Association between oral frailty and physical frailty among older adults: a meta-analysis

Abstract

Background

This meta-analysis aimed to assess whether oral frailty could be associated with higher odds of physical frailty among older adults.

Methods

A systematic literature search of observational studies was conducted across PubMed, Web of Science, and Embase from database inception until July 3, 2025. Literature screening, data extraction, and quality assessment were performed independently by two reviewers. Any discrepancies were resolved through discussion or by consultation with a third reviewer. Meta-analysis was performed using STATA 18.0 to calculate pooled odds ratios (ORs) with 95% confidence intervals (CIs) and P interactions. Subgroup analyses were used to explore sources of heterogeneity.

Results

Fourteen studies were ultimately included in this meta-analysis. The meta-analysis showed that the odds of physical frailty in people with oral frailty was 3.11 times as high as that in those without oral frailty (95%CI: 1.89 to 5.10, P < 0.00001). Subgroup analyses revealed age, region, population type, assessment method, and study design as sources of heterogeneity. Sensitivity analysis showed that the stability of the results. P interactions also revealed interactions between population type and age.

Conclusion

This meta-analysis showed that oral frailty may be positively associated with physical frailty. Early identification of oral frailty could help prevent or delay physical frailty onset through timely interventions.

Trial registration

CRD420251183292.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12903-026-08899-y.

Introduction

Physical frailty (PF), characterized by multisystem physiological deterioration, heightens susceptibility to adverse outcomes such as falls, disability, hospitalization, and mortality when facing stressors including acute illness, surgery, or environmental challenges [1]. For instance, frail older adults face a 1.8-fold increased risk of falls, a 2.5-fold increased risk of disability, and a 1.5-fold increased risk of mortality compared with their robust counterparts [2–4]. A systematic review and meta-analysis of reports across 62 countries suggested a pooled prevalence of 12% of PF in population-based studies by using physical phenotype measures [5]. Studies showed that PF-associated healthcare costs impose severe financial burdens on the families and society [6].

Associated factors for the onset or progression of PF encompass multidimensional domains, including [2]physical factors such as reduced muscle strength and mobility limitations, mental health aspects such as cognitive dysfunction and depressive symptoms, social determinants characterized by isolation and diminished social connections, and nutritional factors involving insufficient protein intake and micronutrient deficiencies [7]. Early identification and treatment of these associated factors may prevent PF onset and progression. However, significant research gaps still remained on this topic, prompting researchers to look for other associated factors that can be intervened in.

There is growing interest in the concept of oral frailty (OF) as a potential precursor or contributor to physical frailty. OF is characterized by the progressive accumulation of minor declines in oral function, including tooth loss, masticatory dysfunction, and diminished communication capacity [8]. Beyond the inherent decline in oral function, OF exacerbated adverse health consequences, such as nutritional and metabolic problems, psychosocial dysfunction, cognitive decline and deterioration of overall health [8]. Evidence showed that OF triggers these negative consequences through three primary mechanisms: (1) physical/nutritional pathway: Tooth loss and weakened masticatory muscles impair chewing function, potentially leading to nutritional deficiencies [9, 10]; (2) psychosocial pathway: OF accompanied by halitosis and tooth loss negatively impact interpersonal communication and facial aesthetics, increasing the risk of social isolation [9]; (3) neurological pathway: Masticatory dysfunction may reduce cerebral blood flow, potentially contributing to neuronal loss, and further leading to a decrease in the number of neurons and consequent cognitive impairment [11].

Although OF can induce nutritional compromise, cognitive decline, and psychosocial impairment—cardinal manifestations of PF—the relationship between OF and PF remains inconclusive. Some studies supported the concept that OF was a associated factor for PF, whereas other studies oppose it. Given the recent emergence of this research field, with most related studies published within the past five years or limited to longitudinal designs, few meta analyses investigated the relationship between OF and PF [12]. This meta-analysis aimed to evaluate whether oral frailty is a associated factor for physical frailty, with the goal of informing preventive strategies and improving health outcomes in older adults.

