Introduction

Multimorbidity, defined as the co-occurrence of multiple chronic health conditions, is emerging as a major public health concern, particularly in high-income countries [1]. However, surprisingly high prevalence rates of multimorbidity are also observed in low- and middle-income countries [2,3]. In Iran, the prevalence of multimorbidity is estimated to be 21.1%, with higher rates among older individuals and females [4]. Multimorbidity is associated with adverse health outcomes, including impaired physical and mental function, reduced quality of life, frailty, and increased mortality risk [5–8]. These effects may result from the synergistic accumulation of diseases rather than their individual impacts [9].

The coexistence of two or more chronic health conditions in an individual necessitates complex and structured medical care and increases healthcare utilization and financial burden [10,11]. Therefore, current single-disease-oriented public health and healthcare strategies are deemed inadequate, reflecting the need for a comprehensive understanding of multimorbidity patterns, and integrated treatment approaches [12]. To establish effective health policies and treatment plans for multimorbidity, it is essential to identify multimorbidity patterns in populations [13].

Cardio-metabolic-related disorders, such as hypertension, dyslipidemia, type two diabetes mellitus (T2DM), and stroke, which share pathophysiological pathways are commonly observed in multimorbid individuals [12,14,15]. Multimorbidity is a complex phenomenon that requires statistical approaches to group populations into subgroups with similar combinations of chronic diseases. Latent class analysis (LCA), a subset of structural equation modeling, is a useful method for identifying subgroups of individuals with common characteristics [16]. Using health-related data, this probabilistic modeling algorithm can explore the latent variables behind observed categorical variables that are statistically associated [17], to identify common combinations of chronic diseases in a population and help develop targeted interventions for specific subgroups of patients with multimorbidity [12]. Several studies have utilized LCA to identify multimorbidity patterns and associated factors [12,18–21]. For example, a study in South Africa identified four latent classes among the multimorbid population, with women comprising 70% of the population, and found that multimorbidity occurred across all age groups [20].

Noncommunicable diseases have a disproportionate distribution in societies, hence identifying subgroups using LCA in populations with shared characteristics can reflect the different healthcare priorities and needs of these societies [21]. Unlike previous studies that primarily assessed the effects of risk factors on individual chronic diseases or multimorbidity, this research employs LCA to uncover hidden patterns within chronic diseases. Given the potential of LCA in identifying disease patterns and associated factors, as well as the limited use of LCA for multimorbidity analysis in Iran, this study aimed to identify multimorbidity patterns and investigate their relationship with various risk factors using baseline cohort data from southern Iran.

Materials and methods

Results

A total of 10,131 participants were included in the study, with a mean age of 48.64 ± 9.58 years, and the majority were women (54.8%). The flow diagram of the included participants is depicted in Fig 1. Among the participants, 54.2% were literate, 50.5% were employed, and approximately 88% were married. The prevalence of multimorbidity was 40.3%, with multimorbid individuals having a mean of 2.8 chronic conditions. Dyslipidemia (45.1%), hypertension (20.1%), and T2DM (19.7%) were the most prevalent chronic conditions. Table 1 summarizes detailed descriptive information about the study participants.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Descriptive information of the FACS participants.

https://doi.org/10.1371/journal.pone.0335177.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Flow diagram of the included participants.

(A) Between October 2014 and September 2016, a total of 11,097 people aged 35 to 70 years from 24 villages in the Sheshdeh rural area were evaluated. (B) Out of these, 10,622 qualified, and 10,131 were enrolled (participation rate: 95.4%). (C) All participants who were recruited were incorporated into this study.

https://doi.org/10.1371/journal.pone.0335177.g001

LCA was performed with models ranging from 2 to 6 classes. The model selection criteria, including G², AIC, and BIC, were used to determine the best-fitting model. Due to the magnitude of the G² reference statistics and the high number of estimated parameters, the distribution of this index was unknown; therefore, the role of AIC and BIC in selecting the model was more pronounced – the best model is the one with the lowest values. The three-class model had the lowest BIC value, while the four-class model had the lowest AIC value. However, for better interpretability, the three-class model was selected as the best fit. The performance indicators of the models are shown in table 2.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Model selection statistics from LCA models with different latent classes.

https://doi.org/10.1371/journal.pone.0335177.t002

The probability of item responses for multimorbid conditions across each multimorbidity cluster is shown in Fig 2. Participants were assigned to one of three classes based on the highest membership probability. The healthy class had a prevalence of 66.8% and low probabilities of all studied chronic diseases. The dyslipidemia class, with a prevalence of 14.1%, had a high probability of dyslipidemia (77.9%). The cardio-metabolic class, with a prevalence of 19.1%, had high probabilities of hypertension (67.7%), T2DM (43.7%), and dyslipidemia (50.9%).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Item response probability for multimorbid conditions across each multimorbidity cluster.

