About half of older adults have two or more chronic conditions at the same time: a systematic review and meta-analysis

Abstract

Background

Population aging has led to a rising prevalence of multimorbidity among older adults, posing a significant global public health challenge. This study aims to comprehensively integrate the evidence regarding the prevalence of multimorbidity among the older adults and the influencing factors through a systematic review and meta-analysis.

Methods

We conducted a systematic review and meta-analysis in accordance with Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. The research question was structured using the PICOS framework: Population: adults aged ≥60 or ≥65; Intervention/Phenomenon of Interest: multimorbidity (≥2 chronic conditions); Comparator: not applicable (prevalence synthesis) or various demographic groups (risk factor analysis); Outcomes: prevalence of multimorbidity and pooled odds ratios (OR) of risk factors; Study design: cross-sectional studies. We conducted a comprehensive search of multiple databases up to July 2025 to identify cross-sectional studies that reported the prevalence and/or risk factors of multimorbidity (defined as ≥2 chronic conditions) in adults aged ≥60 or ≥65. The pooled prevalence was estimated using a random-effects model, and heterogeneity was explored through subgroup analysis and meta-regression.

Results

Forty-nine studies involving 729,043 older adults were included. The global pooled prevalence of multimorbidity was 46.0% (95% CI: 38.0%−55.0%), with substantial heterogeneity (I² = 99.98%). Meta-regression identified survey location as a key source of heterogeneity. Subgroup analyses revealed a higher prevalence in females (48%) than males (41%) and a lower prevalence in studies from China (40%) compared to other countries (53%). Significant risk factors included age ≥ 70 years (OR = 1.43), female sex (OR = 1.59), BMI ≥ 28 kg/m² (OR = 1.97), low education (OR = 1.44), single status (OR = 1.32), and low income (OR = 1.46).

Conclusion

Multimorbidity is highly prevalent among the global older adults, with significant geographical and demographic variations. Modifiable risk factors such as obesity, low education, and low income are crucial targets for intervention to alleviate the burden of multimorbidity.

Introduction

Global population aging has become an important social challenge in the 21st century (1). According to statistics from the World Health Organization (WHO), the global proportion of people aged 60 and over has increased significantly over the past few decades, rising from 9.2% in 2000 to 12.3% in 2020; this age group is expected to account for 20% of the global population by 2050 (2). The trend of population aging is not only obvious in developed countries, such as Japan, Germany and Italy, but also gradually showing this feature in developing countries, such as China and India (3), with the latter accounting for over 60% of the global older adults population by 2030 (4).

Population aging directly increases the pressure on public health systems, as the risk of chronic diseases rises significantly with age. Multimorbidity—defined as the coexistence of two or more chronic conditions (5)—is particularly prevalent in the older adults, driven by physiological changes (e.g., immunosenescence, inflammaging), unhealthy lifestyles, and long-term accumulation of disease risk (6). A systematic review has shown that the prevalence of multimorbidity in adults aged ≥ 65 years exceeds 70% in high-income countries (7), and this coexistence of chronic diseases not only impairs patients' quality of life (e.g., reduced mobility, increased disability) but also raises medical costs—patients with multimorbidity have 2–5 times higher healthcare expenditures than those with a single chronic disease, imposing heavy burdens on individuals, families, and societies (8).

In view of the high prevalence rate of multimorbidity in the global older adults population and the serious health burden brought by it, more and more researchers have begun to pay attention to the phenomenon of multimorbidity in the older adults and conducted a large number of related studies (9). Although previous meta-analyses have documented the prevalence of multimorbidity in older adults, significant research gaps remain. First, most early reviews only reported prevalence as a broad range (e.g., 30%−70%) without calculating a pooled estimate (10), which limits cross-study quantitative comparison and fails to provide a global reference value (11). Second, many studies focused solely on prevalence and neglected quantitative analysis of risk factors—for example, only describing “age and gender are associated with multimorbidity” but not reporting pooled odds ratios (OR) or 95% confidence intervals (CI) (12). Third, there is substantial heterogeneity in the definition of multimorbidity across studies: some defined it as ≥2 chronic conditions, others as ≥3, and a few even included acute diseases, leading to inconsistent prevalence estimates that are difficult to synthesize (13). Fourth, existing reviews often have geographical or language restrictions: most focused on high-income countries in Europe and North America, and English-language publications dominated, overlooking a large body of evidence from low- and middle-income countries (e.g., China, India, Vietnam) and non-English literature, which may introduce selection bias (14). Finally, some studies restricted the age range to ≥65 years (15), excluding data on adults aged 60–64 years—a group with increasing multimorbidity risk in developing countries—resulting in incomplete population coverage. Consequently, a comprehensive, up-to-date meta-analysis that synthesizes global prevalence data from both English and Chinese literature and concurrently investigates key risk factors is urgently needed.

The primary aim of this study is to fill the above gaps by synthesizing global evidence on multimorbidity in older adults. To this end, we will conduct a systematic synthesis and meta-analysis of cross-sectional studies to determine the global prevalence of multimorbidity in older adults and its key influencing factors. Ultimately, we aim to generate robust scientific evidence to support the formulation of targeted interventions and health policies, contributing to the global response to the challenges of aging and multimorbidity.

Materials and methods

Registration

This study was prepared in strict accordance with the PRISMA declaration specification. To elaborate on the specific reporting content of this study, the items in the PRISMA 2020 checklist are listed in Supplementary Table 1. In addition, detailed information on the study protocol was made available online through the Prospective Register for Systematic Reviews (https://www.crd.york.ac.uk/PROSPERO/view/CRD42024602631, PROSPERO: CRD42024602631).

Research question and PICOS framework

The research question was explicitly defined according to the PICOS framework:

Population: community-dwelling or institutionalized older adults, defined as individuals aged 60 years and over or 65 years and over.