Method

This meta-analysis was executed in alignment with the Preferred Reporting-Items for Systematic Reviews and Meta-Analysis (PRISMA) statement and was registered with PROSPERO prior to the initiation of the preliminary exploration (registration No: CRD420251183292).

Inclusion and exclusion criteria

Given the varying screening scales used across studies to assess OF and PF, we adopted a conservative approach by including all observational studies (including cross-sectional, cohort, and case-control designs) that examined the association between OF and PF. In addition, only studies published in English were eligible. The exclusion criteria were as follows: (a) reviews, case reports, conference abstracts, protocols, comments, or conferences; (b) laboratory studies (c) participants aged below 60 years; (d) duplicate publications; and (e) studies for which full-text was unavailable or data were insufficient for extraction.

Search strategy

A comprehensive literature search was conducted in the PubMed, Web of Science, and Embase databases from inception to July 3, 2025. The search terms consisted of a combination of Medical Subject Headings (MeSH) and text words. The keywords used were: “older”, “adults”, “frailty”, and “oral”. The detailed search strategy for the PubMed database was: (“aged” OR “old” OR “older” OR “elder” OR “elderly”) AND (“people” OR “adults” OR “individuals”) AND (“frailty” OR “frail” OR “weakness”) AND “oral” (details were shown in supplementary Table S1).

Study selection and data extraction

The initial search results were imported into EndNote 21.4 for study selection. Duplicate publications were removed by automatically and manually. Two independent reviewers conducted a three-phase screening process: (1) title/abstract screening using eligibility criteria, (2) full-text assessment of potentially eligible studies, and (3) final inclusion determination, with discrepancies resolved through discussion or third-party adjudication. For the included studies, the reviewers performed standardized data extraction using a piloted form capturing: (a) study identification (author, year, and country); (b) participant characteristics (region, age, and participants’ health status ); (c) methodology (study design, sample size, and assessment tools); and (d) outcome data (odds ratios (ORs) with 95% confidence intervals (CIs)). If effect sizes (e.g., odds ratios) were not directly reported in the original studies, they would be recalculated from available raw data or the corresponding authors were contacted to obtain missing information. Inter-rater reliability was assessed using Cohen’s kappa statistic, with a κ value of 0.85 for study selection and 0.92 for data extraction, indicating excellent agreement. Any disagreements were resolved through discussion or consultation with a third reviewer.

Quality assessment

The quality of the included original studies was assessed independently by two evaluators. Three longitudinal studies were included in this meta-analysis; however, only baseline data were extracted. Consequently, these studies were analysed as cross-sectional investigations. All included studies underwent a quality assessment using the JBI Critical Appraisal Tools [13, 14], focusing on key domains: (a) clear definition of sample inclusion criteria; (b) detailed description of study subjects and settings; (c) valid and reliable measurement of exposures; (d) utilization of objective, standardized criteria for condition assessment; (e) robust measurement of outcomes with established methodologies; (f) appropriate application of statistical analyses; (g) comprehensive identification of potential confounding factors; and (h) explicit documentation of strategies implemented to address identified confounders. The items included in the questionnaire were answered as follows: yes, no, unclear, or not applicable. Studies were stratified into three quality tiers: high methodological quality (≥ 5 “yes” responses), moderate methodological quality (3–4 “yes” responses), or low methodological quality (0–2 “yes” responses) [15].

