Collagen Supplementation for Skin and Musculoskeletal Health: An Umbrella Review of Meta-Analyses on Elasticity, Hydration, and Structural Outcomes

Abstract

Collagen is a protein with multiple roles within the human body, such as supporting fibroblast formation in the dermis, replacing dead skin cells, protecting organs, giving structure, strength and elasticity to the skin, and a primary role in blood clotting. The aim of the present study was to carry out an umbrella review with integrated meta-analyses to capture the breadth of health outcomes associated with collagen supplementation intake. This umbrella review of systematic reviews includes a search of the use of collagen in PubMed, Embase, Scopus, and Web of Science until March 2025. The effect size for each health outcome was calculated using standardized mean differences, relative risks, or odds ratios, along with their corresponding 95% CIs. Separate meta-analyses were conducted for each outcome and pooled effect sizes were calculated using the inverse variance method under a random-effects model. Meta-regression analyses were conducted to explore potential sources of heterogeneity, and the grading of evidence was carried out using the GRADE. Among 573 papers, 16 systematic reviews for a total of 113 RCTs and 7983 patients were included. In relation to skin, musculoskeletal health, and osteoarthritis conditions, collagen supplementation was consistently associated with favorable outcomes. Regarding oral health and cardiometabolic parameters, the impact of collagen supplementation yielded mixed results. Collagen supplementation demonstrates consistent and clinically meaningful benefits for dermal, bone, and muscular health.

Level of Evidence: 3 (Therapeutic)

Collagen is a structural protein made up of amino acids, and it is found in human connective tissues, including the skin, tendons, cartilage, and bones. It makes up ∼25% to 30% of all proteins in the human body.¹ There are 28 various types of collagens with the most common being Types I to IV, with Type I comprising >90% of the collagen in the human body.² Type I collagen is found in the bone, tendons, skin, artery walls, cornea, fibrocartilage, and teeth.³ Collagen has multiple roles within the human body, such as supporting fibroblast formation in the dermis, replacing dead skin cells, protecting of organs, giving structure, strength and elasticity to the skin, and a primary role in blood clotting. Importantly, as the human ages, levels of collagen are reduced through decreased synthesis, increased degradation, or both.⁴ Indeed, it has been suggested that the body's collagen production slows down over time, decreasing by 1% a year beginning in early adulthood.⁵ Moreover, studies have found that the female skin loses ∼30% of its collagen during the first 5 years of menopause.⁶ A decline in collagen can have several detrimental impacts on human health. For example, skin aging through reduced elasticity, dryness, and thinning; cartilage breakdown thus contributing to joint pain and potentially osteoarthritis; muscle weakness; bone loss and fractures; wound healing; digestive complications; and blood flow complications. Owing to the known aging reduction in collagen levels and the detrimental impact on multiple health outcomes, the general public has turned to collagen supplementation to combat this and improve multiple aspects of health. Indeed, the global collagen supplements market size was valued at USD 1.99 billion in 2021 and is expected to expand at a compound annual growth rate of 5.5% from 2022 to 2028.⁷

There has been a large increase in scientific publications in relation to the health benefits of collagen supplementation over recent years, with multiple systematic reviews and meta-analyses published in this area.^8-11 However, to date, no attempt has been made to collate, synthesize, and appraise this wide spanning literature into 1 informative report. Therefore, the aim of the present study is to carry out an umbrella review with integrated meta-analyses (ie, the syntheses and appraisal of existing systematic reviews with meta-analyses) to capture the breadth of health outcomes associated with collagen supplementation intake. Thus, this umbrella review with integrated meta-analyses aims to inform the public, clinicians, and the private industry on guidance for collagen supplementation in relation to potential health benefits.

METHODS

Protocol and Registration

This umbrella review, including systematic reviews with meta-analysis, was conducted following the recommendations of the Cochrane handbook for systematic literature reviews to carry out the screening and selection of studies and reported according to the updated 2020 Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA) guidelines.^12,13 The protocol is freely available in PROSPERO (CRD420251089971).

Participants, Intervention, Control, Outcomes, Study Design (PICOS) Question and Eligibility Criteria

Following the Participants, Intervention, Control, Outcomes, Study Design (PICOS) question, we included (1) human participants: any; (2) intervention: any collagen supplementation; (3) controls: placebo or no collagen supplementation; (4) outcomes: all health outcomes; and (5) study design: systematic reviews with meta-analysis. We excluded the following studies: (1) systematic review without meta-analysis (owing to having no meta-analytical findings), (2) narrative review (owing to having no meta-analytical findings), and (3) reviews published in any language other than English.