Intervention/Phenomenon of Interest: The presence of multimorbidity, defined as the coexistence of two or more chronic non-communicable diseases.

Comparator: for the prevalence synthesis, no comparator was applicable. For the analysis of influencing factors, comparisons were made between different demographic and socioeconomic groups (e.g., male vs. female, high vs. low education).

Outcomes: the primary outcome was the prevalence of multimorbidity. Secondary outcomes were the pooled effect sizes (Odds Ratios, OR) of factors associated with multimorbidity.

Study design: observational cross-sectional studies specifically designed to assess prevalence.

Search strategy

A comprehensive systematic search was performed in PubMed, Web of Science, Embase, Cochrane Library, CNKI, WANFANG, VIP, and CBM from their inception to July 30, 2025. The inclusion of the four major Chinese databases (CNKI, WANFANG, VIP, and CBM) was essential to capture the extensive body of relevant literature published in Chinese on multimorbidity among the older adults in China, which is a key population of interest in this global analysis and is often underrepresented in international reviews that rely solely on English-language databases (16). The search strategy was developed and optimized iteratively based on the PICO (Population, Phenomenon of Interest, Study Design) framework with the assistance of an experienced medical librarian to ensure comprehensiveness and precision. To balance sensitivity and specificity, we employed a combination of Medical Subject Headings (MeSH/Entree) and free-text terms for the key concepts: (1) Population (P): terms related to older adults (e.g., “Aged”, “Elderly”, “Older adults”, “Older people”, “Geriatric”); (2) phenomenon of Interest (I): terms related to multimorbidity (e.g., “Multimorbidity”, “Comorbidity”, “Multiple chronic conditions”, “Multiple long-term conditions”, “Multiple morbidity”); (3) study design (O): terms related to study design and outcome (e.g., “Cross-Sectional Studies”, “Prevalence”, “Epidemiology”). Within each conceptual group, synonyms and related terms were combined using the Boolean operator “OR”. The three conceptual groups were then combined using the Boolean operator “AND” to form the final search strategy. The detailed search strategies for each database are provided in Supplementary Table 2. Furthermore, to ensure comprehensiveness, we supplemented the database searches with a literature tracking approach by manually reviewing the reference lists of all included articles to identify any additional studies that met our eligibility criteria.

Inclusion and exclusion criteria

Studies were included in this meta-analysis if they met all of the following criteria: (1) Population: the study population consisted of older adults, defined as individuals aged 60 years and over or 65 years and over. (2) Concept: this study specifically examined the phenomenon of “coexistence of multiple diseases”, which refers to the situation where an individual suffers from two or more chronic non-communicable diseases simultaneously (such as hypertension, diabetes, COPD, etc.). (3) Study design: the study employed a cross-sectional design (To ensure the comparability and quality of prevalence data, this meta-analysis exclusively included cross-sectional studies that were specifically designed to assess prevalence. While baseline data from longitudinal studies were considered, they were ultimately excluded because their primary objective is typically not to report prevalence, which can lead to inconsistent data availability and potential bias in baseline reporting). (4) Outcomes: the study reported at least two of the following outcomes: the prevalence of multimorbidity, or provided sufficient data for its calculation. Risk factors for multimorbidity, assessed using logistic regression and reported as Odds Ratios (OR) with corresponding 95% Confidence Intervals (CI). Studies were excluded based on the following criteria: (1) Unclear definitions or data: studies with an ambiguous definition of multimorbidity or unclear descriptions of the specific chronic conditions assessed. (2) Inappropriate study design: studies that were not cross-sectional (e.g., cohort studies, case-control studies, randomized controlled trials). (3) Publication type: non-original research articles, such as case reports, reviews, editorials, commentaries, conference abstracts, or letters to the editor. (4) Language: articles published in languages other than English or Chinese. (5) Data accessibility: studies for which the full text was unavailable, or from which essential data could not be extracted or calculated. To prevent data duplication from overlapping populations, if multiple studies were identified from the same region or cohort, we included only the study with the largest sample size or the most recently published data.

Study selection and data extraction

All literature in the databases was imported into Endnote software for literature selection and management. First, duplicate publications were removed. Subsequently, two researchers independently conducted a preliminary screening of the titles and abstracts according to the set inclusion and exclusion criteria to exclude irrelevant references. Finally, the researchers read the full text and screened again to determine the final literature to be included. If differences of opinion arose between the two researchers during this process, these differences were resolved through negotiation or consultation with a third researcher. After determining the selected research list, two independent researchers conducted two rounds of data extraction for each included study to minimize errors. Data extraction was carried out using a standardized Microsoft Excel spreadsheet, which included: first author, publication year, survey time, survey location, study design, sample size, sample gender ratio, prevalence of chronic diseases, prevalence of multimorbidity, and relevant influencing factors. The relevant influencing factors include: age, gender, marital status, body mass index, educational level, smoking status, drinking status, exercise habits, income level, and place of residence. For studies that reported the number of individuals with two or more chronic conditions but did not explicitly report the prevalence percentage, the prevalence was calculated during data extraction using the formula: prevalence (%) = (number of individuals with multimorbidity/total study sample size) × 100%. This calculated value was then used in the meta-analysis.

