Prevalence and potential influencing factors of social frailty among community-dwelling older adults in China: systematic review and meta-analysis Authors: Jinwei Tang, Chen Shen, Ling Wu, Chun Gao, Huiping Xue Published: 2026 ABSTRACT AbstractObjectiveBased on the local context in China, this study systematically reviews and explores the prevalence of social frailty among community-dwelling older adults in China and its potential influencing factors.MethodsA systematic search was conducted in databases including CNKI, VIP, Wanfang Data, the Chinese Biomedical Literature Database, PubMed, the Cochrane Library, Web of Science, and Embase, covering the period from the inception of each database to November 9, 2025. Two researchers independently performed the systematic literature search, data extraction, and article quality assessment. Meta-analysis was conducted using Stata 15.1 software, incorporating cross-sectional and cohort studies; a random-effects model was used to calculate the pooled prevalence of social frailty and pre-frailty; odds ratios (OR) and 95% confidence intervals (CI) were used to assess associated influencing factors; the I2statistic was used to assess heterogeneity, and subgroup analysis and meta-regression were performed.ResultsAn initial search identified 5,488 articles; after screening, a total of 24 studies (n = 242,591) were ultimately included. The random-effects model showed that the pooled prevalence of social frailty among community-dwelling older adults in China was 25.0% (95% CI: 19.4%–31.0%, I2 = 99.47%, P < 0.001),and the prevalence of pre-social frailty was 42.0% (95% CI: 37.6%–46.3%, I2 = 94.80%, P < 0.001). Gender (female), advanced age (≥ 85 years), depression, physical frailty, housing satisfaction, marital status, and physical frailty were significantly associated with the risk of social frailty (OR: 1.36–4.15).ConclusionsPre-social frailty and social frailty are common health challenges faced by community-dwelling older adults in China. Recent data indicate that the prevalence of these conditions remains high; although the prevalence is the very high heterogeneity and should be interpreted with caution, it nevertheless underscores the necessity and urgency of implementing effective interventions. Early identification and intervention for individuals at risk of social frailty are of critical importance for advancing the Healthy China strategy and achieving the goals of active aging.PROSPERO registration numberCRD420251246136.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12877-026-07600-w.Keywords: Older adults, Community-dwelling, Social frailty, Risk Factor, Systematic Review, Meta-Analysis Introduction The global population is rapidly aging, and China’s population aged 60 and older continues to grow. According to the Seventh National Population Census in 2020, China’s population aged 60 and older reached 264 million, accounting for 18.7% of the total population, with the aging process continuing to deepen [1]. Against the backdrop of intensifying global population aging, frailty has emerged as a key public health issue. Frailty is a widely used concept describing a complex state of increased vulnerability resulting from adverse health outcomes associated with aging. In recent years, the concept of frailty has evolved from a narrow focus on physiological frailty to a more comprehensive framework encompassing multiple dimensions—including physiological, psychological, and social aspects [2]. In 2017, Bunt [3] and colleagues first proposed the concept of social frailty, defined as a continuous state of risk in which an individual is at risk of losing, or has already lost, the essential resources, activities, or capabilities required to meet their basic social needs. Social frailty is a dynamic and multidimensional concept; although its manifestations vary among individuals across different cultural and social contexts, its core mechanisms remain consistent. In recent years, research has begun to focus on social frailty among older adults in Chinese communities; however, due to differences in sample sources, measurement tools,and geographical distribution, existing studies have reported inconsistent findings regarding prevalence and influencing factors, and there is a lack of systematic integration. Given China’s unique family structures characterized by a low birth rate and “hollowed-out”households, as well as its urban–rural dichotomy and rapid urbanization, the prevalence of social frailty and its associated factors may exhibit distinct local characteristics. Therefore, we believe there is a need to conduct a systematic review and meta-analysis among community-dwelling older adults in China to comprehensively synthesize existing evidence, clarify the current status of social frailty and its primary influencing factors, and provide evidence-based support for the development of targeted intervention strategies. Consequently, this study aims to address three questions: ① What is the combined prevalence of social frailty and pre-frailty among community-dwelling older adults in China?