Statistical analysis

Recording and analysis of the study data were performed utilizing STATA 18.0. ORs were calculated through random-effects meta-analysis using raw data extracted from each study [16]. The precision of effect sizes was expressed as 95% CIs. Statistical heterogeneity across included studies was assessed using I^² statistics, and Galbraith and L’Abbe plots. Subgroup analyses were performed through stratification by age, population characteristics, assessment tools, study designs, and regions to identify potential sources of heterogeneity. P-interaction values were calculated to assess the significance of effect size variations across predefined subgroups, thereby determining whether observed heterogeneity could be attributed to these potential effect modifiers. Leave-one-out sensitivity analyses were used to determine the stability of pooled effect sizes through the sequential exclusion of individual studies. Publication bias was evaluated visually through funnel plots. A P value less than 0.05 was considered to indicate significance.

Results

Study results

The initial database search yielded a total of 2,967 articles. There were 1,408 studies identified as duplicates. During the initial title and abstract screening phase, a total of 1,538 records that did not meet our inclusion criteria were excluded. These excluded publications comprised included review articles (n = 233), case reports (n = 401), conference abstracts (n = 13), study protocols (n = 16), laboratory investigations (n = 87), non-English publications (n = 48), and studies unrelated to our research topic (n = 740). As a result, 21 studies were selected for a full-text evaluation. Four studies were excluded due to the absence of effect estimates (e.g., OR with 95% CI) or sufficient raw data (e.g., 2 × 2 contingency tables) to calculate the association between OF and PF [17–20]. The remaining 17 studies underwent further exclusion evaluation: two studies [21, 22] were excluded because they utilized data from the same cross-sectional survey at the same time point and region, and one study [23] was excluded because the summary data in table headers did not match the sum of the detailed data in the tables, and this inconsistency could not be resolved by contacting the authors. Consequently, 14 studies were ultimately included in this meta-analysis (details were shown in Fig. 1).

Fig. 1.

Flowchart of the selection of studies

Characteristics of the included studies

The included studies were published between 2018 and 2025, with sample sizes ranging from 111 to 11,374. Geographically, almost all of the studies were conducted in Asia: 8 were from Japan [9, 24–30], 5 were from China [31–35]. Only 1 study originated from Northern Europe—Finland [36]. In terms of OF assessment tools, 3 studies [24, 26, 28] employed the Oral Frailty Index-6 (OFI-6), 6 studies [25, 27, 30, 32, 34, 35] utilized the Oral Frailty Index-8 (OFI-8), 2 studies [9, 29] adopted the Oral Frailty 5-Item Checklist (OF-5), and 3 studies [31, 33, 36] adopted self-definition OF criteria. In terms of PF assessment tools, 3 studies [24, 31, 36] employed the Fried’s phenotype criteria, 4 studies [25, 27, 28, 30] utilized the Frailty Screening Index (FSI), 3 studies [9, 26, 29] adopted the Japanese version of the Cardiovascular Health Study criteria (J-CHS criteria), 2 studies [34, 35] utilized the FRAIL scale, and 2 studies [32, 33] adopted the SOF index. In terms of the types of study, 11 were cross-sectional studies [24–28, 31–36] and 3 were longitudinal studies [9, 29, 30]. In terms of population characteristics, the subjects of 11 studies [9, 24, 26–31, 33, 35, 36] were community populations and the subjects of 3 studies were special populations including a stroke population [34], type 2 diabetes mellitus population [25] and cognitive decline population [32]. Twelve studies reported that OF was positively correlated with PF [9, 24–30, 32, 34–36], while the remaining 2 studies [31, 33] reported nonsignificant associations (details were shown in Table 1).

Table 1.