(A) Class 1 (Healthy; prevalence = 0.668), Class 2 (Dyslipidemia; prevalence = 0.141), and Class 3 (Cardio-metabolic conditions; prevalence = 0.191) notes: ¹ The probability of a “No” response can be calculated by subtracting the item-response probabilities from 1. ² Item-response probabilities >0.5 are colored light orange to facilitate interpretation.

https://doi.org/10.1371/journal.pone.0335177.g002

The cardio-metabolic class had the highest mean age (55.36 ± 8.75 years), with 53.7% of men and 46.3% of women classified as healthy. Women were significantly more likely than men to belong to the dyslipidemia and cardio-metabolic classes (P < 0.001). Illiteracy, low socioeconomic status, unemployment, and unmarried status were significantly higher among cardio-metabolic class participants compared to the other two classes (P < 0.001 for all). Additionally, more than 59% of healthy class participants slept 8 hours or more per day, while cardio-metabolic class participants had significantly lower daily sleep hours and physical activity index (P < 0.001). The mean number of co-occurring diseases was significantly higher in the cardio-metabolic and dyslipidemia classes compared to the healthy class (P < 0.001) (Table 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Comparisons of demographic and clinical data among participants by latent class membership.

https://doi.org/10.1371/journal.pone.0335177.t003

A multinomial logistic regression model was developed, adjusting for age, sex, education level, occupation, marital status, socioeconomic status, physical activity, and sleep duration. Results showed that with every year increase in age, the odds of belonging to dyslipidemia (odds ratio (OR) = 1.04, 95% CI: 1.03–1.05, P < 0.001) and cardio-metabolic conditions (OR = 1.10, 95% CI: 1.09–1.11, P < 0.001) classes significantly increased compared to the healthy class. Women had significantly higher odds than men of belonging to dyslipidemia (OR = 2.49, 95% CI: 2.05–3.02, P < 0.001) and cardio-metabolic (OR = 3.35, 95% CI: 2.79–4.03, P < 0.001) classes. Eemployed status was a protective factor against belonging to the cardio-metabolic conditions class (OR = 0.66, 95% CI: 0.55–0.80, P < 0.001). Conversely, very high socioeconomic status increased the odds of belonging to both dyslipidemia (OR =1.44, 95% CI: 1.16–1.78, P = 0.001) and cardio-metabolic conditions (OR = 1.35, 95% CI: 1.10–1.65, P = 0.003) classes. Additionally, sleeping for 8 hours or more per day (OR = 0.74, 95% CI: 0.63–0.87, P < 0.001) was associated with reduced odds of membership in the cardio-metabolic conditions class. Moreover, every unit increase in MET decreased the odds of membership in dyslipidemia classes (OR = 0.98, 95% CI: 0.97–0.98, P < 0.001) and cardio-metabolic conditions (OR = 0.97, 95% CI: 0.96–0.98, P < 0.001). Furthermore, the high dietary intake group had higher odds of belonging to the dyslipidemia class (OR = 1.57, 95% CI: 1.16–2.11, P = 0.003) compared to the healthy class (Table 4).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Summary findings of multinomial logistic regression of association between multimorbidity and studied risk factors.

https://doi.org/10.1371/journal.pone.0335177.t004

Discussion

The primary aim of this study was to identify distinct patterns of multimorbidity in the adult population of southern Iran, with an emphasis on understanding how multiple chronic conditions co-occur, rather than examining each condition separately. Using latent class analysis (LCA), we explored the clustering of chronic diseases within individuals and investigated the socio-demographic and lifestyle factors associated with these patterns. The present study identified three distinct classes: a relatively healthy class, a dyslipidemia class, and a cardio-metabolic disorders class encompassing hypertension, T2DM, and dyslipidemia. This pattern-based approach provides a more integrated understanding of multimorbidity, highlighting disease combinations that may share common etiological pathways and require targeted prevention and management strategies. Comparing our findings with previous studies is challenging due to variations in study environments, definitions, disease spectra, demographic factors, data collection methods, and participants’ baseline health statuses [28–31]. Nonetheless, we attempted to compare our identified disease patterns with those reported in prior literature.