② Do prevalence rates vary by gender, region, urban/rural status, and time period? ③ Which factors are significantly associated with social frailty? Methods The study protocol has been registered on PROSPERO (CRD420251246136) and was conducted in accordance with the PRISMA [4] and MOOSE [5] guidelines. Search strategies We systematically searched databases including CNKI, VIP, Wanfang Data, CBM, PubMed, Cochrane Library, Web of Science, and Embase from their inception to November 9, 2025. Details of the search strategy were as follows: (Social frailty OR Social vulnerability OR social exclusion OR social isolation) AND (Frailty OR Frailty syndrome) AND (Aged OR Elderly OR Older adult OR Older adults). Selection criteria Inclusion and exclusion criteria: (a) Study design includes cross-sectional or cohort studies; (b) Participants are community-dwelling older adults in China (excluding hospitalized or nursing home residents), aged ≥ 60 years;(c) Outcome measures focused on the prevalence of social frailty and its potential determinants; (d) Diagnostic criteria: Social frailty must be diagnosed using validated instruments such as the Makizako Social Frailty Index (MSFI), Help, Activity, Loneliness, Economy, and Conversation Scale (HALEFT),the Tilbury Frailty Index (TFI), Social Frailty Index (SFI), Social Frailty Screening Questionnaire (SFSQ), Social Frailty Questionnaire (SFQ), Social Vulnerability Index (SVI), Lubben Social Network Scale (LSNS-6), and Bunt; (e) Relevant studies were included regardless of literature quality. Exclusion criteria: (a) Literature types such as reviews, conference papers, or abstracts; (b) Duplicate publications; (c) Literature where full text or required data were unavailable; Study selection Two researchers independently screened the literature. When disagreements arose, a third researcher was consulted. The process was as follows: first, duplicate documents were removed; then, titles and abstracts were reviewed to exclude obviously irrelevant studies; next, full texts of remaining studies were examined; finally, studies meeting inclusion criteria were selected. All disagreements were resolved through tripartite discussion. Data were extracted using a standardized form, including: study authors, publication year, Chinese region, mean age, study type, sample size, sample source, prevalence of social frailty/pre-frailty, influencing factors, and assessment tools. Quality assessment Two researchers assessed the quality of included cross-sectional and cohort studies using the Agency for Healthcare Research and Quality (AHRQ) recommended cross-sectional study evaluation tool and the Newcastle–Ottawa Scale (NOS), respectively. Disagreements were resolved through discussion involving a third researcher. The AHRQ tool comprises 11 items answered as "yes," "no," or "unclear," with "yes" scoring 1 point and "no" or "unclear" scoring 0 points, yielding a total range of 0–11 points. The NOS tool evaluates cohort studies across 8 items, with a maximum score of 9 points, categorizing studies as poor, fair, or good quality. Based on the quality scores, the included studies were classified into three categories: low risk of bias (AHRQ ≥ 8 or NOS ≥ 7), moderate risk of bias (AHRQ 4–7 or NOS 4–6), and high risk of bias (AHRQ ≤ 3 or NOS ≤ 3). Statistical analysis In addition, this study analyzed the prevalence of social frailty among older adults in the community and its 95% confidence interval (CI). We calculated pooled odds ratios (OR) and 95% CIs, identified factors associated with social frailty, and tested for heterogeneity using Stata 15.1. If the I2 statistic was less than 50%, the heterogeneity was classified as moderate or low, and a fixed-effects model was used for analysis; if I2 was 50% or higher, a random-effects model was used. The significance level for the meta-analysis was set at α = 0.05. To investigate heterogeneity, we predefined the following subgroup analyses: gender, study design, social frailty diagnostic tool, study quality, sample size, region of sample origin, Chinese region, and publication year. Differences between subgroups were compared using the Q-test. Furthermore, a random-effects meta-regression model was employed to examine the effects of publication year, study design, type of diagnostic tool, and sample size on heterogeneity. Results Study selection The initial search yielded 5,488 relevant articles from sources including CNKI (n = 262), VIP (n = 333), Wanfang (n = 27), China Biomedical Literature Database (n = 38), PubMed (n = 594),Web of Science (n = 4064), Embase (n = 130), and the Cochrane Library (n = 40). After removing 1267 duplicates, 2864 studies were excluded based on title and abstract screening. Following full-text review, an additional 1333 studies were excluded, resulting in 24 studies ultimately included in the meta-analysis. The specific screening process is illustrated in Fig. 1. Study selection flowchart Study characteristics A total of 24 studies were included, and their basic characteristics are summarized in Table 1. These studies