The main characteristics of included studies

Study	Study type	OF criteria	PF criteria	Control	Territory	Population	Age	Sample size	OR (95%CI)
Fei et al. (2024) [31]	cross-sectional study	Self definition	Fried’s criteria	Non OF1	China	Community population	≥ 60	307	1.11[0.55, 2.26]
Hiltunen et al. (2021) [36]	cross-sectional study	Self definition	Fried’s criteria	Non OF1	Finland	Community population	≥ 65	349	0.07[0.04, 0.11]
Hironaka et al. (2020) [24]	cross-sectional study	OFI-6	Fried’s criteria	Non OF1	Japan	Community population	≥ 65	682	3.38[1.29, 8.85]
Ishii et al. (2022) [25]	cross-sectional study	OFI-8	FSI	Non OF1	Japan	Type 2 diabetes population	≥ 75	111	4.29 [1.02, 17.96]
Iwasaki et al. (2024) [9]	longitudinal study	OF-5	J-CHS criteria	Oral robust	Japan	Community population	≥ 65	1206	1.74[1.10, 2.77]
Komatsu et al. (2021) [26]	cross-sectional study	OFI-6	J-CHS criteria	Non OF1	Japan	Community population	≥ 65	380	3.20[1.69, 6.06]
Kuo et al. (2022) [32]	cross-sectional study	OFI-8	SOF index	Non OF1	China	Cognitive decline population	≥ 75	308	2.57[1.39, 4.76]
Lin et al. (2022) [33]	cross-sectional study	Self definition	SOF index	Non OF1	China	Community population	≥ 65	908	1.85[0.66, 5.18]
Ma et al. (2025) [34]	cross-sectional study	OFI-8	FRAIL scale	Non OF1	China	Stroke population	≥ 60	451	0.35[0.19, 0.65]
Maeda et al. (2024) [27]	cross-sectional study	OFI-8	FSI	Non OF1	Japan	Community population	≥ 65	1386	2.87[2.13, 3.85]
Nakagawa et al. (2024) [28]	cross-sectional study	OFI-6	FSI	Non OF1	Japan	Community population	≥ 75	2727	1.98[1.63, 2.39]
Tanaka et al. (2023) [29]	longitudinal study	OF-5	J-CHS criteria	Oral robust	Japan	Community population	≥ 65	2044	3.69[2.71, 5.02]
Wang et al. (2024) [35]	cross-sectional study	OFI-8	FRAIL scale	Non OF1	China	Community population	≥ 60	478	3.49[2.24, 5.44]
Watanabe et al. (2024) [30]	longitudinal study	OFI-8	FSI	Non OF1	Japan	Community population	≥ 65	11,374	2.36 [2.08, 2.67]

1: Oral robust + Pre-oral frail

Abbreviations: CI confidence interval, FSI The Frailty Screening Index, J-CHS Japanese version of the Cardiovascular Health Study, OF oral frail, OFI oral frail index, OR odds ratio, PF physical frail, SOF The Study of Osteoporotic Fractures

Quality assessment of the included studies

All included studies met the predefined high-quality threshold according to the JBI appraisal criteria. The quality scores were distributed as follows: eight studies [9, 24, 26, 27, 29, 34–36] achieved the maximum score (8 points), two studies [28, 30] scored 7 points, and four studies [25, 31–33] received 6 points. Quality deductions primarily resulted from three methodological limitations: (a) insufficient specification of sample inclusion/exclusion criteria (n = 1), (b) failure to identify potential confounding variables (n = 4), and (c) the absence of documented strategies to address confounding factors (n = 5) (details were shown in Fig. 2a and b). The detailed item-by-item quality assessment for each included study is presented in supplementary Table S2.

Fig. 2.

Quality assessment of the included studies. a Risk of bias graph. b Risk of bias summary

The association between OF and PF

The meta-analysis included 14 studies that examined the association between OF and PF in older adult. The results showed that the odds of PF in people with OF was 3.11 times as high as that in non-OF populations (95%CI: 1.89 to 5.10, P < 0.00001) (details were shown in Fig. 3). Substantial heterogeneity (I² = 95.88%) among studies was observed. Galbraith and L’Abbe plots showed that the outlier study [34] reported by Ma et al. was a potential source of heterogeneity(details were shown in Fig. 4a and b). After excluding this study [34], the OR value was 2.46 (95% CI: 2.05 to 2.94, P < 0.00001) and the I² value decreased to 59.67% (details were shown in Fig. 4c). The results remained stable.

Fig. 3.