Information Sources and Search Strategies

For this umbrella review, several relevant bibliographic databases were comprehensively searched, including PubMed, Embase, Scopus, and Web of Science from database inception up to March 6, 2025.

The following search was used in PubMed: ((“collagen” [Title/Abstract] OR “collagen supplement*” [Title/Abstract] OR “collagen administration” [Title/Abstract] OR “collagen hydrolysate” [Title/Abstract] OR “collagen peptides” [Title/Abstract]) AND (“health” [Title/Abstract] OR “mental health” [Title/Abstract] OR “well-being” [Title/Abstract] OR “heart” [Title/Abstract] OR “myocardium” [Title/Abstract] OR “cardiac fibrosis” [Title/Abstract] OR “atherosclerosis” [Title/Abstract] OR “arterial stiffness” [Title/Abstract] OR “hypertension” [Title/Abstract] OR “endothelial dysfunction” [Title/Abstract] OR “coronary artery disease” [Title/Abstract] OR “stroke” [Title/Abstract] OR “thrombosis” [Title/Abstract] OR “metabolic syndrome” [Title/Abstract] OR “diabetes” [Title/Abstract] OR “insulin resistance” [Title/Abstract] OR “obesity” [Title/Abstract] OR “weight loss” [Title/Abstract] OR “fat metabolism” [Title/Abstract] OR “testosterone” [Title/Abstract] OR “oestrogen” [Title/Abstract] OR “menopause” [Title/Abstract] OR “andropause” [Title/Abstract] OR “hormone replacement therapy” [Title/Abstract] OR “lung” [Title/Abstract] OR “pulmonary fibrosis” [Title/Abstract] OR “COPD” [Title/Abstract] OR “asthma” [Title/Abstract] OR “gut” [Title/Abstract] OR “gut barrier function” [Title/Abstract] OR “leaky gut” [Title/Abstract] OR “microbiome” [Title/Abstract] OR “inflammatory bowel disease” [Title/Abstract] OR “IBD” [Title/Abstract] OR “Crohn's Disease” [Title/Abstract] OR “osteoarthritis” [Title/Abstract] OR “rheumatoid arthritis” [Title/Abstract] OR “joint health” [Title/Abstract] OR “musculoskeletal injury” [Title/Abstract] OR “knee injury” [Title/Abstract] OR “meniscal injury” [Title/Abstract] OR “tendon tear” [Title/Abstract] OR “cartilage regeneration” [Title/Abstract] OR “bone density” [Title/Abstract] OR “osteoporosis” [Title/Abstract] OR “myofascial tissue” [Title/Abstract] OR “fibromyalgia” [Title/Abstract] OR “depression” [Title/Abstract] OR “anxiety” [Title/Abstract] OR “stress” [Title/Abstract] OR “cortisol” [Title/Abstract] OR “sleep” [Title/Abstract] OR “fatigue” [Title/Abstract] OR “psychological well-being” [Title/Abstract] OR “wound healing” [Title/Abstract] OR “hair loss” [Title/Abstract]) AND (“review” [Title/Abstract] OR “systematic review” [Title/Abstract] OR “meta-analysis” [Title/Abstract])) AND (meta-analysis[Filter]).

The search was then adapted to the other databases.

Study Selection

The selections were independently carried out by 2 review authors (J.G. and D.P.), with consensus meetings to discuss the studies for which divergent selection decisions were made by the 2 review authors. A third senior member of the review team (L.S.) was involved, if necessary. The study selection process involved, first, a selection based on title and/or abstracts, then a selection of studies retrieved from this first step based on the full-text manuscripts by the same 2 authors.

Data Collection and Data Items

The following information was extracted from eligible full-text articles: first author name and year of publication, data on the characteristics of the population considered and controls, dosage, follow-up duration, and health outcomes. We extracted data in relation to estimates at the single study level and categorized this data as risk ratio (RR), odds ratio (OR), hazard ratio, mean difference, and standardized mean difference (SMD). Data were collected using a standardized Microsoft Excel data extraction form (Microsoft Corporation, Washington). Two authors (J.G. and D.P.) led the data extraction, and this was systematically double-checked by senior authors (L.S. and R.R.). Any errors identified in extraction by the senior authors were discussed with all reviewers in a consensus meeting and subsequently corrected.