were published between 2018 and 2025, covering communities or rural areas across China, and involved a total of 242,591 participants. The study designs included 20 cross-sectional studies and 4 cohort studies, with sample sizes ranging from a minimum of 188 to a maximum of 222,179 participants. Various tools were used to diagnose social frailty across the studies, including the HALFT, MSFI, SFSQ, SFI, TFI, and SVI. The HALFT, MSFI, SFI, and SFSQ are all diagnostic tools developed based on the theory of social need satisfaction; among these, the HALFT scale was the most widely used, with 15 studies employing this tool. Table 1Characteristics of included studiesAuthorsYearRegion in ChinaMean AgeStudy TypeSample SizeSample SourceNumber of men and womenPrevalence of social frailtyPrevalence of Pre-Social FrailtySocial Frailty Diagnostic Toolinfluencing factorsQuality assessment scoreQi X [6]2023Nationwide ≥ 60Cross-sectional study222,179Rural, CommunityMale: 116,038Female: 106,14115.2%HALFTa, b, c, f7Sun, Y [7]2025East China≥ 60Cross-sectional study1,453Rural, CommunityMale: 662Female: 79114.5%31.5%HALFT7Hu X [8]2024East China71.73 ± 6.4Cross-sectional study1175MixedMale: 530Female: 64515.7%HALFT8Qin X [9]2025East China70.82 ± 7.432Cross-sectional study1611MixedMale: 930Female: 68144.2%HALFT7ZeKun Chen [10]2021East China75.3 ± 3.9Cohort study1764RuralMale: 944Females: 8203.5%36.8%HALFT7Li Zhixiao [11]2024Central China ≥ 60Cross-sectional study295CommunityMale: 136Female: 15929.8%53.9%HALFT9L. MA [12]2018North China≥ 60Cohort study1,697CommunityMale: 193Female: 2497.7%45.6%HALFT7Sun [13]2023Southwest≥ 60Cooperative study460CommunityMixed24.3%MSFIb9Wu [14]2023East China76.5 ± 6.9Cross-sectional study1,251RuralMale: 280Females: 97121.7%34.7%Bunt5Chen [15]2023East China70.33 ± 5.357Cross-sectional study208CommunityMale: 83Females: 12545.7%SFSQ6Yu Shan [16]2025Other regions70.80 ± 7.48Cross-sectional study1826Rural and communityMale: 837Females: 98962.3%SFIa, d5Hao Juan [17]2022Central China≥ 60Cross-sectional study454Rural, CommunityMen: 192Women: 26211.2%54.4%HALFTa, b, c, d5Zhang Wenya [18]2021Central ChinaSocially Frail Group:71.4 ± 6.4Non-socially frail group:68.3 ± 6.0Cross-sectional study359Rural, communityMale: 149Females: 21010.9%HALFT5Qin Yanmei [19]2025North China≥ 60Cross-sectional study1,450CommunityMale: 702Female: 74838.6%HALFTe5Zhao Linlin [20]2025North China74.09 ± 6.03Cross-sectional study919CommunityMale: 329Female: 59011.2%HALFTe, f4Liu Yue [21]2024East China73.61 ± 5.73Cohort study300RuralMale: 139Female: 16122.3%41.0%SFS-88Tang Long [22]2025Other regions≥ 60Cross-sectional study235CommunityMen: 64Female: 17157.9%SFI8Liu Linfeng [23]2024Southwest≥ 60Cross-sectional study215CommunityMale: 121Female: 9416.3%HALFT4Chen Yingwei [24]2022Southwest73.8 ± 5.6Cross-sectional study2,987Rural, communityMale: 1,358Females: 1,62925.9%41.4%HALFT6Peng Xinyu [25]2021Central China≥ 60Cross-sectional study313CommunityMale: 184Female: 12915.7%41.2%HALFT6Liu Hang [26]2022Central China≥ 60Cross-sectional study194CommunityMale: 104Female: 9039.2%TFI5Song Ge [27]2022North ChinaSocially Frail Group 74.55 ± 6.34 Non-Socially Frail Group 73.22 ± 5.33Cross-sectional study516CommunityMale: 224Female: 29212.8%HALFT7Bai Huiqiong [28]2023North China71.57 ± 6.45Cross-sectional study188CommunityMale: 83Females: 10565.4%SVI-C6Zhang Yuxi [29]2025East China71.69 ± 6.54Cross-sectional study542CommunityMale: 267Female: 27530.1%SFI5a: Gender; b: Marital status; c: Satisfaction with living environment; d: Depression; e: Physical frailty; f: Age (≥ 85 years vs 65–69 years) Characteristics of included studies Male: 116,038 Female: 106,141 Male: 662 Female: 791 Male: 530 Female: 645 Male: 930 Female: 681 Male: 944 Females: 820 Male: 136 Female: 159 Male: 193 Female: 249 Male: 280 Females: 971 Male: 83 Females: 125 Male: 837 Females: 989 Men: 192 Women: 262 Socially Frail Group: 71.4 ± 6.4 Non-socially frail group: 68.3 ± 6.0 Male: 149 Females: 210 Male: 702 Female: 748 Male: 329 Female: 590 Male: 139 Female: 161 Men: 64 Female: 171 Male: 121 Female: 94 Male: 1,358 Females: 1,629 Male: 184 Female: 129 Male: 104 Female: 90 Male: 224 Female: 292 Male: 83 Females: 105 Male: 267 Female: 275 a: Gender; b: Marital status; c: Satisfaction with living environment; d: Depression; e: Physical frailty; f: Age (≥ 85 years vs 65–69 years) Methodological quality Table 1 presents the results of the methodological quality assessment for all 24 studies. Among the cross-sectional studies, 3 were rated as having a low risk of bias, and 17 were rated as having a moderate risk of bias. Among the cohort studies, all 4 were rated as having a low risk of bias. The most common causes of risk of bias included failure to report non-response rates and failure to specify the number and causes of missing data. Prevalence of social frailty and social pre-frailty A total of 24 studies involving 242,591 participants were included in the meta-analysis of the prevalence of social frailty. The prevalence of social frailty reported in the studies ranged from 3.5% to 65.4%. Based