2014 to 2024). These eight conditions were relatively highly prevalent NCDs in rural Southwest China, and this focused selection may leads to an underestimation of overall multimorbidity prevalence in the study population. Methods Data sources and population Data were obtained through two waves of community-based, cross-sectional health interviews and physical examinations conducted in four rural regions of Yunnan Province across a decade: 2013–2014 and 2023–2024. Yunnan, located in southwest China, is one of the least economically developed provinces in the country. Rural residents constituted 59.5% of the provincial population in 2013, and 47.1% in 2023. In 2014, all counties in Yunnan were divided into four strata (defined as low, lower-middle, upper-middle, and high) based on their economic status, as assessed by per capita gross domestic product. From each stratum, one county was randomly selected, yielding a total of four counties. A three-stage stratified random sampling method was then applied to recruit a representative sample of rural residents aged ≥ 35 years from the four selected counties. This age threshold was chosen to align with the target population for NCD management under the National Basic Public Health Services Project. These sampling methods have been detailed previously [11]. In 2024, a consistent three-stage stratified random sampling procedure was used in the same four rural counties to select survey participants. Data collection and measurement In both survey cycles, a pre-tested structured questionnaire [12] was administered to each consenting participant in a face-to-face interview by uniformly trained investigators. Information on sociodemographic characteristics (sex, age, educational level, household income, and walking time to the nearest clinic) and medical histories of the studied chronic diseases (obesity, hypertension, diabetes, dyslipidemia, CHD, stroke, COPD, and asthma) were collected. Anthropometric measurements, including height, weight, systolic and diastolic blood pressure (BP), fasting blood glucose (FBG), as well as pre- and post-bronchodilator spirometry (pre- and post-bronchodilator forced expiratory volume in one second (FEV1) and forced vital capacity (FVC)) were obtained from all participants by physicians. Data collection followed standardized protocols, detailed previously [11, 13]. Definitions Hypertension was defined as systolic BP ≥ 140 mmHg, diastolic BP ≥ 90 mmHg, and/or current use of antihypertensive medications [14]. Diabetes was defined as FBG ≥ 7.0 mmol/L (126 mg/dL), a self-reported history of diabetes, and/or current use of medication for diabetes [15]. COPD was defined by the Global Initiative for Chronic Obstructive Lung Disease (GOLD)’s criteria of a post-bronchodilator FEV1/FVC < 70% [16], and/or by a self-reported physician diagnosis of COPD. Body mass index (BMI) was calculated using the following formula: BMI = mass (kg)/squared height (m). Obesity was then defined as BMI > 25 kg/m2, based on WHO/IASO/IOTF recommendations for Asian adults [17]. Dyslipidemia, CHD, stroke, and asthma were defined based on participant self-reporting and subsequently confirmed by medical records. Multimorbidity was defined as having two or more of the eight chronic diseases in this study. Education level was classified as illiterate (defined as inability to read or write a short simple statement with understanding on everyday life) and primary (defined as completing grades 1–6) or higher. Annual household income was divided into two groups according to the median value (US$ 729 in 2014 and US$1699 in 2024): low and high. Yunnan ranks among the top three Chinese provinces with the poorest spatial accessibility to medical facilities, particularly in rural areas with limited healthcare options. SEP directly shapes residential location, which turn affects proximity to medical resources. Thus, access to medical services was appropriately included as an indicator for SEP construction in the study setting. Specifically, access to medical services was defined as good when walking time from a participant’s home to the nearest healthcare facility was less than 30 min and poor when it exceeded 30 min. Socioeconomic position (SEP) was a composite index constructed jointly across the two survey waves through principal component analysis (PCA), based on three indicators: educational level, annual household income, and access to medical services. Statistical analysis All data management and analyses were performed using R software (version 4.1.2). Categorical variables were summarized by calculating counts and percentages. Continuous variables were assessed for