Assessment of Risk of Bias

Initially a single author (J.G.) rated the methodological quality of the included systematic reviews using “A MeaSurement Tool to Assess systematic Reviews 2 (AMSTAR 2).” This tool ranks the quality of a meta-analysis, following 16 predefined items, into 1 of 4 categories that range from “critically low” to “high.”¹⁴ A second reviewer (D.P.) double-checked the evaluation. The most relevant aspects of AMSTAR 2 include:

• protocol registered before commencement of the review (Item 2);

• adequacy of the literature search (Item 4);

• justification for excluding individual studies (Item 7);

• risk of bias from individual studies being included in the review (Item 9);

• appropriateness of meta-analytical methods (Item 11);

• consideration of risk of bias when interpreting the results of the review (Item 13);

• assessment of presence and likely impact of publication bias (Item 15).

We categorized the overall AMSTAR 2 score as:

• High: No or 1 noncritical weakness: the systematic review provides an accurate and comprehensive summary of the results of the available studies that address the question of interest.

• Moderate: More than 1 noncritical weakness: the systematic review has more than 1 weakness but no critical flaws. It may provide an accurate summary of the results of the available studies that were included in the review.

• Low: One critical flaw with or without noncritical weaknesses: the review has a critical flaw and may not provide an accurate and comprehensive summary of the available studies that address the question of interest.

• Critically low: More than 1 critical flaw with or without noncritical weaknesses: the review has more than 1 critical flaw and should not be relied on to provide an accurate and comprehensive summary of the available studies.

Data Synthesis and Grading of the Evidence

We calculated the effect size for each individual study using SMDs, RRs, or ORs, along with their corresponding 95% CIs, based on prespecified changes in outcomes between intervention and control groups. We carried out separate meta-analyses for each outcome (eg, fat-free mass, tendon morphology), utilizing the Paule–Mandel estimator to estimate between-study heterogeneity (τ²).¹⁵ To pool effect sizes, we employed the inverse variance method under a random-effects model. The I² statistic was used to evaluate heterogeneity, which describes the percentage of total variation across studies because of heterogeneity rather than chance. To visualize both individual study results and overall estimates, forest plots were produced. Meta-regression analyses were conducted to explore potential sources of heterogeneity, including publication year, product used in the control group, intervention dosage, and duration (when applicable). All analyses were conducted using R software (version 4.3.0; R Core Team, Vienna, Austria) and RStudio (version 2023.03.1; Posit, Boston, MA), employing the “meta” package (functions “metagen,” “forest,” and “metareg”).¹⁶

To evaluate the evidence from meta-analyses, we implemented the GRADE (Grading of Recommendations, Assessment, Development, and Evaluation) assessment. Where possible, we included the GRADE reported by the authors of the meta-analyses. The GRADE framework considers important domains for the judgment of the certainty of the evidence. These domains include: study design, risk of bias, inconsistency, indirectness, imprecision, and other aspects, such as publication bias and the outcomes of interest.¹⁷ The certainty of the evidence is then categorized into the following domains: very low (the true effect is probably markedly different from the estimated effect), low (the true effect might be markedly different from the estimated effect), moderate (the true effect is probably close to the estimated effect), or high (there is a lot of confidence that the true effect is similar to the estimated effect).¹⁷ The results of data analysis were imported into the GRADEpro Guideline Development Tool (McMaster University, 2015; developed by Evidence Prime, Inc.).

RESULTS

Literature Search

As shown in Figure 1, among 573 papers initially screened, we evaluated 32 full texts. After excluding 16 full texts, mainly because they were narrative reviews or systematic reviews without meta-analysis, we finally included 16 systematic reviews with meta-analyses (Appendix A).

Figure 1.

PRISMA flow chart.

Main Findings of the Umbrella Review

The 16 systematic reviews included a total of 113 randomized controlled trials (RCTs) involving 7983 participants. The included studies assessed a wide range of health outcomes across 5 major domains. Collectively, our findings illustrate the multifaceted potential of collagen supplementation across distinct health domains, with several outcomes supported by moderate-to-high certainty of evidence. Meta-regression findings further inform the nuances of collagen efficacy, emphasizing the roles of dosage, duration, and study context in shaping outcomes (Tables 1-5). Detailed GRADE assessments can be found in Supplemental Table 3. Units of measurement and the measurement tools for all outcomes are described in Supplemental Table 1. The PRISMA checklist is provided in Appendix B.