on a random-effects model, the pooled prevalence of social frailty among community-dwelling older adults in China was 25% (95% CI: 19.4%–31.0%). There was significant heterogeneity among the included studies (I2 = 99.47%, Q = 4,891.648, P < 0.001). Nine studies reported the prevalence of pre-social frailty, with estimates ranging from 31.5% to 54.4%. The random-effects model indicated a pooled prevalence of pre-social frailty of 42% (95% CI: 37.6%–46.3%), with significant heterogeneity also observed among studies (I2 = 94.80%, Q = 64.705, P < 0.001). Subgroup analysis To identify sources of heterogeneity, subgroup analyses were conducted based on sample size, gender, study design, region of sample origin, Chinese region, publication year, study quality, and diagnostic criteria for social frailty; the results are shown in Table 2. Additionally, meta-regression analyses were performed (Table 3). Univariate and multivariate meta-regression analyses indicated that publication year (β = 0.167, 95% CI = 0.034–0.300, P = 0.016) and diagnostic tool (β = 0.273, 95% CI = 0.018–0.527, P = 0.037) were significantly associated with the prevalence of social frailty; the remaining factors did not show significant associations. Table 2Subgroup analysisSubgroupNumber of studiesPrevalence of social frailty95% CII2P-valueGender Male1022.615.1–30.199.4< 0.001 Female1022.714.9–30.599.5< 0.001Study Design Cross-sectional study2029.023.1–34.999.4< 0.001 Cohorte study414.07.4–20.698.2< 0.001Social Frailty Diagnostic Tool Specific Social Frailty Scale2224.219.6–28.999.5< 0.001 Other scales252.447.6–57.3< 0.001< 0.001Quality of the literature Low risk of bias722.514.9–30.298.9< 0.001 Moderate risk of bias1728.121.5–34.799.4< 0.001Sample size > 5001323.317.2–29.499.7< 0.001 ≤ 5001130.621.1–40.097.9< 0.001Sample source region Community2025.621.1–30.199.0< 0.001 Rural1029.920.1–39.799.78< 0.001ChinaCentral China619.514.5–24.594.5< 0.001 East China923.117.1–29.099.4< 0.001 North China624.416.7–32.099.3< 0.001 Subgroup analysis Table 3Results of multiple regression analysisCovariatesβSE95% CIPPublication Date0.1670.0640.034–0.3000.016Study Design−0.0870.080−0.256—0.0810.290Diagnostic Tools0.2730.1220.018—0.5270.037Sample size−0.0350.063−0.167—0.0960.582 Results of multiple regression analysis Influencing factors associated with social frailty A total of 3 studies analyzed gender, 2 addressed depression, 2 focused on housing satisfaction, 2 examined age groups (≥ 85 years vs. 65–69 years), 3 analyzed marital status, and 2 studied physical frailty. The meta-analysis results revealed that gender, depression, housing satisfaction, age, marital status, and physical frailty are the primary influencing factors for social frailty among community-dwelling older adults; specific data are presented in Table 4. Table 4Results of the meta-analysis on influencing factors for social frailty among community-dwelling older adultsCorrelatesNumber of StudiesOR (95% CI)P-valueAnalytical ModelHeterogeneityp-valueI2Gender31.36 [1.30–1.43]< 0.001Fixed0.3504.9%Depression22.82 [1.93–4.10]< 0.001fixed0.7960%Housing Satisfaction22.13 [2.01–2.24]< 0.001fixed0.9900%Age (≥ 85 years vs 65–69 years)23.25 [1.69–6.26]< 0.001fixed0.8440%Marital Status34.32 [4.11–4.54]< 0.001random0.000280.0%Physical Weakness24.15 [1.20–14.39]0.025random0.00190.7% Results of the meta-analysis on influencing factors for social frailty among community-dwelling older adults Sensitivity analysis and publication bias To assess the robustness of the results, a sensitivity analysis was conducted by sequentially excluding individual studies (Fig. 2). The estimated overall prevalence of social frailty fluctuated between 23.4% and 26.3%, showing no significant difference from the overall estimate, indicating that the results are relatively stable. A funnel plot (Fig. 3) and Egger’s test were used to assess publication bias. When all 24 studies were initially included, the funnel plot exhibited some asymmetry, and the Egger’s test results indicated significant publication bias (t = − 2.46, p = 0.022). To further assess the potential impact of this bias on the pooled effect size, we conducted a trim-and-fill analysis. The results showed that correcting the funnel plot asymmetry required adding 7 studies; the adjusted pooled effect size (increasing from 0.161 to 1.165) suggested that the original analysis may have overestimated the effect due to the absence of negative results. Given that the sample size of the study by Qi X et al. (n = 222,179) far exceeded that of other studies and might have significantly influenced the overall bias assessment, we conducted a sensitivity analysis after excluding this study. A redrawn funnel plot and Egger’s test indicated that publication bias was no longer significant (t = 0.18, p = 0.857). Applying the trim-and-fill method again, the results showed that no additional studies needed to be included; the pooled effect size was 0.257 (95% CI: 0.243–0.271) for the fixed-effects model and 0.270 (95% CI: 0.195–0.344) for the random-effects model, which were close to the pooled