normality using Kolmogorov-Smirnov tests and expressed as mean ± standard deviation (SD). To compare the differences in variables between survey years, t-tests were conducted for continuous variables and chi-square tests for categorical variables. PCA was employed to develop a composite SEP index. Two-tailed P values of < 0.05 were considered statistically significant. ARM, based on the Apriori algorithm was applied to identify frequent combinations of the eight studied NCDs that exhibit strong associations. There were three commonly used measurement ratios to determine notable rules: support, confidence, and lift. For association rules, such as “{A}→{B}”, support was interpreted as the frequency of the co-occurrence of both {A} and {B} in the dataset. Confidence referred to the conditional probability of {B} occurring, given that {A} was already present. Lift was defined as the ratio of the observed support to that expected if {A} and {B} were independent; lift {A, B} = P (A∩B)/[P(A)×P(B)]. A lift value greater than 1 indicated a positive correlation between A and B. The general ARM procedure encompassed two steps: first, identifying all frequent combinations of chronic diseases that exceeded the minimum support threshold; second, generating strong association combinations from these frequent chronic disease combinations, which satisfy the minimum threshold for both confidence and lift. Considering the sample size and relevant prior literature [18], the thresholds for association rules were set as follows: support > 1%, confidence > 50%, and lift > 1. Results The number of participants aged ≥ 35 years invited to join the two surveys was 10,000 in 2014 and 10,200 in 2024. Of those invited, 9,886 in 2014 and 10,012 in 2024 provided consent to participate, yielding an overall response rate of 98.9% and 98.2%, respectively. PCA results showed a KMO value of 0.54, which is above the acceptable limit of 0.50 [19]. Bartlett’s test of sphericity was significant (P < 0.0001), indicating correlations between variables were sufficiently large for PCA. Only the first component with eigenvalues > 1, which explained 50.7% of the total data variance. This single component was therefore retained as the SEP index. The factor loadings of the three indicator variables were as follows: education level (0.64), annual household income (0.65), and access to medical services (0.59). Subsequently, the SEP index was categorized into three levels: low, medium, and high. The demographic characteristics and physical measurements of the study participants in 2014 and 2024 is shown in Table 1. The gender distribution of the participants did not differ between the two survey years (P > 0.05). The proportion of individuals aged 35–44 years, illiterate individuals, and those with poor access to medical services, low income level, and low SEP was greater in 2014 than in 2024 (P < 0.05). In addition, the mean values of all measurements, including BMI, BP, FBG, FEV1, FVC, and FEV1/FVC, were higher in 2024 than 2014 (P < 0.05). Table 1General characteristics and physical measurements of the study population by survey yearCharacteristicSurvey year2014 (n = 9886)2024 (n = 10012)Sex Male4847 (49.0)4940 (49.3) Female5039 (51.0)5070 (50.7)Age 35–44 years2306 (23.3)720 (7.2) ** 45–54 years2585 (26.1)2144 (21.4) 55–64 years2296 (23.2)2843 (28.4) 65–74 years1780 (18.0)2689 (26.9) ≥ 75 years919 (9.3)1616 (16.1)Education Illiterate1839 (18.6)1619 (16.2) ** Primary (grade 1–6) or higher8047 (81.4)8393 (83.8)Annual household income Low5336 (54.0)4729 (47.2) ** High4550 (46.0)5283 (52.8)Access to medical services Poor3062 (31.0)2819 (28.2) ** Good6824 (69.0)7193 (71.8)SEP Low2738 (27.7)2325 (23.2) ** Medium3186 (32.2)3066 (30.6) High3962 (40.1)4621 (46.2)BMI (kg/m2, mean ± SD)22.8±3.423.5±3.6 **Systolic BP (mm Hg, mean ± SD)124.6±20.1135.2±21.1 **Diastolic BP (mm Hg, mean ± SD)78.6±11.983.7±11.8 **FBG (mmol/L, mean ± SD)6.4±2.16.8±2.5 **Spirometry† (mean ± SD) FEV1, L1.6 ± 0.71.8±0.0.6 ** FVC, L2.0 ± 0.82.3±0.7 ** FEV1/FVC, %80.3±7.283.0 ± 13.2**† pre-bronchodilator; *P < 0.05, **P < 0.01SEP = socioeconomic position, BP = blood pressure, BMI = Body mass index, FBG = Fasting blood glucose, FEV1 = forced expiratory volume in one second, FVC = forced vital capacity, SD = standard deviation Table 1General characteristics and physical measurements of the study population by survey yearCharacteristicSurvey year2014 (n = 9886)2024 (n = 10012)Sex Male4847 (49.0)4940 (49.3) Female5039 (51.0)5070 (50.7)Age 35–44 years2306 (23.3)720 (7.2) ** 45–54 years2585 (26.1)2144 (21.4) 55–64 years2296 (23.2)2843 (28.4) 65–74 years1780 (18.0)2689 (26.9) ≥ 75 years919 (9.3)1616 (16.1)Education Illiterate1839 (18.6)1619 (16.2) ** Primary (grade 1–6) or higher8047 (81.4)8393 (83.8)Annual household income Low5336 (54.0)4729 (47.2) ** High4550 (46.0)5283 (52.8)Access to medical services Poor3062 (31.0)2819 (28.2) ** Good6824 (69.0)7193 (71.8)SEP Low2738 (27.7)2325 (23.2) ** Medium3186 (32.2)3066 (30.6) High3962 (40.1)4621 (46.2)BMI (kg/m2, mean ± SD)22.8±3.423.5±3.6 **Systolic BP (mm Hg, mean ± SD)124.6±20.1135.2±21.1 **Diastolic BP (mm Hg, mean ± SD)78.6±11.983.7±11.8 **FBG (mmol/L, mean ± SD)6.4±2.16.8±2.5 **Spirometry† (mean ± SD) FEV1, L1.6 ± 0.71.8±0.0.6 ** FVC, L2.0 ± 0.82.3±0.7 ** FEV1/FVC, %80.3±7.283.0 ± 13.2**† pre-bronchodilator; *P < 0.05, **P < 0.01SEP = socioeconomic position, BP = blood pressure, BMI = Body mass index, FBG = Fasting blood glucose, FEV1 = forced expiratory volume in one second, FVC = forced vital capacity, SD = standard deviation General characteristics and physical measurements of the study population by survey year † pre-bronchodilator; *P < 0.05, **P < 0.01 SEP = socioeconomic position, BP = blood pressure, BMI = Body mass index, FBG = Fasting blood glucose, FEV1 = forced expiratory volume in one second, FVC = forced vital capacity, SD = standard deviation Table 2 presents the age-adjusted prevalence of the eight studied chronic conditions in rural Yunnan, by survey year and sex. In 2024, the prevalence of the eight chronic diseases was as follows: hypertension (52.4%), diabetes (10.4%), obesity (11.7%), dyslipidemia (6.8%), CHD (5.0%), COPD (4.2%), stroke (3.3%), and asthma (0.8%). In both 2014 and 2024, men had significantly higher prevalence of COPD than women, whereas women had higher prevalence of dyslipidemia than men (P < 0.05). From 2014 to 2024, the age-standardized prevalence of hypertension, diabetes, dyslipidemia, CHD, stroke, COPD, and obesity increased significantly (P < 0.01), while the age-standardized prevalence of asthma remained statistically unchanged (P > 0.05). Table 2Age-standardized prevalence of eight chronic NCDs by sex and survey year in rural Yunnan Province, ChinaDisease20142024Malen (%)Femalen (%)Alln (%)Malen (%)Femalen (%)Alln (%)Hypertension1663 (34.3) **1854 (36.8)3519 (35.6)2618 (53.0)2632 (51.9)5246 (52.4) **Diabetes451 (9.3)428 (8.5)880 (8.9)538 (10.9)507 (10.0)1041 (10.4) **Dyslipidemia174 (3.6) **262 (5.2)435 (4.4)306 (6.2) **375 (7.4)681 (6.8) **Obesity320 (6.6)368 (7.3)722 (7.3)534 (10.8) **639 (12.6)1171 (11.7) **CHD155 (3.2)146 (2.9)297 (3.0)282 (5.7) **223 (4.4)501 (5.0) **Stroke131 (2.7) **91 (1.8)227 (2.3)168 (3.4)167 (3.3)330 (3.3) **COPD218 (4.5) **121 (2.4)336 (3.4)252 (5.1) **167 (3.3)421 (4.2) **Asthma44 (0.9)35 (0.7)79 (0.8)44 (0.9)41 (0.8)80 (0.8)*P < 0.05, **P < 0.01CHD = coronary heart disease, COPD = chronic obstructive pulmonary disease Table 2Age-standardized prevalence of eight chronic NCDs by sex and survey year in rural Yunnan Province, ChinaDisease20142024Malen (%)Femalen (%)Alln (%)Malen (%)Femalen (%)Alln (%)Hypertension1663 (34.3) **1854 (36.8)3519 (35.6)2618 (53.0)2632 (51.9)5246 (52.4) **Diabetes451 (9.3)428 (8.5)880 (8.9)538 (10.9)507 (10.0)1041 (10.4) **Dyslipidemia174 (3.6) **262 (5.2)435 (4.4)306 (6.2) **375 (7.4)681 (6.8) **Obesity320 (6.6)368 (7.3)722 (7.3)534 (10.8) **639 (12.6)1171 (11.7) **CHD155 (3.2)146 (2.9)297 (3.0)282 (5.7) **223 (4.4)501 (5.0) **Stroke131 (2.7) **91 (1.8)227 (2.3)168 (3.4)167 (3.3)330 (3.3) **COPD218 (4.5) **121 (2.4)336 (3.4)252 (5.1) **167 (3.3)421 (4.2) **Asthma44 (0.9)35 (0.7)79 (0.8)44 (0.9)41 (0.8)80 (0.8)*P < 0.05, **P < 0.01CHD = coronary heart disease, COPD = chronic obstructive pulmonary disease Age-standardized prevalence of eight chronic NCDs by sex and survey year in rural Yunnan Province, China *P < 0.05, **P < 0.01 CHD = coronary heart disease, COPD = chronic obstructive pulmonary disease Age-adjusted prevalence of multimorbidity among eight NCDs in rural Yunnan, stratified by survey year and socioeconomic status, is summarized in Table 3. The prevalence of NCDs multimorbidity increased significantly from 14.5% in 2014 to 22.8% in 2024 (P < 0.01). This upward trend was consistently observed across all subgroups, by sex, age, level of education, level of income, access to medical services, and SEP (P < 0.05). In addition, the increase in the prevalence of NCDs multimorbidity was most pronounced among illiterate individuals (absolute change: +10.8%), followed by those with high SEP (+ 9.9%) and high household income level (+ 9.5%). Advancing age and better access to medical services were both associated with higher prevalence of NCDs multimorbidity in both 2014 and 2024 (P < 0.05). Moreover, in 2024, prevalence of NCDs multimorbidity