Table 1.

Meta-regression Analyses for Muscle-Related Outcomes

			95% CI
Outcome	Factor	β	Lower	Upper	P-value
Fat-free mass	Publication year (per 1 year)	−.04	−0.16	0.08	.563
	Product used in control group
	Maltodextrin	Ref.
	Silicea	.78	−0.01	1.58	.053
	Dosage (per 1 g/day)	.08	−0.001	0.16	.053
	Duration (per 1 week)	−.39	−0.79	0.01	.053
Tendon morphology	Publication year (per 1 year)	−.86	−2.21	0.50	.216
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	−.02	−0.09	0.06	.633
	Duration (per 1 week)	−.09	−0.49	0.30	.638
Tendon mechanical properties	Publication year (per 1 year)	.76	−0.68	2.19	.302
	Control	NA	NA	NA	NA
	Dosage (per 1 g/day)	.05*	0.01*	0.09*	.020*
	Duration (per 1 week)	−.24*	−0.40*	−0.07*	.006*
Muscle architecture	Publication year (per 1 year)	.08	−0.04	0.20	.187
	Product used in control group
	Maltodextrin	Ref.
	Silicea	−.35	−0.81	0.11	.137
	Dosage (per 1 g/day)	−.03	−0.08	0.02	.173
	Duration (per 1 week)	.09	−0.03	0.21	.127
Maximal strength	Publication year (per 1 year)	−.05	−0.12	0.02	.147
	Product used in control group
	Maltodextrin	Ref.
	Silicea	.22	−0.10	0.54	.173
	Dosage (per 1 g/day)	.003	−0.03	0.04	.851
	Duration (per 1 week)	−.01	−0.07	0.04	.660
Maximal strength recovery (24 h)	Publication year (per 1 year)	.02	−0.20	0.25	.842
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	.0004	−0.10	0.11	.993
	Duration (per 1 week)	.01	−0.07	0.09	.749
Maximal strength recovery (48 h)	Publication year (per 1 year)	.11	−0.12	0.34	.340
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	−.52	−0.16	0.05	.337
	Duration (per 1 week)	02	−0.06	0.10	.639
Reactive strength recovery (24 h)	Publication year (per 1 year)	.05	−0.20	0.29	.700
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	−.04	−0.23	0.15	.700
	Duration (per 1 week)	.01	−0.07	0.10	.739
Reactive strength recovery (48 h)	Publication year (per 1 year)	.06	−0.26	0.39	.699
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	.02	−0.25	0.28	.897
	Duration (per 1 week)	.04	−0.04	0.12	.339
Immediate muscle soreness	Publication year (per 1 year)	−.03	−0.18	0.12	.694
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	−.005	−0.09	0.08	.913
	Duration (per 1 week)	.03	−0.05	0.11	.484
Muscle soreness (24 h)	Publication year (per 1 year)	.18	−0.06	0.12	.694
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	−.06	−0.17	0.05	.266
	Duration (per 1 week)	.04	−0.06	0.13	.431
Muscle soreness (48 h)	Publication year (per 1 year)	.03	−0.19	0.25	.785
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	−.05	−0.16	0.05	.342
	Duration (per 1 week)	.03	−0.10	0.17	.629

β, unstandardized beta coefficient; CI, confidence interval; NA, not applicable; P, statistical significance. *Indicates significant findings.

Table 5.