Methodological quality assessment

Two researchers independently assessed the methodological quality of the final included literature. They systematically scored the included studies using the cross-sectional study evaluation criteria recommended by the Agency for Healthcare Research and Quality (AHRQ) (17). The AHRQ methodological checklist is widely recognized as an effective tool for assessing the quality of cross-sectional studies and is currently accepted and used by researchers in multiple fields (18). The checklist can be found at http://www.ncbi.nlm.nih.gov/books/NBK35156/. The tool contains 11 assessment items relating to data sources, inclusion and exclusion criteria, timing and order of inclusion of participants, influence of evaluator subjective factors, assessment of quality assurance, interpretation of data exclusions, control for confounding variables, management of missing data, integrity of data collection, and follow-up. Each assessment item is graded with “Yes”, “No” or “Unclear”, where “Yes” was awarded 1 point, and the remaining options were not awarded points. Depending on the score, a score of 0–3 indicated “low” methodological quality, a score of 4–7 indicated “medium” methodological quality, and a score of 8–11 indicates “high” methodological quality. In addition, the process of quality evaluation was independently conducted by two researchers, and the consistency of their scores was evaluated by Cohen's Kappa coefficient (19). For literature with inconsistent evaluation results, the researchers resolved these differences by discussing them with a third investigator.

Subgroup analysis and meta-regression

To investigate potential sources of the significant heterogeneity observed among the included studies, we performed both subgroup analyses and meta-regression. These analyses were pre-specified to explore the influence of key study-level characteristics on the pooled prevalence of multimorbidity. The moderators examined included age, gender, study quality (high vs. medium), survey region, survey year, and sample size. We aimed to assess whether these factors significantly influenced the overall pooled estimate.

Data analysis and synthesis

In this study, Stata 18.0 software was used to conduct a meta-analysis of the included literatures. To ensure the consistency of the analysis, all the included studies met the “multimorbidity definition” criteria. We extracted the prevalence of multimorbidity among the older adults from the included studies. When the prevalence of multimorbidity was not explicitly reported, it was calculated using the formula: prevalence = (number of individuals with ≥ 2 chronic conditions)/(total number of participants) × 100%, following the method used in previous meta-analyses (20). Pooled results were estimated using a random-effects model for the prevalence of multimorbidity and reported using weighted point estimates based on the prevalence of multimorbidity and 95% confidence intervals (95% CI). The heterogeneity among the included studies was assessed using the Chi-squared (χ²) test, with a significance level of P < 0.10. The magnitude of heterogeneity was quantified by the I² statistic. If I² ≤ 50% and P > 0.10, the heterogeneity among studies was considered small, and the fixed-effect model was adopted. If I² > 50% and P ≤ 0.10, it indicated moderate or high heterogeneity among studies, and a random-effects model was used (21). We performed a sensitivity analysis to test the robustness of the results by excluding one study at a time and observing changes in the pooled prevalence in the remaining studies. For the pooled effect size of the influencing factors, the combined OR value and 95% CI of the influencing factors were obtained by comparing the fixed-effect model and the random-effect model. If the results of the two models are similar, it can be inferred that the results of this study have good reliability and stability. In addition, the funnel plot and Egger's test were used to evaluate potential publication bias. A result of P > 0.05 was considered to indicate a low risk of publication bias, while P < 0.05 indicated statistically significant asymmetry.

Results

Search results

The literature screening and selection process is illustrated in the PRISMA flow diagram (Figure 1). Our initial search across all databases yielded a total of 29,142 records. After removing 6,413 duplicates, 22,729 unique articles remained for screening.

Figure 1.

PRISMA flow diagram of literature search and selection.

In the first stage of screening, we reviewed titles and abstracts, which led to the exclusion of 18,649 records irrelevant to the research topic and 3,678 records of an inappropriate study type (e.g., reviews, case reports, editorials). This process left 402 articles for full-text eligibility assessment.

Upon detailed full-text review, a further 306 articles were excluded for the following reasons: non-representative sample or setting (n = 148), inappropriate study design (n = 79), data unavailable (n = 49), use of previously published data (n = 27), and unclear research objectives (n = 3).

This left 96 articles for the final eligibility check. From this pool, we excluded 34 studies due to missing essential data, such as specific comorbidity information or sex-specific prevalence, and an additional 13 studies because their study population included participants younger than 60 years of age.

Ultimately, 49 original studies with complete data on the prevalence and/or risk factors of multimorbidity in older adults were included in the final meta-analysis.

Studies characteristics

All features included in this study are shown in Table 1. Finally, 49 studies were selected for analysis. The studies included 729,043 older adults respondents, of whom 347,779 had two or more chronic conditions at the same time. The studies were published between 2010 and 2022, with sample sizes ranging from 279 to 229,493 people. Of the 49 studies, 26 were from China, while the remaining 23 were from 21 countries, including the United States, Japan, South Korea, India, Vietnam, Germany, Switzerland and others. Of these, 17 studies clearly reported various factors influencing multimorbidity. All included studies scored from a minimum of 4 to a maximum of 8 on the quality assessment.

Table 1.

Characteristic summary of included studies on the prevalence of multimorbidity in the older adults.