decreased as levels of education increased, whereas a higher prevalence of NCDs multimorbidity was observed among those with higher income level and SEP (P < 0.05). Table 3Age-standardized prevalence of multimorbidity of eight chronic NCDs by survey year and socioeconomic status in rural Yunnan Province, ChinaCharacteristic20142024Number of NCDsn (%)Number of NCDsn (%)01≥ 201≥ 2Sex Male2656 (54.8)1493 (30.8)698 (14.4)1778 (36.0)2016 (40.8)1146 (23.2) Female2696 (53.5)1607 (31.9)736 (14.6)1907 (37.6)2029 (40.0)1136 (22.4)Age (years) 35–441840 (79.8)362 (15.7)104 (4.5) **426 (59.2)206 (28.6)88 (12.2) ** 45–541644 (63.6)680 (26.3)261 (10.1)984 (45.9)806 (37.6)354 (16.5) 55–641182 (51.5)767 (33.4)347 (15.1)978 (34.4)1177 (41.5)685 (24.1) 65–74698 (39.2)694 (39.0)388 (21.8)680 (25.3)1170 (43.5)839 (31.2) ≥ 75344 (37.4)383 (41.7)192 (20.9)304 (18.8)832 (51.5)480 (29.7)Education Illiterate923 (50.2)640 (34.8)276 (15.0)460 (28.4)742 (45.8)418 (25.8) ** Primary (grade 1–6) or higher4410 (54.8)2470 (30.6)1175 (14.6)3240 (38.6)3298 (39.3)1855 (22.1)Annual household income Low2860 (53.6)1724 (32.3)752 (14.1)1778 (37.6)1991 (42.0)965 (20.4) ** High2484 (54.6)1379 (30.3)687 (15.1)1928 (36.5)2045 (38.7)1310 (24.8)Access to medical services Poor1825 (59.6)842 (27.5)395 (12.9) **1139 (40.4)1122 (39.8)558 (19.8) ** Good3528 (51.6)2259 (33.1)1044 (15.3)2554 (35.5)2913 (40.5)1726 (24.0)SEP Low1539 (56.2)849 (31.0)350 (12.8)858 (36.9)993 (42.7)474 (20.4) ** Medium1571 (49.3)1096 (34.4)519 (16.3)1119 (36.5)1272 (41.5)675 (22.0) High2223 (56.1)1177 (29.4)563 (14.2)1742 (37.7)1765 (38.2)1114 (24.1)All5348 (54.1)3104 (31.4)1434 (14.5)3694 (36.9)4035 (40.3)2283 (22.8)*P < 0.05, **P < 0.01NCDs = non-communicable diseases, SEP = socioeconomic position, CHD = coronary heart disease, COPD = chronic obstructive pulmonary disease Table 3Age-standardized prevalence of multimorbidity of eight chronic NCDs by survey year and socioeconomic status in rural Yunnan Province, ChinaCharacteristic20142024Number of NCDsn (%)Number of NCDsn (%)01≥ 201≥ 2Sex Male2656 (54.8)1493 (30.8)698 (14.4)1778 (36.0)2016 (40.8)1146 (23.2) Female2696 (53.5)1607 (31.9)736 (14.6)1907 (37.6)2029 (40.0)1136 (22.4)Age (years) 35–441840 (79.8)362 (15.7)104 (4.5) **426 (59.2)206 (28.6)88 (12.2) ** 45–541644 (63.6)680 (26.3)261 (10.1)984 (45.9)806 (37.6)354 (16.5) 55–641182 (51.5)767 (33.4)347 (15.1)978 (34.4)1177 (41.5)685 (24.1) 65–74698 (39.2)694 (39.0)388 (21.8)680 (25.3)1170 (43.5)839 (31.2) ≥ 75344 (37.4)383 (41.7)192 (20.9)304 (18.8)832 (51.5)480 (29.7)Education Illiterate923 (50.2)640 (34.8)276 (15.0)460 (28.4)742 (45.8)418 (25.8) ** Primary (grade 1–6) or higher4410 (54.8)2470 (30.6)1175 (14.6)3240 (38.6)3298 (39.3)1855 (22.1)Annual household income Low2860 (53.6)1724 (32.3)752 (14.1)1778 (37.6)1991 (42.0)965 (20.4) ** High2484 (54.6)1379 (30.3)687 (15.1)1928 (36.5)2045 (38.7)1310 (24.8)Access to medical services Poor1825 (59.6)842 (27.5)395 (12.9) **1139 (40.4)1122 (39.8)558 (19.8) ** Good3528 (51.6)2259 (33.1)1044 (15.3)2554 (35.5)2913 (40.5)1726 (24.0)SEP Low1539 (56.2)849 (31.0)350 (12.8)858 (36.9)993 (42.7)474 (20.4) ** Medium1571 (49.3)1096 (34.4)519 (16.3)1119 (36.5)1272 (41.5)675 (22.0) High2223 (56.1)1177 (29.4)563 (14.2)1742 (37.7)1765 (38.2)1114 (24.1)All5348 (54.1)3104 (31.4)1434 (14.5)3694 (36.9)4035 (40.3)2283 (22.8)*P < 0.05, **P < 0.01NCDs = non-communicable diseases, SEP = socioeconomic position, CHD = coronary heart disease, COPD = chronic obstructive pulmonary disease Age-standardized prevalence of multimorbidity of eight chronic NCDs by survey year and socioeconomic status in rural Yunnan Province, China *P < 0.05, **P < 0.01 NCDs = non-communicable diseases, SEP = socioeconomic position, CHD = coronary heart disease, COPD = chronic obstructive pulmonary disease Table 4 displays the results of association rules analysis conducted by survey year, which are the strong association rules discovered according to the established thresholds criteria. ARM yielded 5 association rules in 2014 and 10 rules in 2024, with all NCD combinations involving hypertension. These five association rules identified in 2014 persisted in 2024, and the prevalence of these shared chronic disease combinations increased between the two survey years. Compared to 2014, two new binary NCDs multimorbidity patterns (hypertension combined with diabetes and COPD) and three ternary NCDs multimorbidity patterns (hypertension + diabetes + CHD, hypertension + diabetes + dyslipidemia, and hypertension + dyslipidemia + obesity) emerged in 2024. In 2014, the most common combination of the eight studied chronic diseases was hypertension + obesity, whereas in 2024 hypertension + diabetes was the most prevalent combination. In both 2014 and 2024, hypertension + stroke had the highest confidence and lift in binary NCDs multimorbidity patterns. Table 4Multimorbidity patterns of eight chronic NCDs by survey year in rural Yunnan Province, ChinaSurvey YearRulesSupport (%)Confidence (%)LiftCount2014hypertension + stroke1.275.42.749hypertension + CHD1.362.72.252hypertension + dyslipidemia2.457.22.095hypertension + obesity4.250.21.8166hypertension + diabetes + obesity1.162.52.2452024hypertension + stroke3.180.41.4311hypertension + CHD4.272.91.3424hypertension + dyslipidemia5.272.41.3518hypertension + obesity7.669.91.3764hypertension + diabetes8.374.01.3832hypertension + COPD2.757.31.1274hypertension + diabetes + dyslipidemia1.883.51.5177hypertension + diabetes + CHD1.181.61.5115hypertension + diabetes + obesity1.683.01.5156hypertension + dyslipidemia + obesity1.278.61.4121NCDs = non-communicable diseases, CHD = coronary heart disease, COPD = chronic obstructive pulmonary disease. All identified association rules include hypertension Table 4Multimorbidity patterns of eight chronic NCDs by survey year in rural Yunnan Province, ChinaSurvey YearRulesSupport (%)Confidence (%)LiftCount2014hypertension + stroke1.275.42.749hypertension + CHD1.362.72.252hypertension + dyslipidemia2.457.22.095hypertension + obesity4.250.21.8166hypertension + diabetes + obesity1.162.52.2452024hypertension + stroke3.180.41.4311hypertension + CHD4.272.91.3424hypertension + dyslipidemia5.272.41.3518hypertension + obesity7.669.91.3764hypertension + diabetes8.374.01.3832hypertension + COPD2.757.31.1274hypertension + diabetes + dyslipidemia1.883.51.5177hypertension + diabetes + CHD1.181.61.5115hypertension + diabetes + obesity1.683.01.5156hypertension + dyslipidemia + obesity1.278.61.4121NCDs = non-communicable diseases, CHD = coronary heart disease, COPD = chronic obstructive pulmonary disease. All identified association rules include hypertension Multimorbidity patterns of eight chronic NCDs by survey year in rural Yunnan Province, China NCDs = non-communicable diseases, CHD = coronary heart disease, COPD = chronic obstructive pulmonary disease. All identified association rules include hypertension Table 5 outlines the multimorbidity patterns of the eight studied NCDs by survey year and SEP. In 2014, 6, 6, and 5 association rules were selected for participants with low, medium, and high SEP, respectively. In 2024, 8, 8, and 10 association rules were selected for participants with low, medium, and high SEP, respectively. The number of identified association rules increased across SEP subgroups, with the largest increase observed in those with high SEP. In both 2014 and 2024, the most frequent NCD multimorbidity pattern among individuals with low and medium SEP was hypertension + diabetes, whereas hypertension + obesity was the most common pattern among those with high SEP. There were four NCDs multimorbidity patterns across all SEP subgroups both in 2014 and 2024: hypertension combined with stroke, CHD, dyslipidemia, and obesity. Table 5Multimorbidity patterns of eight chronic NCDs by survey year and SEP in rural Yunnan Province, ChinaSurvey yearSEPRulesSupport (%)Confidence (%)LiftCount2014Lowhypertension + stroke1.270.22.233hypertension + CHD1.870.02.249hypertension + diabetes3.766.22.0102hypertension + obesity3.158.11.886hypertension + dyslipidemia1.853.81.749hypertension + COPD1.551.91.640Mediumhypertension + CHD2.371.61.973hypertension + dyslipidemia3.165.81.898hypertension + stroke1.664.61.751hypertension + diabetes5.458.71.6172hypertension + obesity4.156.91.5132hypertension + diabetes + obesity1.168.01.834Highhypertension + stroke1.275.42.749hypertension + CHD1.362.72.252hypertension + dyslipidemia2.457.22.095hypertension + obesity4.250.21.8166hypertension + diabetes + obesity1.162.52.2452024Lowhypertension + stroke4.083.61.492hypertension + dyslipidemia3.975.21.391hypertension + obesity6.275.11.3145hypertension + diabetes7.877.71.3181hypertension + CHD4.368.71.2101hypertension + COPD3.063.11.170hypertension + diabetes + dyslipidemia1.489.21.533hypertension + diabetes + obesity1.375.61.331Mediumhypertension + stroke2.977.91.488hypertension + diabetes8.076.91.4246hypertension + CHD4.072.91.3124hypertension + dyslipidemia5.376.31.3164hypertension + obesity6.468.31.2196hypertension + diabetes + CHD1.2901.636hypertension + diabetes + obesity1.389.11.641hypertension + diabetes + dyslipidemia1.787.91.551Highhypertension + stroke2.879.91.5131hypertension + CHD4.375.11.4199hypertension + dyslipidemia5.769.41.3263hypertension + obesity9.269.01.3423hypertension + diabetes8.870.81.3405hypertension + diabetes + obesity1.883.21.684hypertension + dyslipidemia + CHD1.178.51.551hypertension + diabetes + CHD1.379.71.559hypertension + dyslipidemia + obesity1.577.51.569hypertension + diabetes + dyslipidemia2.079.51.593NCDs = non-communicable diseases, SEP = socioeconomic position, CHD coronary heart disease, COPD = chronic obstructive pulmonary disease. All identified association rules include hypertension, and rules are not mutually exclusive across different SEP groups Table 5Multimorbidity patterns of eight chronic NCDs by survey year and SEP in rural Yunnan Province, ChinaSurvey yearSEPRulesSupport (%)Confidence (%)LiftCount2014Lowhypertension + stroke1.270.22.233hypertension + CHD1.870.02.249hypertension + diabetes3.766.22.0102hypertension + obesity3.158.11.886hypertension + dyslipidemia1.853.81.749hypertension + COPD1.551.91.640Mediumhypertension + CHD2.371.61.973hypertension + dyslipidemia3.165.81.898hypertension + stroke1.664.61.751hypertension + diabetes5.458.71.6172hypertension + obesity4.156.91.5132hypertension + diabetes + obesity1.168.01.834Highhypertension + stroke1.275.42.749hypertension + CHD1.362.72.252hypertension + dyslipidemia2.457.22.095hypertension + obesity4.250.21.8166hypertension + diabetes + obesity1.162.52.2452024Lowhypertension + stroke4.083.61.492hypertension + dyslipidemia3.975.21.391hypertension + obesity6.275.11.3145hypertension + diabetes7.877.71.3181hypertension + CHD4.368.71.2101hypertension + COPD3.063.11.170hypertension + diabetes + dyslipidemia1.489.21.533hypertension + diabetes + obesity1.375.61.331Mediumhypertension + stroke2.977.91.488hypertension + diabetes8.076.91.4246hypertension + CHD4.072.91.3124hypertension + dyslipidemia5.376.31.3164hypertension + obesity6.468.31.2196hypertension + diabetes + CHD1.2901.636hypertension + diabetes + obesity1.389.11.641hypertension + diabetes + dyslipidemia1.787.91.551Highhypertension + stroke2.879.91.5131hypertension + CHD4.375.11.4199hypertension + dyslipidemia5.769.41.3263hypertension + obesity9.269.01.3423hypertension + diabetes8.870.81.3405hypertension + diabetes + obesity1.883.21.684hypertension + dyslipidemia + CHD1.178.51.551hypertension + diabetes + CHD1.379.71.559hypertension + dyslipidemia + obesity1.577.51.569hypertension + diabetes + dyslipidemia2.079.51.593NCDs = non-communicable diseases, SEP = socioeconomic position, CHD coronary heart disease, COPD = chronic obstructive pulmonary disease. All identified association rules include hypertension, and rules are not mutually exclusive across different SEP groups Multimorbidity patterns of eight chronic NCDs by survey year and SEP in rural Yunnan Province, China NCDs = non-communicable diseases, SEP = socioeconomic position, CHD coronary heart disease, COPD = chronic obstructive pulmonary disease. All identified association rules include hypertension, and rules are not mutually exclusive across different SEP groups Discussion The findings indicated a substantial increase in the prevalence of multimorbidity of the eight studied chronic diseases across all socioeconomic gradients in rural Southwest China between 2014 and 2024. Further, this study revealed that all identified NCDs multimorbidity patterns were associated with hypertension, and that significant socioeconomic disparities in temporal shifts of these patterns. Among the chronic diseases investigated, hypertension (52.4%) was the most prevalent, followed by diabetes (10.4%). The overall prevalence of hypertension and diabetes in the study also exceeded previously reported estimates in other Chinese regions and other Asian countries [20–23], suggesting hypertension and diabetes remain a major public health problem among adults residing in rural communities. Notably, hypertension, diabetes, dyslipidemia, obesity, CHD, stroke, and COPD increased significantly over the ten-year study period, a finding in line with other epidemiological studies [22, 24]. In contrast, the prevalence of asthma remained stable over the past decade, and the complex causes (environmental, medical, etc.) underlying this stability require further investigation. Collectively, the growing epidemic of these NCDs highlights mounting pressure on the healthcare system in rural Southwest China, given its relatively limited and unevenly distributed medical resources. Similar to findings in prior studies [24, 25], our study uncovered sex disparities in the prevalence and temporal trends of common chronic conditions. Specifically, both in 2014 and 2024, the prevalence of COPD was higher in males than in females, whereas dyslipidemia was more prevalent among females compared to males. In addition, over the study period, men exhibited a greater increase in the prevalence