Our study estimated the prevalence of multimorbidity in the general population of southern Iran to be 40.3%, with LCA identifying rates of 14.1% for dyslipidemia and 19% for cardiometabolic classes. For comparison, a meta-analysis of 37 studies reported global multimorbidity prevalence ranging from 3.5% to 70%. Another systematic review involving 15.4 million adults found an overall prevalence of 37.2%, with notable regional variations: South America (45.7%), North America (43.1%), Europe (39.2%), and Asia (35%). Our findings align with similar studies conducted in Asia but are slightly higher than those reported for Europe and North America. These results highlight significant regional and demographic differences in the prevalence of multimorbidity. It is important to note that although most studies applied a similar general definition of multimorbidity (i.e., the coexistence of two or more chronic conditions), the number and type of conditions considered varied substantially across studies, ranging from a few to several dozen diseases. Despite this heterogeneity, hypertension, diabetes, dyslipidemia, and stroke were among the most frequently included conditions. These are largely consistent with the disease spectrum considered in our study, which enhances the comparability of our findings. However, differences in disease lists and data sources across studies may partly explain the variability observed in multimorbidity prevalence estimates. This variability underscores the importance of conducting region-specific investigations, such as ours, to better understand local patterns and inform targeted healthcare interventions [32,33]. Additionally, our study identified a high likelihood of co-occurrence among dyslipidemia, hypertension, and T2DM, consistent with findings from other studies. For instance, Fortin et al. reported hypertension, hyperlipidemia, and rheumatological diseases as common disorders in multimorbidity, while another study found hypertension, hyperlipidemia, and T2DM among 45 chronic diseases [34,35].

The regression analysis demonstrated that older age and female sex were strongly associated with belonging to multimorbidity classes. This finding is consistent with previous research and may be explained by age-related accumulation of chronic conditions and women’s longer life expectancy, biological differences, and greater healthcare utilization. Interestingly, very high socioeconomic status was also associated with increased multimorbidity risk, which may reflect lifestyle-related factors such as dietary habits, stress, and sedentary behavior observed in affluent populations. Conversely, protective factors such as employment, higher physical activity, and adequate sleep suggest that social engagement, healthy lifestyles, and restorative behaviors play an important role in mitigating multimorbidity. These results emphasize the need for integrated prevention strategies focusing not only on biological determinants but also on modifiable behavioral and social factors [12,36,37].

Women had a significantly higher likelihood of belonging to dyslipidemia and cardio-metabolic classes compared to men, aligning with other studies indicating a higher prevalence of multimorbidity among women [19,38,39]. In contrast to our study, Petty et al. found no association between sex and the higher risk of multimorbidity [40]. This gender disparity in multimorbidity prevalence may be attributed to factors such as longer lifespan, biological differences, and socio-cultural factors [38,41].

Our study also showed an association between unemployment and increased odds of belonging to cardio-metabolic classes. Although information on this association is scarce, some evidence might be consistent with ours. That is, people with low levels of education may have limited health literacy, low socioeconomic status, lower chance of employment, and reduced access to healthcare resources; therefore, they may have poor management in preventing and controlling disease risk factors. A study of Danish adults found that low education greatly increased the likelihood of belonging to class 6 (Complex Cardio-metabolic Disorders) and class 7 (Complex Respiratory Disorders) of multimorbidity [19]. Other studies also have shown that a low level of education is consistently associated with an increased risk of multimorbidity [42–44].

Longer sleep duration (>8 hours per day) was inversely associated with membership in the cardio-metabolic class. This finding is consistent with studies linking short sleep duration to increased mortality risk in individuals with chronic diseases such as T2DM, hypertension, cardiovascular disease, and obesity [45,46]. In a large cohort study conducted in the United States, a U-shaped relationship was reported between sleep duration and all-cause and cardiovascular mortality compared to normal sleep; that is, sleeping less than 5 hours and more than 9 hours increased total mortality by 16% and 11%, respectively [46]. These findings might highlight the importance of adequate sleep-in disease prevention and management.

Additionally, higher levels of physical activity, measured in metabolic equivalent of task (MET) units, were associated with a reduced likelihood of membership in lipid disorders and cardiometabolic classes, consistent with findings from previous studies [47]. However, a study by Autenrieth et al. observed no link between physical activity and multimorbidity in women, while Hudon et al. found no such relationship in either sex after adjusting for age, education, income, and employment. Both studies highlighted limitations, including reliance on self-reported data for physical activity and chronic diseases, as well as potential recall bias [48,49]. Beyond these methodological limitations, variations in the type and number of comorbidities examined may also account for these discrepancies. Nonetheless, with the aging global population and the rising prevalence of multimorbidity, our findings underscore the importance of physical activity as a cornerstone of prevention and health promotion efforts [50,51].