Forest plot of the association between oral frailty and physical frailty

Fig. 4.

Assessment of heterogeneity across studies. a Galbraith plot. b L’Abbe plot. c Forest plot

A funnel plot revealed that five studies [9, 28, 30, 31, 34] fell outside the confidence interval region, suggesting the presence of potential publication bias (details were shown in Fig. 5). Nevertheless, this finding did not serve as a criterion for excluding these studies, and all five were retained in the subsequent pooled effect size analysis.

Fig. 5.

Assessment of publication bias

To further evaluate the robustness of the pooled results, a sensitivity analysis was conducted using the leave-one-out method. This procedure demonstrated that omitting any single study did not change the overall direction of the findings, with the pooled OR ranging from 2.46 to 3.30 across all iterations (details were shown in Fig. 6). These results confirmed that the primary findings of the meta-analysis were stable.

Fig. 6.

Result of sensitivity analysis

Subgroup analysis and multivariate meta-regression

Although no statistically significant between-group differences were observed across most subgroup categorizations, two trends were noted. When stratified by age, the OR value was 6.22 for participants older than 60 years, 2.57 for those older than 65 years, and 2.05 for those older than 75 years. Moreover, the I² values decreased from 97.91% to 50.73% and finally to 0.00% (details were shown in Fig. 7a). When grouped by population type, interestingly, the OR value was 9.63(95% CI: 1.33 to 69.61) in special populations and 2.42(95% CI: 1.99 to 2.94) in community populations (details were shown in Fig. 7b).

Fig. 7.

Forest plot of subgroup analysis (a) age and (b) populations

In term of heterogeneity, several points were noteworthy. Analysis by study type revealed persistently high heterogeneity within both subgroups, with I² values of 94.43% and 82.49%,. Studies conducted in Japan exhibited lower heterogeneity (I² =66.19%) than those from non-Japanese regions did (I² = 95.27%) (details were shown in supplementary table S3). The analysis stratified by assessment tools revealed significant variations. Within the subgroup analysis based on the “assessment of OF”, the OFI-8 assessment method exhibited the highest level of heterogeneity (I² = 97.69%) (details were shown in supplementary table S3). In the subgroup analysis based on the “assessment of PF”, the subgroup utilizing Fried’s criteria demonstrated the lowest heterogeneity (I² = 36.16%) (details were shown in supplementary table S3).

In univariate meta-regression analyses, population type exhibited a statistically significant association with effect size (P < 0.005) (details were shown in Table 2). Considerable residual heterogeneity remained (tau²=0.38). A multivariate meta-regression model incorporating population type, age, assessment tools, population types, study designs, and regions revealed a significant synergistic interaction between population type and age (β = -1.55, 95% CI: -2.23 to -0.87; P < 0.05). These covariates collectively explained 83.75% of the heterogeneity. No significant difference was detected in the regression coefficients of the other covariates.

Table 2.

The results of subgroup forest and meta-regression

Subgroup	Study number	Sample size	Category	OR	95%LCI	95%UCI	P> Q2	I2	P for subgroup difference	P interaction
Age	3	1858	≥ 60	6.22	0.60	64.06	0.00	97.91%	0.19	0.258
	8	27,809	≥ 65	2.57	2.09	3.15	0.08	50.73%
	3	4609	≥ 75	2.05	1.71	2.46	0.38	0.00%
OF criteria	3	5114	OFI-6	2.39	1.62	3.53	0.22	40.10%	0.26	0.416
	6	22,804	OFI-8	5.09	1.83	14.16	0.00	97.69%
	5	6358	Other criteria	2.00	1.28	3.11	0.00	69.02%
PF criteria	3	1530	FPC	1.76	1.05	2.97	0.18	36.16%	0.42	0.954
	4	24,749	FSI	2.34	1.96	2.79	0.13	50.36%
	3	4911	J-CHS	2.77	1.72	4.46	0.03	69.58%
	4	3086	Other criteria	5.69	1.15	28.02	0.00	96.07%
Population	11	32,936	Community population	2.42	1.99	2.94	0.01	66.62%	0.17	0.005
	3	1340	Special population	9.63	1.33	69.61	0.00	94.47%
Study type	11	11,300	Cross-sectional	3.33	1.76	6.31	0.00	94.43%	0.47	0.641
	3	22,976	Longitudinal study	2.53	1.71	3.74	0.01	82.49%
Territory	8	30,407	Japan	2.56	2.08	3.15	0.02	66.19%	0.58	0.654
	6	3869	Not Japan	3.58	1.11	11.52	0.00	95.27%