of hypertension, diabetes, dyslipidemia, and CHD than women, while women showed a relatively higher increase in prevalence of obesity, stroke, and COPD than men. These gender differences likely relate to complex interactions among biological (genetic and hormonal), environmental, and behavioral factors [26]. The results suggest gender specific interventions to mitigate chronic disease risk should be considered in future NCD prevention and management efforts. Our study revealed that the prevalence of NCDs multimorbidity in the study population in 2024 was lower than the rate observed in rural Chinese populations (58.3%, involving 14 chronic conditions) [27] and in Canadian populations (33%, involving 14 chronic conditions) [28], but greater than the rate observed in Brazil (18.3%, involving 9 chronic conditions) [29] as well as in Bangladesh (8.4%, involving 6 chronic conditions) [30]. This notable discrepancy in NCD multimorbidity prevalence maybe attributed to variations in geography or in differing methodologies (for example, differing study population, types, and number of NCDs included) between studies [3]. Moreover, prevalence of NCDs multimorbidity significantly increased from 2014 to 2024, aligning with prior trends observed in China [22] as well as in other countries, including the United States [31]. In this way, our findings underscore the urgent need for more effective prevention strategies to curb the epidemic of NCDs multimorbidity in rural Chinese communities. The present study uncovered a considerable increase in the prevalence of NCDs multimorbidity among both sexes over the ten-year study period. However, no statistically significant difference was observed in the distribution of NCD multimorbidity between females and males, which was not consistent with previous studies [31–33]. However, NCDs multimorbidity prevalence was positively correlated with advancing age, a result in accordance with previous research globally [3, 34], as health naturally deteriorates with age [26]. Age-stratified analyses further revealed a rising prevalence of multimorbidity across all age groups, with older adults experiencing a greater increase compared to younger individuals, suggesting attention should be directed to addressing NCDs multimorbidity in rural elderly populations. The present study also indicated that both education level and access to medical services had significant impacts on NCDs multimorbidity. Despite the fact that prevalence of NCDs multimorbidity has increased substantially across subgroups stratified by education level and access to medical services from 2014 to 2024, participants with lower education level and better access to medical services had larger relative increases than their counterparts. This inverse association between educational level and prevalence of NCDs multimorbidity has been reported in many prior studies in China [35] as well as in other countries [32]. Individuals with lower educational attainment tend to have lower health literacy and are less likely to engage in healthy behaviors [36]. Further, the discrepancy in the association between access to medical services and NCDs multimorbidity may be explained by the fact that those with better access to healthcare tend to be more knowledgeable about their chronic health conditions. In this study, NCDs multimorbidity was more prevalent among individuals with higher income and SEP. This finding is broadly consistent with previous studies conducted in LMICs [34]. However, it contrasts with research from high-income countries, where lower income and SEP are associated with an increased risk of multimorbidity [37]. Two possible factors may explain these differences. First, amidst rural China’s rapid socioeconomic development and epidemiological transition, unhealthy behaviors, such as high-calorie diets, physical inactivity, and heavy drinking, may initially emerge among more affluent or higher SEP groups [38]. Second, individuals with higher SEP typically have better access to health screenings and medical care than those of lower SEP, leading to higher rates of disease detection and longer survival with multimorbidity, which may have contributed to the higher prevalence observed among higher SEP groups. Furthermore, SEP and income were additionally correlated with temporal trends of NCDs multimorbidity in the survey regions: across all income and SEP groups, NCDs multimorbidity levels increased, with higher-income and higher-SEP participants experiencing larger increases over the ten-year study period than their counterparts. The findings thereby highlight the need for tailored interventions that address the unique