Author, year	Area	Time of survey	Sampling method	Sample size, (n)	Age	Gender		Quality score	Comorbidity rate, n (%)	Influencing factors of comorbidity	References
						Male, n (%)	Female, n (%)
Liu YT et al. 2022	China	2019	Cluster sampling	1,552	≥60	837 (53.92)	715 (46.08)	8	657 (42.33)	1, 5	(22)
Chen YT et al. 2023	China	2018	Convenience sampling	1,262	≥60	613 (48.57)	649 (51.43)	8	275 (21.79)	3, 4, 8, 9	(36)
Li GX et al. 2023	China	2021	Random sampling	1,310	≥60	586 (44.73)	724 (55.27)	8	685 (52.29)	7	(52)
Guo D et al. 2022	China	2018	Stratified quota sampling	7,507	≥65	3,734 (49.74)	3,773 (50.26)	7	5,383 (71.71)	1, 3	(37)
Kong Y et al. 2021	China	2021	Multistage cluster sampling	10,062	≥60	4,930 (48.99)	5,132 (51.01)	4	2,007 (19.95)	/	(50)
Li X et al. 2019	China	2017	Multi-stage stratified random sampling	4,833	≥60	2,199 (45.49)	2,634 (54.51)	6	778 (16.10)	/	(51)
Zhang H et al. 2019	China	2015	Multistage stratified cluster sampling	23,718	≥60	10,533 (44.41)	13,185 (55.59)	5	13,097 (55.22)	/	(56)
Qi YT et al. 2023	China	2018	Multilevel random sampling	7,354	≥60	3,727 (50.68)	3,627 (49.32)	7	4,792 (65.16)	1, 2, 5, 9, 10	(23)
Liu XX et al. 2023	China	2019	Multi-stage random sampling	107,177	≥60	48,841 (45.57)	58,336 (54.43)	5	14,511 (13.54)	/	(53)
Tian L et al. 2023	China	2021	Cluster sampling	15,899	≥65	6,787 (42.69)	9,112 (57.31)	5	5,476 (34.44)	/	(54)
Zhu PY et al. 2023	China	2019	Multi-stage random sampling	8,335	≥60	3,484 (41.79)	4,851 (58.21)	7	1,385 (16.62)	1, 3, 5–9	(24)
Yu ZJ et al. 2023	China	2020	/	1,849	≥60	801 (43.32)	1,048 (56.68)	4	1,021 (55.22)	1, 3	(59)
He YZ et al. 2023	China	2021	Cluster sampling	2,010	≥65	963 (47.91)	1,047 (52.09)	8	623 (30.99)	1, 2, 4, 9	(84)
Wang WH et al. 2024	China	2021	Multistage stratified cluster sampling	20,724	≥65	9,787 (47.22)	10,937 (52.78)	5	9,759 (47.09)	/	(61)
Mu YJ et al. 2023	China	2019	Cluster sampling	4,475	≥65	1,905 (42.56)	2,570 (57.44)	8	2,728 (60.96)	/	(60)
Yao YL et al. 2022	China	2020	/	2,506	≥60	1,041 (41.54)	1,465 (58.46)	4	566 (22.59)	/	(62)
Cao M et al. 2021	China	2019	Multistage stratified cluster sampling	1,336	≥60	645 (48.28)	691 (51.72)	8	490 (36.68)	/	(57)
Hou YT et al. 2020	China	2016	Multi-stage random sampling	622	≥65	264 (42.44)	358 (57.56)	6	317 (50.97)	1–3, 4, 9	(85)
Liu DN et al. 2023	China	2017	Multi-stage stratified random sampling	12,507	≥60	5,298 (48.25)	7,209 (51.75)	5	6,925 (55.37)	/	(58)
Zhang XQ et al. 2024	China	2021	/	1,491	≥60	596 (39.97)	895 (60.03)	8	1,192 (79.95)	/	(55)
Zhou FK et al. 2023	China	2018–2020	Cluster sampling	8,221	≥65	3,772 (45.88)	4,449 (54.12)	6	2,402 (29.22)	/	(63)
Marzban M et al. 2024	Iran	2015	/	2,426	≥60	1,161(47.85)	1,265(52.15)	5	1,946 (80.21)	2, 4, 5, 7, 8, 9	(86)
Oliveira-Figueiredo DST et al. 2024	Brazil	2020	Random sampling	22,728	≥60	10,193 (44.84)	12,535 (55.16)	7	11,728 (51.60)	1–3, 5, 6	(25)
Ko S et al. 2024	India	2018	Multi-stage stratified sampling	12,316	≥60	6,158 (50.00)	6,158 (50.00)	8	3,062 (24.86)	/	(64)
Su W et al. 2024	China	2020–2021	Stratified random sampling	1,161	≥60	499 (42.98)	662 (57.02)	8	366 (31.53)	1, 8, 9	(83)
Kohler S et al. 2024	Dares Salaam	2017–2018	Random sampling	555	≥60	243 (43.78)	312 (56.21)	7	465 (83.78)	/	(65)
Lee C et al. 2024	Korea	2019	/	1,041	≥65	233 (22.38)	808 (77.62)	6	936 (89.91)	/	(66)
Maimaitiwusiman Z et al. 2023	China	2019	Multi-stage random sampling	86,510	≥60	41,188 (47.61)	45,322 (52.39)	7	28,894 (33.40)	/	(67)
Reyes-Ortiz CA et al. 2023	Colombia	2015	Multi-stage stratified random sampling	18,873	≥60	8,757 (46.39)	10,116 (53.61)	7	8,172 (43.30)	1–3, 5, 7, 10	(26)
You L et al. 2023	China	2022	Multistage stratified cluster sampling	7,774	≥60	3,876 (49.89)	3,898 (50.11)	8	3,830 (49.27)	/	(68)
Yang K et al. 2023	China	2021	Convenience sampling	4,803	≥60	2,276 (47.38)	2,527 (52.62)	8	1,883 (39.21)	/	(69)
Honda Y et al. 2022	Japan	2013	Stratified random sampling	23,340	≥65	10,266 (43.98)	13,074 (56.02)	5	9,574 (41.02)	/	(70)
Lynch DH et al. 2022	United States	2005–2014	Multi-stage random sampling	7,261	≥60	3,652 (50.30)	3,609 (49.70)	6	4,965 (68.38)	/	(71)
Balakrishnan S et al. 2022	Eastern Nepal	2020	Multistage cluster sampling	843	≥60	431 (51.12)	412 (48.88)	8	192 (22.78)	1, 5, 6–8, 10	(27)
Keomma K et al. 2022	São Paulo	2015	Complex probability sampling	1,019	≥60	387 (37.97)	632 (62.03)	7	408 (40.04)	2, 5, 9	(28)
Shariff Ghazali S et al. 2021	Malaysia	2018	Two-stage stratified cluster sampling	3,966	≥60	1,868 (47.10)	2,098 (52.90)	5	1,543 (38.91)	/	(72)
Lin WQ et al. 2022	China	2020	Multi-stage stratified random sampling	31,708	≥65	14,046(44.29)	17,662 (55.71)	5	4,819 (15.20)	/	(73)
Sara HH et al. 2018	Bangladesh	2017	Two-stage stratified random sampling	566	≥60	432 (76.32)	134 (23.68)	7	319 (56.36)	/	(74)
Smith L et al. 2022	Irish	2009–2011	Multistage stratified cluster sampling	2,941	≥65	1,323 (44.98)	1,618 (55.02)	8	2,041 (69.40)	/	(75)
Jovic D et al. 2016	Serbia	2013	Two-stage stratified random sampling	2,749	≥65	1,174 (42.70)	1,575 (57.30)	5	1,578 (57.40)	/	(76)
Puth MT et al. 2017	Germany	2012–2013	Two-stage random sampling	7,152	≥60	3,454 (48.29)	3,698 (51.71)	5	4,882 (68.26)	/	(77)
Bähler C et al. 2015	Switzerland	2013	/	229,493	≥65	98,124 (42.76)	131,369(57.24)	5	175752 (76.58)	/	(78)
Ha NT et al. 2015	Vietnam	2010	Multi-stage random sampling	2,400	≥60	834 (34.75)	1,566 (65.25)	7	941 (39.21)	1–6, 8, 10	(29)
Yadav UN et al. 2021	Nepal	2018	Multistage cluster sampling	794	≥60	400 (50.38)	394 (49.62)	8	116 (14.61)	/	(79)
Hien H et al. 2014	Burkina Faso	2012	Random sampling	389	≥60	215 (55.27)	174 (44.73)	8	252 (64.78)	/	(80)
Aye SKK et al. 2019	Myanmar	2016	Multi-stage random sampling	4,859	≥60	1,841 (37.89)	3,018 (62.11)	7	1,613 (33.20)	2, 4, 5, 6	(87)
Asante D et al. 2022	Australia	2013–2017	Random sampling	5,920	≥60	2,442 (41.25)	3,478 (58.75)	7	1,973 (33.33)	/	(81)
Abdulazeez ZU et al. 2021	Nigeria	2018	Random sampling	279	≥60	93 (33.40)	186 (66.70)	8	201 (72.04)	/	(82)
Nugraha S et al. 2020	Indonesia	2018	Random sampling	427	≥60	137 (32.08)	290 (67.92)	5	259 (60.66)	/	(83)