Meta-regression Analyses for Skin-Related Outcomes

			95% CI
Outcome	Factor	β	Lower	Upper	P-value
Adverse events	Publication year (per 1 year)	−.03	−0.17	0.12	.731
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	NA	NA	NA	NA
	Duration (per 1 week)	NA	NA	NA	NA
Area reduction	Publication year (per 1 year)	NA	NA	NA	NA
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	NA	NA	NA	NA
	Duration (per 1 week)	NA	NA	NA	NA
Complete wound healing	Publication year (per 1 year)	−.01	−0.11	0.08	.768
	Control	NA	NA	NA	NA
	Dosage (per 1 g/day)	NA	NA	NA	NA
	Duration (per 1 week)	NA	NA	NA	NA
Cutaneous elasticity	Publication year (per 1 year)	−.10	−1.02	0.81	.829
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	.01	−0.71	0.73	.978
	Duration (per 1 week)	.08	−0.66	0.82	.826
Cutaneous elasticity (arbitrary unit)	Publication year (per 1 year)	−.08	−0.71	0.56	.813
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	.01	−0.24	0.25	.950
	Duration (per 1 week)	−.26*	−0.42*	−0.11*	.001*
Cutaneous elasticity (MPa)	Publication year (per 1 year)	−.62	−3.14	1.91	.636
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	−1.00	−7.31	5.30	.755
	Duration (per 1 week)	NA	NA	NA	NA
Cutaneous hydratation (arbitrary unit)	Publication year (per 1 year)	NA	NA	NA	NA
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	NA	NA	NA	NA
	Duration (per 1 week)	NA	NA	NA	NA
Cutaneous hydratation (µm)	Publication year (per 1 year)	.25*	0.11*	0.39*	<.001*
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	.05	−0.12	0.22	.565
	Duration (per 1 week)	.11	−0.02	0.25	.084
Healing velocity	Publication year (per 1 year)	NA	NA	NA	NA
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	NA	NA	NA	NA
	Duration (per 1 week)	NA	NA	NA	NA
Adverse events (during healing)	Publication year (per 1 year)	−.05	−0.12	0.02	.145
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	NA	NA	NA	NA
	Duration (per 1 week)	NA	NA	NA	NA
Area reduction (%)	Publication year (per 1 year)	−.05	−0.18	0.06	.354
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	NA	NA	NA	NA
	Duration (per 1 week)	NA	NA	NA	NA
Wounds closed (%)	Publication year (per 1 year)	.04	−0.28	0.35	.822
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	NA	NA	NA	NA
	Duration (per 1 week)	NA	NA	NA	NA
Recurrence of ulceration	Publication year (per 1 year)	−.18	−0.67	0.30	.455
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	NA	NA	NA	NA
	Duration (per 1 week)	NA	NA	NA	NA
Skin elasticity	Publication year (per 1 year)	−.30*	−0.50*	−0.10*	.003*
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	.01	−0.17	0.18	.943
	Duration (per 1 week)	.06	−0.16	0.27	.611
Skin hydratation	Publication year (per 1 year)	−.08	−0.19	0.02	.123
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	.01	−0.07	0.09	.732
	Duration (per 1 week)	−.03	−0.14	0.07	.554
Skin hydratation (arbitrary unit)	Publication year (per 1 year)	NA	NA	NA	NA
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	NA	NA	NA	NA
	Duration (per 1 week)	NA	NA	NA	NA
Skin hydratation (µm)	Publication year (per 1 year)	.12	−0.15	0.38	.387
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	.05	−0.11	0.22	.508
	Duration (per 1 week)	.09	−0.08	0.25	.300
Skin hydratation (%)	Publication year (per 1 year)	NA	NA	NA	NA
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	NA	NA	NA	NA
	Duration (per 1 week)	NA	NA	NA	NA
Skin roughness	Publication year (per 1 year)	.14	−0.40	0.68	.606
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	.07	−0.27	0.41	.671
	Duration (per 1 week)	−.07	−0.35	0.21	.618
Wound healing rate	Publication year (per 1 year)	−.005	−0.03	0.02	.721
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	NA	NA	NA	NA
	Duration (per 1 week)	NA	NA	NA	NA
Wound relative reduction (%)	Publication year (per 1 year)	−1.69	−5.95	2.57	.438
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	NA	NA	NA	NA
	Duration (per 1 week)	NA	NA	NA	NA

β, unstandardized beta coefficient; CI, confidence interval; NA, not applicable; P, statistical significance. *Indicates significant findings.

Collagen Supplementation and Muscle Outcomes

In relation to musculoskeletal health, collagen supplementation was found to exert beneficial effects on various muscle-related parameters. It led to a moderate increase in fat-free mass (moderate certainty, SMD = 0.48; 95% CI, 0.21-0.75; 454 participants, 8 RCTs) and also improved muscle architecture (moderate certainty, SMD = 0.39; 95% CI, 0.15-0.63; 183 participants, 5 RCTs). A modest gain in maximal strength was observed (high certainty, SMD = 0.18; 95% CI, 0.02-0.34; 533 participants, 11 RCTs). Collagen intake appeared to enhance tendon morphology (low certainty, SMD = 0.65; 95% CI, 0.04-1.26; 127 participants, 4 RCTs). However, the intervention did not yield statistically significant changes in tendon mechanical properties (low certainty, SMD = 0.04; 95% CI, −0.57 to 0.64; 127 participants, 4 RCTs).