Abbreviations: FPC Fried’s phenotype criteria, FSI The Frailty Screening Index, J-CHS Japanese version of the Cardiovascular Health Study, LCI lower confidence interval, OF oral frailty, OR odds ratio, PF physical frailty, UCI upper confidence interval

Discussion

This meta-analysis, comprising 14 studies, compared with individuals without OF, individuals with OF were more than three times more likely to develop PF. The decreasing trend in the relationship between PF and OF across population types did not reach significance. An interaction between population type and age was also found. The interpretation of these findings, however, was complicated by substantial conceptual and methodological heterogeneity in the assessment of both OF and PF across the included studies.

The concept of OF was constantly evolving in the literature, and there was no consensus [37]. Despite the lack of conceptual consistency, OF was defined as the cumulative decline in oral functions, including mastication, swallowing, articulation, and aesthetics [8]. Among the various tools originating from these oral functions, three tools, OFI-6, OFI-8, and OFI-5, were mainly used in epidemiological studies. The OFI-6 combined two subjective reports with four objective clinical measures [22], largely enhancing its reliability and multidimensional nature. However, its reliance on specialized equipment substantially limited its clinical utility [26]. In marked contrast, the OFI-8 included eight self-report items covering oral functions, health behaviours and social participation [21]. While this assessment allowed for broad use, it was susceptible to social desirability bias and other response inaccuracies, limiting its reliability [38, 39]. The subgroup analysis also revealed that owing to the nature of the OFI-8, heterogeneity within the OFI-8 group remained notably high and there was no substantial reduction relative to the overall prestratification heterogeneity. The OFI-5, derived from the Kihon Checklist, shared the fundamental reliance of the OFI-8 on subjective reporting through five targeted questions [8]. Unlike the OFI-8, this assessment was confined to three core oral functions and did not encompass broader behavioural or social aspects [29]. These methodological differences directly contributed to heterogeneity among the studies.

The concept of PF was also continually evolving within the geriatric and medical literature [40]. PF was characterized by the accumulation of deficits, typically manifested through indicators such as weakness, slowness, low physical activity, exhaustion, and unintentional weight loss [26, 41].The assessment of PF typically utilized several established tools, such as the Fried’s phenotype criteria, the FSI, the J-CHS, and so forth. Fried’s phenotype criteria combines objective measures such as grip strength and walking speed with self-reported items. This method offered high reliability but required specialized equipment [1]. In contrast, the FSI encompassed a concise set of items that mirror core components of PF, such as mobility, weight loss, and exhaustion. It stood out as a brief and practical instrument designed for rapid assessment [42]. The J-CHS represented a hybrid approach, integrating brief physical tests with self-reports; this balanced efficiency with broader assessments [43]. The observed heterogeneity could be largely attributed to differences in methodological approaches across studies. The subgroup analysis revealed varying degrees of heterogeneity reduction across all groups, with the most pronounced decrease observed in studies utilizing Fried’s phenotype criteria.