1: senior age, 2: gender (female), 3: obesity, 4: educational level (high), 5: smoking, 6: drinking, 7: exercise situation, 8: marital status, 9: income level, 10: place of residence.

Methodological quality assessment results

The quality evaluation results of this study showed that 16 studies (32.65%) were rated as high quality, while 33 studies (67.35%) were rated as medium quality. All included studies defined multimorbidity and were cross-sectional studies. More than half of the studies (51.02%) clarified the inclusion and exclusion criteria for participants. In most of the included studies, assessments to assure study quality (89.79%) and summaries of patient response rates and data integrity (71.43%) were reported. However, none of the studies described the impact of the evaluators' subjective factors on the findings, nor did they detail how missing data was dealt with in their analyses. Details are given in supplementary Table 3.

Meta-analysis of the prevalence of multimorbidity

A meta-analysis of the prevalence of multiple comorbidities was conducted on the 49 included studies using a random-effects model. The results showed that the overall combined prevalence based on all 49 studies (n = 729,043) was 46.0% (95% CI: 38.0%−55.0%), with extremely high heterogeneity (I² = 99.98%, P = 0.001). The forest map is shown in Figure 2.

Figure 2.

Meta-analysis of the prevalence of multimorbidity in the older adults.

To explore the sources of heterogeneity, we performed additional analyses to examine the prevalence of multimorbidity using subgroup analysis and meta-regression methods. The results of our subgroup analysis indicated that gender, age, research quality, study location, survey time, and sample size were not important factors affecting the heterogeneity of this meta-analysis. Although most subgroup differences did not reach statistical significance, we retained Table 2 to transparently present the direction and magnitude of heterogeneity across study characteristics. These findings help identify potential sources of clinical and methodological diversity, which are informative for future research design and policy planning. However, meta-regression results showed that study location was an important factor affecting the heterogeneity of this meta-analysis.

Table 2.

Subgroup analysis and Meta-regression analysis of the prevalence of multimorbidity in the older adults.

Analysis	Studies, n	Subgroup				Meta-regression
		Pooled estimate	95% CI	I 2	P -value	Coefficient	95% CI	P -value
Gender						0.072	(−0.018, 0.161)	0.117
Male	37	0.41	(0.35–0.47)	99.88%	< 0.001
Female	37	0.48	(0.41–0.55)	99.93%	< 0.001
Age						0.074	(−0.059, 0.208)	0.270
≥60	36	0.44	(0.38–0.51)	99.95%	< 0.001
≥65	13	0.52	(0.35–0.69)	99.99%	< 0.001
Research quality						−0.071	(−0.219, 0.076)	0.334
Medium	33	0.48	(0.39–0.57)	99.99%	< 0.001
High	16	0.41	(0.27–0.54)	99.88%	< 0.001
Study location						0.132	(0.019, 0.246)	0.023
China	26	0.40	(0.33–0.47)	99.96%	< 0.001
Other	23	0.53	(0.44–0.63)	99.95%	< 0.001
Survey time						−0.098	(−0.227, 0.031)	0.133
~2020	35	0.49	(0.39–0.59)	99.99%	< 0.001
2021~	14	0.39	(0.30–0.49)	99.92%	< 0.001
Sample size						0.138	(−0.047, 0.324)	0.141
10,000~	13	0.39	(0.23–0.56)	100.00%	< 0.001
1,001–9,999	28	0.48	(0.40–0.56)	99.89%	< 0.001
~1,000	8	0.53	(0.38–0.55)	99.59%	< 0.001

Sensitivity analysis

The robustness of the findings was assessed by the leave-one method, and each study was individually excluded for analysis. The prevalence of multimorbidity in the older adults ranged from 45.56% to 47.15%, indicating that the overall prevalence of comorbidity fluctuated less, which meant that the study results were stable and reliable. In addition, a sensitivity analysis of 33 medium-quality studies showed that the prevalence of multimorbidity in the older adults ranged from 46.82% to 48.83%, further suggesting that the findings were also stable and reliable in medium-quality studies.