Moreover, our results also showed the association between dietary intake and multimorbidity, particularly with dyslipidemia. Medium or high intake levels were found to increase the odds of belonging to the dyslipidemia class. The importance of diet in multimorbidity prevention has been widely recognized. For example, a study of 35,372 participants in the UK cohort found that higher daily energy intake was associated with an 8% increased risk of multimorbidity (HR: 1.08 (95% CI: 1.01, 1.16)) [52]. Therefore, it might represent the significance of promoting healthy dietary patterns as a strategy for mitigating multimorbidity risks.

In addition, the association between very high socioeconomic status and increased multimorbidity risk underscores the need to explore lifestyle factors and stressors in affluent populations. These findings emphasize the importance of a holistic, multi-level approach to reducing the burden of multimorbidity and improving the quality of life for vulnerable groups.

The findings of this study had significant public health implications. The high odds of multimorbidity among women and older adults represent the need for targeted interventions. For women, preventive care and chronic disease management programs are essential, as they are at higher risk for multiple conditions. For older adults, integrating multimorbidity management into primary care systems is critical to improve health outcomes. Moreover, the association between very high socioeconomic status and increased multimorbidity risk underscores the need to explore lifestyle factors and stressors in affluent populations. Additionally, employed participants showed decreased odds of experiencing cardio-metabolic conditions, and higher physical activity along with adequate sleep were identified as protective factors. Thus, promoting physical activity, healthy sleep habits, and addressing the risk factors associated with socioeconomic disparities are essential components of an effective public health strategy. Finally, our results indicate that medium or high dietary intake increased the odds of belonging to the dyslipidemia class, emphasizing the importance of nutrition interventions in reducing the burden of chronic diseases. These findings suggest that public health policies should focus on both preventive measures and the management of multimorbidity in vulnerable groups.

A notable strength of this study is the use of LCA, which allowed us to identify distinct subgroups of multimorbidity based on 11 chronic conditions. However, the study has several limitations. First, the use of multinomial logistic regression with class membership as the outcome variable, derived from probabilistic LCA, might not fully account for the inherent uncertainty in class assignment, which represents a methodological consideration. Longitudinal studies are recommended in the future, particularly when follow-up data from the FACS become available. Second, despite performing quality control at FACS, reliance on self-reported baseline data introduces the possibility of reporting bias, particularly recall errors, which warrants caution when interpreting the results. Third, the findings of this study are based on a cohort of adults aged 35 years and older residing in a rural district of Fasa, Iran. This specific geographic focus limits the generalizability of our results to other regions, particularly urban areas or populations with different socio-economic or cultural characteristics. Furthermore, the demographic and health profiles of the Fasa cohort may not fully represent the broader Iranian population or other middle-income countries. Therefore, the external validity of the study findings may be restricted. Fourth, other unmeasured chronic diseases or risk factors were not included in the analysis, which could introduce residual confounding. Finally, the available data does not allow us to assess the impact of multimorbidity on long-term outcomes such as mortality, which could provide further insights into the burden of chronic diseases.

Conclusion

This study aimed to identify multimorbidity patterns and their associated risk factors among adults in southern Iran using latent class analysis. We identified three distinct classes of multimorbidity: a relatively healthy group, a dyslipidemia group, and a cardio-metabolic group, with an overall multimorbidity prevalence of 40.3%. Our findings indicate that older age, female sex, very high socioeconomic status, and higher dietary intake were significantly associated with increased risk of belonging to multimorbidity classes, while employment, greater physical activity, and adequate sleep were protective factors. These results highlight the heterogeneous nature of multimorbidity and emphasize the role of both demographic and lifestyle determinants in shaping disease patterns. Recognizing such patterns is crucial for developing targeted prevention and management strategies, particularly for vulnerable subgroups such as women and older adults. Future longitudinal research using cohort follow-up data will be essential to confirm these associations, explore causal pathways, and guide the design of comprehensive, population-specific interventions to reduce the burden of multimorbidity.

Acknowledgments

The authors express their gratitude to all personnel at the Non-communicable Diseases Research Center (NCDRC) and Clinical Research Development Unit of Vali Asr Hospital of Fasa University of Medical Sciences.