In addition to heterogeneity caused by assessments, similar evidence was also observed in other subgroup analyses. For instance, stratification by geographic region revealed that studies from Japan exhibited the most substantial reduction in heterogeneity following stratification. This suggests that shared regional and cultural characteristics including genetic homogeneity, similar lifestyle practices, and comparable healthcare systems [44] may contribute to the homogeneity among the studies. Interestingly, a notable trend observed in the age-stratified analysis was that heterogeneity completely diminished to zero in the oldest age category. This suggested that the oldest-old population represents a selective survivor cohort, potentially possessing a more homogeneous physiological profile because of the attrition of more vulnerable individuals at younger ages [45].

After the potential sources of heterogeneity were explored, several clear and important trends in the OR value emerged in the subgroup analysis. First, when grouped by age, the association between OF and PF was strongest in younger older adults and diminished with advancing age. The presence of competing risks including chronic disease, and severe cognitive impairment in the oldest-old may diminish the relative contribution of OF to the overall PF phenotype [46]. Moreover, the aetiology of PF itself may differ across age groups—OF in younger older adults might be a primary driver of functional decline, whereas in the older adults, it may represent one component of a broader, multisystem deterioration [47, 48]. Secondly, grouped by population type, the association between OF and PF was stronger in special population than in community population. This may be because special populations were more vulnerable and faced with more health challenges. In these individuals, the presence of OF may act as a critical stressor that accelerates functional decline through nutritional and neurological pathways [7]. Furthermore, the disease burden in these populations may have a synergistic effect, in which oral health deterioration and physical decline mutually reinforce each other, amplifying the observed association [49]. Notably, subgroup analyses revealed higher odds ratios in special populations, including patients with stroke, diabetes, and cognitive decline. However, these estimates were accompanied by wide confidence intervals, indicating statistical instability, and were derived from a limited number of studies. Therefore, these findings should be interpreted with caution and considered hypothesis-generating rather than conclusive. Moreover, the generalizability of these results to the broader older adult population may also be limited, and further research with larger sample sizes is needed to confirm these associations.

Despite the implications of our findings, this study must be interpreted in the context of its limitations. First, the inclusion of data mainly from East Asia may limit the generalizability of the findings. Second, the lack of sex-specific subgroup analysis represents an important gap in understanding potential sex-based differences. Third, the assessment of certain outcomes relied solely on qualitative synthesis rather than quantitative approaches, which restricted deeper mechanistic exploration. Finally, the predominantly cross-sectional design of included studies precludes causal inference; longitudinal studies are needed to clarify the causal relationship between oral frailty and physical frailty.

Conclusions

This meta-analysis revealed that OF could be associated with higher odds of PF. These findings highlight the potential value of incorporating OF screening into routine geriatric assessments. Identifying older adults with OF may provide a critical opportunity for early intervention, informing preventive strategies to delay or avoid the onset of PF. Therefore, healthcare professionals and policymakers should prioritize OF in clinical practice and program planning, and develop targeted preventive measures to improve the quality of life and health status of older adults.

Supplementary Information

Supplementary Material 1. ^{(269.5KB, docx)}

Supplementary Material 2. ^{(67.1KB, docx)}

Abbreviations

FSI

Frailty Screening Index

JBI

Joanna Briggs Institute

J-CHS

Japanese version of the Cardiovascular Health Study criteria

OF

Oral Frailty

OFI-5

Oral Frailty 5-Item Checklist

OFI-6

Oral Frailty Index-6

OFI-8

Oral Frailty Index-8

PF

Physical Frailty

PRISMA

Preferred Reporting Items for Systematic Reviews and Meta-Analyses

Authors’ contributions

Q. Li, T. Zhu, H. Jiang, C. Liu, H. Guo and M. Du contributed to the study conception and design, literature search, data collection, and analysis. The first draft of the manuscript was written by Q. Li and T. Zhu and all authors commented on previous versions.

Funding

This research was supported by the National Natural Science Foundation of China (Grant No. 81771084 and Grant No.82301091).

Data availability

The data supporting this study’s findings are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

All authors consent for publication.

Competing interests

The authors declare no competing interests.