effect sizes before the initial exclusion. These results indicate that the meta-analysis findings are highly sensitive to the large-scale study by Qi X et al., and the publication bias detected in the original analysis primarily stemmed from this study with an exceptionally large sample size. After excluding its excessive influence, the body of evidence comprising the remaining 23 studies did not show significant publication bias, and the pooled effect sizes remained stable, suggesting that the core conclusions possess a certain degree of robustness (Fig. 4). Forest plot of pooled prevalence Sensitivity Analysis Funnel plot Trimming method Discussion Prevalence of social frailty in older adults A meta-analysis of 24 original studies indicates that the pooled prevalence of social frailty among community-dwelling older adults in China is 25%. This suggests that social frailty is a significant health issue among the elderly. However, we found that the pooled estimates exhibited high heterogeneity (I2 = 99.47%, P < 0.001), indicating substantial differences across the included studies. This high heterogeneity directly affects the robustness and generalizability of our pooled estimate. Furthermore, our meta-analysis found that the combined prevalence of pre-social frailty among community-dwelling older adults in China was 42%, indicating that the population in the pre-stage is more extensive than those in a state of social frailty. This finding suggests that implementing screening and interventions for pre-social frailty among China’s community-dwelling older adults may provide a significant opportunity for effectively managing social frailty and improving health outcomes. However, we found that the pooled estimates exhibited high heterogeneity (I2 = 99.47%, P < 0.001), suggesting substantial differences in effect sizes among the included studies; therefore, the overall pooled estimate should be interpreted with caution. Therefore, we conducted a pre-specified subgroup analysis to elucidate the sources of heterogeneity and to more precisely characterize the distribution of social frailty across different subgroups. However, heterogeneity remained significant after subgroup analysis, and we ultimately chose a random-effects model to pool the prevalence estimates. We first hypothesized that differences in diagnostic tools were a key methodological factor contributing to heterogeneity. Due to the lack of a standardized diagnostic tool for social frailty, the included studies utilized various instruments such as HALFT, MSFI, SFI, SFS, TFI, and SVI. Based on the differing theoretical frameworks underlying these diagnostic tools, we grouped diagnostic tools based on the theory of social need fulfillment—including HALFT, MSFI, SFI, and SFS—into one category, while classifying the remaining instruments as “other scales.”Among these, specialized social frailty scales based on the theory of social need satisfaction were used in 22 studies, with a combined prevalence of 24.2%, while studies using other scales had a combined prevalence of 52.4%. The effect size differed significantly between the different scale types (p = 0.001). In-depth analysis revealed that in the group of other scales, such as the TFI scale, social frailty is defined based on three sub-indicators: living alone, lack of social relationships, and lack of social support; however, there are currently no clear criteria for determining social frailty [30]. The SVI scale is a diagnostic tool developed based on the deficit accumulation model; diagnostic tools developed from different theoretical models may affect the accuracy of prevalence estimates. Currently, the HALFT scale is widely used in China for assessing social frailty, whereas other scales may not be specifically designed for this purpose or may lack a Chinese version [31]. This suggests that in future research and clinical screening, diagnostic tools for social frailty must be carefully selected and standardized to facilitate the comparison and integration of results. Furthermore, it is essential to develop or validate a comprehensive diagnostic tool suitable for the Chinese context that integrates both objective and subjective dimensions and encompasses social participation and social support. Furthermore, our analysis revealed a significant difference between the pooled prevalence rates from cross-sectional studies (29%) and those from cohort studies (14%), with the difference between study types being statistically significant (p = 0.001). This indicates that study design influences prevalence results and may serve as an important source of heterogeneity. We attribute this phenomenon to the differing numbers of studies included in each group, which led to variations in prevalence estimates. Concurrently, our subgroup analysis based on study quality revealed no significant difference in pooled prevalence between the low-risk-of-bias and moderate-risk-of-bias groups (P = 0.277), suggesting that study quality is not a primary factor contributing to the high heterogeneity observed in this study. However, it