risk factors for NCDs multimorbidity among populations with high income and SEP. Our results demonstrated that all combinations of the studied NCDs involved hypertension in the two survey years. This pattern of hypertension combined with other NCDs has previously been noted, both in China and internationally [35, 39]. While specific NCDs multimorbidity patterns, including hypertension combined with stroke, CHD, dyslipidemia, and obesity, as well as the triad combination of hypertension + diabetes + obesity, remained consistent between 2014 and 2024, the prevalence of these shared NCDs combinations increased significantly over the ten-year study period, with the largest increase observed in the dyad of hypertension + obesity. Moreover, several new NCDs multimorbidity patterns associated with diabetes and dyslipidemia emerged in 2024, such as hypertension + diabetes + CHD and hypertension + dyslipidemia + obesity. This may partly result from the escalating rate of diabetes and dyslipidemia in the surveyed communities. The findings thereby suggest that integrated management strategies for hypertension-related multimorbidity should be prioritized, with specific emphasis placed on controlling the combinations of hypertension with metabolic disorders (for example, diabetes, dyslipidemia, or obesity) and other cardiovascular diseases to alleviate the development of complex multimorbidity. In alignment with prior studies [7, 40], our results also showed that individual SEP was significantly associated with patterns of NCDs multimorbidity. In both 2014 and 2024, hypertension and diabetes was the most common multimorbidity pattern among participants with low and medium SEP, whereas hypertension combined with obesity was the most prevalent combination among high SEP individuals. Furthermore, although all NCDs multimorbidity patterns identified across socioeconomic strata in 2014 continued to be validated in 2024, high SEP participants had the largest increase in the number of NCDs multimorbidity over the ten-year study period, with multiple ternary multimorbidity patterns (for example, hypertension + dyslipidemia + coronary heart disease and hypertension + dyslipidemia + obesity) emerging in 2024, resulting in more complex multimorbidity patterns compared to other SEP groups. Thus, prevention and management measures for NCDs multimorbidity in rural Southwest China should be tailored to address socioeconomic disparities in multimorbidity patterns. The current study was limited in several respects. First, our assessment of multimorbidity was limited to eight prevalent chronic NCDs (primarily cardiometabolic and respiratory conditions); excluding other common chronic conditions (e.g., cancer, arthritis, chronic kidney disease), which may likely leads to underestimation of overall multimorbidity prevalence and restrict the generalization of our findings to the overall NCDs multimorbidity burdens. Second, the diagnosis of diabetes was based on FBG and/or pharmacological treatment of diabetes, and did not include a glycated haemoglobin (HbA1c) or oral glucose tolerance test; this may have led to an underestimation of diabetes in the participant population. Third, data on certain chronic conditions was collected through self-reporting, including dyslipidemia, CHD, stroke, and asthma, an approach which may underestimate the prevalence of these diseases in the participant population due to underdiagnosis. Fourth, the present results were based on a random sampling of four counties, which limits their generalizability to the broader province and region. Finally, despite age-standardization to minimize bias from population structure shifts, residual confounding related to cross-sectional comparability between the two survey waves cannot be completely excluded, which should be considered when interpreting the temporal trends of NCD multimorbidity. Conclusion In conclusion, this study demonstrates chronic NCDs and their multimorbidity are an increasingly large public health challenge in rural Southwest China, with prevalence of multiple chronic diseases and NCDs multimorbidity rising substantially in the decade from 2014 to 2024. Further, significant socioeconomic disparities were observed in temporal trends of both the prevalence and patterns of NCDs multimorbidity. Future interventions to improve prevention and control of NCDs multimorbidity should be specifically tailored to distinct socioeconomic subgroups: targeted behavior modification interventions for high SEP groups to mitigate unhealthy behaviors; and structured health literacy promotion programs for low-education populations to enhance their NCD self-management.