Regarding recovery-related outcomes, no significant effects were detected for maximal strength recovery at 24 h (low certainty, SMD=0.26; 95% CI, −0.09 to 0.61; 114 participants, 4 RCTs) or 48 h (low certainty, SMD = 0.30; 95% CI, −0.05 to 0.65; 114 participants, 4 RCTs). Similarly, reactive strength recovery at both time points showed no statistical significance (low certainty, SMD = 0.26; 95% CI, −0.09 to 0.61; 114 participants, 4 RCTs).

With respect to muscle soreness, immediate postexercise soreness was not significantly improved (low certainty, SMD = −0.21; 95% CI, −0.57 to 0.15; 104 participants, 4 RCTs), nor was soreness at 24 h (low certainty, SMD = −0.06; 95% CI, −0.45 to 0.33; 114 participants, 4 RCTs) or 48 h (low certainty, SMD = 0.10; 95% CI, −0.41 to 0.60; 122 participants, 5 RCTs). Meta-regression analyses suggested that higher daily dosages were significantly associated with improved tendon mechanical properties (β = .05, P = .020), whereas longer intervention durations appeared to diminish this effect (β = −.24, P = .006; Table 1).

Collagen Supplementation and Oral Health

Regarding oral health outcomes, the impact of collagen supplementation yielded mixed results. A modest yet statistically significant reduction in gingival thickness was identified (low certainty, SMD = −0.44; 95% CI, −0.71 to −0.17; 110 participants, 3 RCTs). Conversely, no statistically significant effects were observed for keratinized mucosa width (moderate certainty, SMD = −0.42; 95% CI, −1.05 to 0.20; 266 participants, 7 RCTs), probing depth (moderate certainty, SMD = −0.08; 95% CI, −0.21 to 0.05; 184 participants, 4 RCTs), or aesthetic satisfaction (very low certainty, SMD = 6.42; 95% CI, −6.32 to 19.16; 100 participants, 2 RCTs). Findings from meta-regression indicated that longer intervention durations were significantly correlated with reductions in keratinized mucosa width (β = −.04, P = .002; Table 2).

Table 2.

Meta-regression Analyses for Oral Health-Related Outcomes

			95% CI
Outcome	Factor	β	Lower	Upper	P-value
Keratinized mucosa width parameter	Publication year (per 1 year)	−.04	−0.25	0.17	.691
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	NA	NA	NA	NA
	Duration (per 1 week)	−.04*	−0.07*	−0.02*	.002*
Gingival thickness parameter	Publication year (per 1 year)	−.13	−0.72	0.46	.657
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	NA	NA	NA	NA
	Duration (per 1 week)	−.02	−0.12	0.08	.657
Probing depth parameter	Publication year (per 1 year)	−0.09	−0.22	0.04	.177
	Control	NA	NA	NA	NA
	Dosage (per 1 g/day)	NA	NA	NA	NA
	Duration (per 1 week)	−.07	−0.22	0.08	.340
Participant aesthetic satisfaction	Publication year (per 1 year)	NA	NA	NA	NA
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	NA	NA	NA	NA
	Duration (per 1 week)	NA	NA	NA	NA

β, unstandardized beta coefficient; CI, confidence interval; NA, not applicable; P, statistical significance. *Indicates significant findings.