Publication bias

Potential publication bias was assessed by visual inspection of a funnel plot and formally tested using Egger's linear regression test. The funnel plot, which plots the study effect size against its standard error, appeared to be largely symmetrical around the pooled effect estimate (Figure 3). However, given the substantial heterogeneity present in this meta-analysis, the interpretation of the funnel plot's symmetry should be approached with caution, as heterogeneity can be a confounding factor for plot asymmetry. To supplement the visual assessment, Egger's test was performed. The result did not provide statistical evidence of significant publication bias (P = 0.798). While this statistical test supports the visual inspection, the potential influence of heterogeneity on these assessments cannot be entirely ruled out.

Figure 3.

Funnel plot of publication bias.

Meta-analysis of influencing factors of multimorbidity

Of the 49 studies included, 17 examined factors influencing multimorbidity in older adults. Through the extraction of data from these 17 studies, 10 risk factors associated with multimorbidity were identified and their effect sizes were pooled for analysis. The results of the analysis showed that the following factors were identified as significant risk factors for multimorbidity in older adults: age (≥70 years), gender (female), body mass index (BMI ≥ 28kg/m²), education level (at least junior high school level), marital status (single, widowed, divorced), and higher income level. However, this study did not find that smoking, alcohol consumption, lack of exercise, and place of residence had significant effects on the occurrence of multimorbidity. Detailed data are shown in Table 3.

Table 3.

Results of meta-analysis of risk factors for multimorbidity in the older adults.

Risk factors	Studies, n	Heterogeneity		Effect model	Pooled estimate
		P	I2 (%)		OR	95% CI	Z	P
Age (≥70)	11	0.001	96.46	random	1.43	1.09–1.89	2.57	0.010*
Gender (Female)	9	0.001	86.80	random	1.59	1.38–1.83	6.44	0.001*
BMI (≥28.0 kg/m2)	7	0.001	96.73	random	1.97	1.28–3.03	3.10	0.002*
Educational	6	0.072	55.18	random	0.56	0.43–0.73	−4.36	0.001*
Smoking	10	0.006	62.92	random	1.09	0.99–1.20	1.71	0.088
Tipple	5	0.001	82.80	random	0.88	0.64–1.20	−0.81	0.416
Exercise (lack)	5	0.001	95.46	random	0.82	0.52–1.30	−0.85	0.397
Marital status	6	0.004	80.44	random	1.32	1.02–1.71	2.09	0.037*
Income (low)	8	0.001	83.93	random	1.46	1.11–1.91	2.71	0.007*
Live in the city	4	0.001	96.35	random	1.09	0.64–1.83	0.31	0.757

CI, confidence interval; ^*P < 0.05

Sensitivity analysis

In this study, sensitivity analysis was conducted under fixed effects model and random effects model for OR values and 95%CI of each influencing factor. The results showed that although there were some differences in the calculation results of smoking and exercise, the OR values and 95%CI of the two models showed significant agreement in the analysis of other influencing factors. This finding indicates that the meta-analysis results of influencing factors in this study have good stability, and the details are shown in Table 4.

Table 4.

Sensitivity analysis of risk factors.

Risk factors	Fixed effect model			Random effects model
	OR	95% CI	P	OR	95% CI	P
Age (≥70)	1.21	1.17–1.26	0.001*	1.43	1.09–1.89	0.010*
Gender (Female)	1.43	1.38–1.49	0.001*	1.59	1.38–1.83	0.001*
BMI (≥28.0kg/m2)	1.57	1.47–1.68	0.001*	1.97	1.28–3.03	0.002*
Educational	0.58	0.51–0.67	0.001*	0.56	0.43–0.73	0.001*
Smoking	1.13	1.09–1.17	0.001*	1.09	0.99–1.20	0.088
Tipple	0.92	0.82–1.03	0.155	0.88	0.64–1.20	0.416
Exercise (lack)	1.12	1.03–1.22	0.011*	0.82	0.52–1.30	0.397
Marital status	1.21	1.09–1.34	0.001*	1.32	1.02–1.71	0.037*
Income (low)	1.51	1.37–1.66	0.001*	1.46	1.11–1.91	0.007*
Live in the city	1.05	0.96–1.15	0.257	1.09	0.64–1.83	0.757

CI, confidence interval; ^*P < 0.05

Publication bias

For the overall prevalence analysis, we visually inspected the funnel plot and conducted Egger's test to assess publication bias. The funnel plot showed a roughly symmetrical distribution (Figure 3), and Egger's test did not indicate significant publication bias (P = 0.798). Additionally, for the “risk factors” meta-analysis, we only conducted a publication bias assessment for the factors with a large number of included studies (≥10 studies), such as age and smoking. Figure 4 shows the funnel plots for age and smoking, and their asymmetry suggests the possibility of some publication bias, which should be taken into account when interpreting the combined results of these two specific risk factors.

Figure 4.

Age and smoking publication bias funnel plot, the picture on the left is age and the picture on the right is smoking.