is important to note that while the subgroup analysis based on population characteristics showed no significant effect of gender on the effect size (p = 0.982), it revealed differences with significant public health implications. Subgroup analysis by gender showed that the combined prevalence of social frailty among older women was 22.7% (95% CI: 0.15–0.31), while that among older men was 22.6% (95% CI: 0.15–0.30). This finding is consistent with the conclusions of the study by Gong Xiaoyan et al. [32]. A possible reason is that women are more likely and more frequently express emotional experiences than men, leading to higher levels of loneliness among older women compared to older men, which in turn increases the risk of social frailty [33]. We divided the sample into urban and rural subgroups to conduct an in-depth, multi-level, and multi-faceted exploration of social frailty among community-dwelling older adults in China. The analysis showed that the pooled prevalence across 10 rural-based studies was 29.9% (95% CI: 0.20–0.30), while the pooled prevalence across 20 community-based studies was 25.6% (95% CI: 0.21–0.30). The difference in effect sizes between urban and rural areas (p = 0.440) was not significant, indicating that geographic location was not a major source of heterogeneity. The prevalence of social frailty among rural older adults was higher than that among urban community-dwelling older adults, a finding consistent with the conclusions of Yu Shan’s study [16]. The cause of this difference may lie in the fact that, as the urban–rural dichotomy has intensified and young and middle-aged rural residents have migrated en masse to cities for work, rural older adults—due to poor health status, low levels of family support, unstable economic resources, and a lack of fixed pensions—may have further widened the gap between rural and urban older adults in terms of social participation or social integration. To investigate whether there are regional differences in the prevalence of social frailty, we conducted a subgroup analysis based on the study participants’ locations, dividing the study population into five subgroups: Central China, East China, North China, Southwest China, and other regions (primarily Northeast and Northwest China). The results indicate that there are differences in the pooled prevalence of social frailty across different regions of China. Among these, the “Other Regions” group had the highest overall prevalence at 32.1% (95% CI: 24.2%–40.0%), followed by North China at 24.4% (95% CI: 16.7%–32.0%) and East China at 23.1% (95% CI: 17.1%–29.0%). The pooled prevalence rates in Southwest and Central China were relatively lower, at 21.7% (95% CI: 17.3%–26.1%) and 19.5% (95% CI: 14.5%–24.5%), respectively. Our subgroup analysis generated the hypothesis that the prevalence of social frailty might vary by geographic region in China (e.g., potentially higher in the north). However, given the substantial unexplained heterogeneity and the limited number of studies in some regions, this finding is exploratory and should not be considered conclusive. Future research with standardized methodologies is needed to test this hypothesis. We believe that this regional disparity may be related to domestic labor migration trends, which are often driven by economic and educational factors. This migration has led to increasingly prominent aging issues in northern regions, directly weakening older adults’ social support networks and potentially exacerbating social frailty [34]. The limited number of studies included for certain regions (such as Southwest China and other areas), coupled with the lack of a nationally standardized social frailty assessment scale, may have affected the stability of the pooled results. Future research should focus on exploring the epidemiological characteristics and causes of social frailty among older adults in urban and rural areas and across different regions of China, with the aim of establishing unified assessment standards and comprehensively understanding its nationwide distribution. Results from a stratified analysis based on publication year indicate that the prevalence of social frailty in China shows an increasing trend over time. The overall prevalence of social frailty was 15.5% between 2018 and 2022, rising to 32% between 2023 and 2025; the effect size differed significantly across publication year groups (p = 0.001), suggesting that temporal trends may be a source of heterogeneity. Only 8 studies were conducted between 2020 and 2022, involving a total of 8,284 participants; in contrast, 16 studies were conducted between 2023 and 2025, involving a total of 234,307 participants. The COVID-19 pandemic severely restricted older adults'’ social participation [35], with significant reductions in both physical and social activities, which may have further exacerbated the incidence of social frailty. More importantly, the substantial increase in sample size and a deeper understanding of the concept of social frailty may account for the higher prevalence estimates observed