Discussion

This study was the first to systematically integrate 49 Chinese and English papers from 22 countries around the world. The aim was to comprehensively understand the prevalence of multimorbidity among the older adults and the related influencing factors through meta-analysis. We found that the problem of multimorbidity is extremely common among the older adults population. We calculated that the global prevalence of multimorbidity was 46.0%, but we also found that this prevalence showed significant differences globally (I² = 99.98%). This also indicates that using a single numerical value to represent the multimorbidity status of the global older adults population is inappropriate, and it is crucial to identify the factors causing these differences. Meanwhile, there are several factors affecting the prevalence of multimorbidity in the older adults, including age (≥70 years), gender (female), body mass index (BMI ≥ 28kg/m²), education level (junior high school or above), marital status (single, widowed, divorced) and higher income level. These factors have a significant impact on multimorbidity among older adults.

Exploration of the sources of heterogeneity

Our analysis revealed that the study location was the most important factor explaining the high heterogeneity of the prevalence of multiple comorbidities. The most significant finding was that the combined prevalence rate of older adults people in China was significantly lower than the aggregated results of other countries. We suggest that this significant difference may stem from complex interactions among factors such as disease spectrum, adopted diagnostic criteria, common lifestyles, and national medical system characteristics. This study effectively overcame language barriers by systematically searching and including both Chinese and English literature, especially a large number of Chinese studies. It compensated for the shortcomings of previous reviews that were limited to English publications and might have overlooked key evidence, providing a more comprehensive and representative picture of the global prevalence of comorbidities. Furthermore, there are significant differences in the definition of multimorbidity included in the studies: some studies define it as ≥2 chronic diseases, while others adopt the standard of ≥3 diseases. This directly leads to the incomparability of prevalence estimates and is another important reason for the heterogeneity. Such inconsistencies in definition are widespread in international literature, reflecting the lack of a unified standard in this field. Future research needs to reach a consensus on this to improve the comparability of results.

Influence of age on multimorbidity

Our meta-analysis reaffirms that advanced age is arguably the most potent, non-modifiable risk factor for multimorbidity, a finding consistent with a vast body of literature (30). The significantly increased risk observed in individuals aged 70 and over (22–26, 50–52) is deeply rooted in the fundamental biological processes of aging. This study confirmed that being 70 years old or older is a risk factor for multimorbidity (OR = 1.43, 95% CI: 1.09–1.89). This further supports the mechanism that 'aging-related physiological changes (such as cellular aging, inflammatory aging) are the core driving factors for multimorbidity (31). However, it should be noted that Smith et al. (2022) conducted a longitudinal study on older adults people in American communities (n = 18,721), finding that for individuals aged 60–69 who had a “hypertension + diabetes” background, the risk of multimorbidity after the age of 70 was 2.3 times higher than that of those without such a background (HR = 3.30, 95% CI: 2.91–3.75) (32). This suggests that the interaction effect of “age + cumulative underlying diseases” might have been overlooked by the cross-sectional design of this study. Future research needs to verify this interaction through longitudinal data.

Influence of gender on multimorbidity

The study results further revealed the gender differences in the prevalence of multimorbidity in the older adults population, showing that the prevalence of multimorbidity in older adults women was significantly higher than that in older adults men (23, 25, 26, 28, 29, 50–53). The higher co-morbidity risk in women is closely related to the physiological changes after menopause and their medical-seeking behaviors. The decline in estrogen levels significantly increases women's susceptibility to various chronic diseases such as cardiovascular diseases and osteoporosis (33). In addition, women usually tend to report symptoms and seek medical help (34, 35), which may partly contribute to their higher prevalence being recorded. These two factors jointly explain the observed gender differences.

Influence of body mass index on multimorbidity

Body mass index (BMI) is one of the important factors affecting the occurrence of multimorbidity in the older adults. Our findings suggest that the risk of multimorbidity in older adults is significantly increased when BMI exceeds 28kg/m² (24–26, 29, 36, 37, 51). Obesity is the core metabolic driving factor of multiple comorbidities, mainly exerting its effects through two pathways: inflammation and insulin resistance. The pro-inflammatory cytokines secreted by adipose tissue (such as TNF-α and IL-6) establish a chronic low-level inflammatory state (38). Meanwhile, obesity directly leads to insulin resistance (39). These two systemic obstacles jointly create an internal environment that is highly prone to triggering a variety of chronic diseases (such as diabetes and cardiovascular diseases), rather than merely increasing the risk of a single disease.

Influence of educational level on multimorbidity

Educational level influences the risk of comorbidity by enhancing health literacy and the ability to access resources. Higher educational levels not only directly empower individuals to adopt healthy behaviors (such as quitting smoking, maintaining a healthy diet) but also effectively utilize medical information (29, 36, 51, 52). It is also usually associated with a higher social and economic status, thereby ensuring better accessibility to medical resources and health protection (40, 41). These advantages collectively facilitate the early management and intervention of chronic diseases and are the key social determinants for reducing the risk of comorbidities.

Influence of marital status on multimorbidity

Marital status significantly influences the risk of multiple diseases through social support mechanisms. Our analysis has confirmed that being unmarried (single, widowed, or divorced) is an important risk factor for multiple diseases (24, 27, 29, 36, 42, 52). Married individuals usually can obtain more stable emotional support and health supervision from their spouses, which is beneficial for the management of chronic diseases (43). Conversely, the lack of social support may exacerbate the emotional stress of non-married older adults individuals and weaken their compliance with healthy behaviors, thereby promoting the occurrence and development of multiple diseases (44). Therefore, when formulating intervention strategies, it is of utmost importance to establish a strong social support network for older adults people living alone.

Influence of economic base on multimorbidity

Our pooled results suggest that lower income levels are an important risk factor against multimorbidity (23, 24, 28, 36, 42). Individuals with higher incomes generally have access to better quality health care services, healthier living environments, and more adequate resources to manage and manage chronic diseases (45). They were also more likely to engage in healthy lifestyles, such as regular exercise, maintaining a nutritionally balanced diet, and limiting tobacco and alcohol consumption. To address the challenges posed by economic disparities, implementing policies that reduce poverty, provide affordable housing and equitable access to health care could have a positive impact on reducing the burden of multimorbidity among older adults.

Influence of other factors on multimorbidity

Regarding factors such as smoking and drinking, this study did not find a significant correlation between these factors and multiple diseases. However, this result should be interpreted with caution (46), this is likely to reflect the fact that the existing related studies are few in number and highly heterogeneous, resulting in insufficient statistical test power and failing to confirm that these factors are truly ‘harmless'. Therefore, we should not dismiss the importance of a healthy lifestyle based on this. Adhering to quitting smoking, limiting alcohol consumption, and maintaining regular exercise remains a fundamental strategy for the older adults to maintain their health and manage chronic diseases. Given the high prevalence of multimorbidity among the older adults, we believe that an integrated care model is necessary to address the complex health care needs of this group (47). In addition, healthcare providers should receive systematic training on the principles of geriatric care, which should include a holistic approach to care that is oriented toward the physical, psychological and social needs of older persons (48). Finally, public health initiatives to address modifiable risk factors, such as obesity, literacy, smoking and alcohol consumption, will help to prevent the occurrence of multimorbidity and thus reduce the health burden on older persons. Community-based healthy aging promotion programs, such as exercise classes, nutrition counseling, and smoking cessation programs, will play a vital role in promoting health and independence for older adults (49).

Limitations

The limitations of this study are mainly reflected in the following four aspects: First, there was extremely high heterogeneity among the included studies, which represents the most prominent limitation of this study. Therefore, the reported overall pooled prevalence rate should be regarded as a rough average estimate, and its interpretation must be approached with extreme caution. Uncertain confounding factors-such as variations in survey methods, survey time points, and geographical diversity-may have contributed to these differences, and such uncertainties may introduce bias when results are combined. We attempted to explore potential sources of heterogeneity and concluded that differences in survey locations were the primary factor contributing to heterogeneity. Additionally, inconsistency in the definition of multimorbidity (e.g., most studies (n = 38) defined it as the presence of ≥2 chronic diseases, while the remaining 11 adopted a stricter threshold of ≥3 diseases) was another important reason, which directly led to incomparability among prevalence estimates. Second, the most significant limitation of this systematic review lies in the initial search strategy, which may have failed to comprehensively capture global evidence. Although we searched both Chinese and English databases, the relatively high proportion of studies from China among the finally included studies (26 from China vs. 23 from other countries worldwide) may not fully represent the global epidemiological status of multimorbidity. This imbalance may have been influenced by our original search strategy, which might have overrepresented terms commonly used in Chinese literature. Although we subsequently expanded the strategy to include more globally relevant terms, the existing evidence base in the literature could not be altered. This highlights a common challenge in global meta-analyses and suggests that the pooled prevalence should be interpreted primarily as a summary of existing identified literature rather than a definitive global average. The absence of certain key terms and the overrepresentation of Chinese studies indicate that our evidence base may be subject to geographical and linguistic biases. Thus, the results of this study should be considered an integration of available evidence from Chinese and English literature rather than a fully balanced global representation. Future research should intentionally ensure equitable geographical representation from the design stage onward. Third, in our main prevalence analysis, we included 49 studies, but only 17 of them were incorporated into the analysis of influencing factors. This limited sample size may reduce the statistical power to detect associations. Moreover, due to the inherent limitations of observational study designs, it is difficult to establish causal relationships between various risk factors and multimorbidity, which complicates the interpretation of the results. Fourth, many included studies did not report how missing data were handled, which may introduce selection bias and lead to underestimation of the true prevalence of multimorbidity. Additionally, publication bias may affect the pooled results, as studies with significant positive outcomes or larger sample sizes are more likely to be published, potentially skewing the overall findings. It is important to note that the lack of significant association between factors such as smoking, alcohol consumption, and physical inactivity with multimorbidity should be interpreted cautiously. This likely does not indicate a true absence of association but may rather reflect the limited number of original studies available for correlation analysis (n < 10), resulting in insufficient statistical power to detect a true effect. More specialized research is needed to deeply explore the complex relationship between these lifestyle factors and multimorbidity.

Conclusion

In conclusion, this meta-analysis establishes that multimorbidity affects a substantial proportion of the world's older adult population, with a pooled prevalence of 46.0%. It is crucial to note that the evidence base of this study is more focused on reflecting the situation in China and regions where literature is widely published in Chinese and English. Consequently, while providing a global average, this figure conceals profound geographical and demographic disparities, underscoring that context-specific understanding, rather than a single estimate, is essential for effective policy. Beyond quantifying the burden, this study identifies a consistent set of associated factors, highlighting that advanced age and female sex are key non-modifiable risks, while obesity, lower education, economic disadvantage, and non-married status represent pivotal modifiable determinants rooted in social patterning. These findings compellingly argue for a dual approach: clinically, a shift toward integrated, patient-centered care models that address co-existing conditions; and at the policy level, multisectoral actions tackling social determinants of health to promote healthy aging and alleviate the burden of multimorbidity.

Funding Statement

The author(s) declare that financial support was received for the research and/or publication of this article. Guangxi Project for the Development and Promotion of Appropriate Medical and Health Technologies (S2024030), Reform Project of Degree and Postgraduate Education (SA2400000928).

Data availability statement

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

Author contributions

XZ: Data curation, Formal analysis, Writing – original draft. ZW: Data curation, Writing – original draft. XY: Formal analysis, Methodology, Writing – review & editing. ZN: Methodology, Supervision, Validation, Writing – review & editing.

Conflict of interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Generative AI statement

The author(s) declare that no Gen AI was used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher's note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpubh.2025.1680745/full#supplementary-material