recently. We also note that among the 8 studies conducted between 2018 and 2022, 4 had sample sizes of fewer than 500; among the 16 studies conducted between 2023 and 2025, 7 had sample sizes of fewer than 500. We believe that the COVID-19 pandemic may have limited the survey scale of early studies to some extent, which may also partially explain why the combined prevalence rates from 2023 to 2025 were higher than those from the previous period. Finally, even after conducting various subgroup analyses and meta-regressions, heterogeneity remained significant, suggesting the presence of unobserved confounding factors (such as differences in the definition of social frailty, sample representativeness, and levels of community services). This significant and not fully explained heterogeneity directly affects the robustness and generalizability of our pooled prevalence estimates. Given that the prevalence of social vulnerability reported across studies ranges from 3.5% to 65.4%, and considering differences in sample size, diagnostic tools, and study regions, the pooled estimate of 25.0% should be viewed as a statistical summary of highly diverse findings rather than an exact or universally applicable national rate. Therefore, this estimate is more suitable for generating hypotheses than for informing specific policy or clinical decisions. Therefore, this estimate is more suitable for generating hypotheses than for cautiously providing a basis for definitive policy or clinical decisions. Factors influencing social frailty in older adults In summary, the factors influencing social frailty among older adults in the community primarily include demographic, health-related, and psychosocial factors. At the health level, physical frailty (OR = 4.15) and depression (OR = 2.82) showed the strongest associations. Physical frailty is significantly associated with social frailty, a finding consistent with previous studies [36]. Furthermore, research indicates that there is a bidirectional relationship between social frailty and physical frailty, and that this relationship changes with age [37]. A longitudinal study in Japan [38] noted that symptoms of physical frailty predict the future onset of social frailty; in particular, gait stability and muscle strength are considered important independent influencing factors for future social frailty. Maintaining and improving physical function through exercise, particularly balance and muscle strength, can serve as a key intervention strategy for addressing social frailty. As a psychological factor, we found that depression is significantly associated with social frailty, consistent with previous research findings [39]. The negative emotions triggered by depression can affect older adults’ willingness and ability to participate in social activities, potentially reducing their social engagement and, consequently, the social support they receive. Our study also indicates that demographic factors such as gender, marital status, and age are closely associated with social frailty. Older women face a 1.36-fold higher risk of social frailty compared to men, consistent with findings by Hyuma Makizako [40] and others. On the one hand, this may be attributed to physical symptoms (e.g., joint pain, insomnia, fatigue) and psychological symptoms (e.g., anxiety, depression) that emerge after menopause, which may limit older women’s capacity for social participation and lead to social frailty [41]. On the other hand, in terms of the division of household roles, older women typically shoulder more family caregiving responsibilities, which objectively consumes their time and energy that could be devoted to broader social interactions, leading to a gradual inward contraction and homogenization of their social support networks [17]. Future interventions could focus on the mental health and social participation of older women, encouraging them to maintain adequate time and space for socializing to reduce the risk of social frailty. Furthermore, our study found that compared to community-dwelling older adults with spouses, those without spouses (including those who are divorced or widowed) face a 4.32-fold higher risk of social frailty. This aligns with the findings of Yang Xiaojiao et al. [42]. The underlying reason is that divorce or widowhood may lead to a reduction in social connections and a weakening of social support networks, thereby triggering or exacerbating feelings of loneliness. As an important emotional bond, the absence of a spouse may intensify older adults’ experience of loneliness and reduce the frequency and depth of their daily social interactions. Although our study found that older adults aged ≥ 85 years had a 3.25-fold higher risk of social frailty compared to those aged 65–69 years, and this difference was statistically significant—a finding similar to that of Makizako et al. [43] —it cannot be denied that social frailty scores tend to increase with age. This may be due to the gradual decline in cognitive function with age, leading older adults to actively or passively reduce their participation in family activities and social interactions, thereby significantly increasing their risk of social isolation [44]. Regarding housing satisfaction as a psychosocial factor, our analysis also found that older adults dissatisfied with their living environment face a 2.13-fold increased risk of social frailty compared to those who are satisfied. We believe that since the primary sphere of activity for Chinese older adults after retirement is the home environment, the importance of their living conditions is particularly pronounced. Therefore, when selecting or planning housing for older adults, it is essential to prioritize age-friendly modifications to interior furnishings. In summary, social frailty is the result of the combined effects of physical decline, psychological state, living environment, and macro-social structures, which aligns with Bunt’s core concept of the multidimensional nature of frailty. This also suggests that future intervention strategies must adopt a multi-level, integrated approach. Limitations We also acknowledge the following limitations of this study: First, the meta-analysis of prevalence rates exhibited high heterogeneity. Although we conducted pre-specified subgroup analyses and meta-regression, and identified diagnostic tools, publication year, and study design as important moderating variables, some sources of heterogeneity remain unexplained. This heterogeneity affects the robustness and generalizability of the pooled estimates. Therefore, the pooled prevalence of 25.0% reported here should be regarded as the midpoint of a range of comprehensive estimates based on the available evidence, rather than an absolute precision. Second, meta-analyses of associated factors were limited by the data reported in the original studies, resulting in our ability to pool only a few factors. Furthermore, heterogeneity among studies regarding certain factors (such as marital status and physical frailty) remained high, suggesting that the relationship between these factors and the outcome may vary across different populations. Third, the absence of gray literature may lead to publication bias; although we verified the robustness of the results using the trim-and-fill method, we cannot completely rule out its impact on the overall effect estimate. Conclusion This systematic review indicates that pre-social frailty and social frailty are notable concerns among community-dwelling older adults in China. Crucially, due to the very high heterogeneity across studies, the pooled prevalence estimates should be considered as a range of possible values rather than a precise figure, the exact prevalence requires further validation through high-quality research. The main influencing factors for social frailty include advanced age, female gender, being unmarried, depressive symptoms, dissatisfaction with housing, and coexisting physical frailty. Existing studies suggest that social support interventions and combined exercise and nutritional interventions may play a positive role in improving social frailty; future high-quality intervention studies should be conducted in Chinese populations to validate these findings [45]. In summary, public health and clinical practice should prioritize the systematic assessment and early management of social frailty among community-dwelling older adults. Efforts should focus on developing or validating social frailty diagnostic tools suitable for the Chinese elderly population, actively promoting relevant intervention studies and tracking long-term outcomes, and gradually establishing a scientific and effective intervention management system to enhance the overall quality of life for older adults. Supplementary Information Supplementary Material 1. Authors’ contributions Jinwei Tang was responsible for the primary drafting of the manuscript, data extraction, and data analysis. Huiping Xue contributed to manuscript revisions, partial data analysis, and proofreading. Ling Wu and Chun Gao provided innovative insights for the paper and prepared Figs. 1, 2, 3 and 4. Huiping Xue and Chen shen offered guidance on the manuscript and conducted a thorough review. All authors contributed to the paper and approved the submitted version. Funding Funded by Nantong Municipal Science and Technology Program Project (Grant Number: MS2202 2063). Data availability All data included in this analysis were extracted from publicly published studies. Detailed references for the incidence rates, effect sizes of risk factors, and their 95% confidence intervals from each original study are fully listed in the tables within the main text and in the appendices. Declarations Ethics approval and consent to participate Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary Material 1. Data Availability Statement All data included in this analysis were extracted from publicly published studies. Detailed references for the incidence rates, effect sizes of risk factors, and their 95% confidence intervals from each original study are fully listed in the tables within the main text and in the appendices.