Collagen Supplementation and Cardiometabolic Outcomes

In the cardiometabolic domain, collagen intake demonstrated varied efficacy across metabolic markers. Although fat-free mass (%) improved significantly (low certainty, SMD = 1.49; 95% CI, 0.57-2.42; 192 participants, 3 RCTs), and fat mass (kg) was reduced (moderate certainty, SMD = −1.21; 95% CI, −2.13 to −0.29; 254 participants, 4 RCTs), body fat percentage did not show a statistically significant change (moderate certainty, SMD = −0.55; 95% CI, −1.47 to 0.36; 329 participants, 5 RCTs). Other indicators such as fasting blood glucose (high certainty, SMD = −7.18; 95% CI, −19.60 to 5.25; 338 participants, 5 RCTs), lipid profiles (eg, low-density lipoprotein cholesterol, high-density lipoprotein cholesterol [HDL-C], total cholesterol, triglycerides [TGs]), and blood pressure showed wide CIs, many of which included the null value, indicating no significant effect. Meta-regression models showed that collagen dosage significantly influenced improvements in body fat mass (%) (β = −.14, P < .001), and fasting glucose (β = −2.66, P < .001), and TGs (β = −1.10, P < .001). Moreover, the year of publication emerged as a significant predictor, positively correlating with BMI (β = .62, P = .002) and inversely with HDL-C levels (β = −1.14, P = .002; Table 3).

Table 3.

Meta-Regression Analyses for Cardiometabolic Outcomes

			95% CI
Outcome	Factor	β	Lower	Upper	P-value
BMI	Publication year (per 1 year)	.62*	0.24*	1.00*	.002*
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	.18*	0.07*	0.30*	.002*
	Duration (per 1 week)	NA	NA	NA	NA
Body fat mass (%)	Publication year (per 1 year)	−.14	−0.60	0.33	.559
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	−.14*	−0.22*	−0.07*	<.001*
	Duration (per 1 week)	NA	NA	NA	NA
Body fat mass (kg)	Publication year (per 1 year)	.14	−0.40	0.68	.619
	Control	NA	NA	NA	NA
	Dosage (per 1 g/day)	.04	−0.12	0.18	.656
	Duration (per 1 week)	NA	NA	NA	NA
Fat-free mass (%)	Publication year (per 1 year)	−.10	−0.47	0.26	.573
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	NA	NA	NA	NA
	Duration (per 1 week)	NA	NA	NA	NA
Fat-free mass (kg)	Publication year (per 1 year)	−.01	−0.56	0.54	.979
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	.12	−0.06	0.30	.177
	Duration (per 1 week)	NA	NA	NA	NA
LDL-c	Publication year (per 1 year)	.20	−2.05	2.44	.866
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	−.91	−1.96	0.15	.092
	Duration (per 1 week)	NA	NA	NA	NA
HDL-C	Publication year (per 1 year)	−1.14*	−1.87*	−0.41*	.002
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	.53	−0.19	1.27	.148
	Duration (per 1 week)	NA	NA	NA	NA
TC	Publication year (per 1 year)	.36	−2.57	3.29	.810
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	−1.90*	−3.60*	−0.21*	.028*
	Duration (per 1 week)	10.08	−22.39	42.54	.543
TG	Publication year (per 1 year)	.62	−1.96	3.19	.640
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	−1.10*	−1.70*	−0.50*	<.001*
	Duration (per 1 week)	NA	NA	NA	NA
SBP	Publication year (per 1 year)	.78	−0.65	2.21	.283
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	−.64	−1.47	0.19	.129
	Duration (per 1 week)	NA	NA	NA	NA
DBP	Publication year (per 1 year)	−.14	−1.15	0.87	.786
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	.07	−0.58	0.72	.835
	Duration (per 1 week)	NA	NA	NA	NA
Fasting blood glucose	Publication year (per 1 year)	3.82*	2.60*	5.03*	<.001*
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	−2.66*	−3.05*	−2.27*	<.001*
	Duration (per 1 week)	−1.99	−50.45	46.46	.936
HbA1c	Publication year (per 1 year)	NA	NA	NA	NA
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	NA	NA	NA	NA
	Duration (per 1 week)	NA	NA	NA	NA

β, unstandardized beta coefficient; CI, confidence interval; DBP, diastolic blood pressure; HbA1c, glycated hemoglobin; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; NA, not applicable; P, statistical significance; SBP, systolic blood pressure; TC, total cholesterol; TG, triglycerides. *Indicates significant findings.

Collagen Supplementation and Osteoarthritis-Related Outcomes

In individuals with osteoarthritis, collagen supplementation was consistently associated with symptom relief. Significant reductions were observed in self-reported pain (high certainty, SMD = −0.35; 95% CI, −0.47 to −0.22; 2687 participants, 25 RCTs), visual analog scale (VAS) scores (high certainty, SMD = −10.13; 95% CI, −19.88 to −0.38; 572 participants, 7 RCTs), Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) total score (high certainty, SMD = −7.24; 95% CI, −10.24 to −4.23; 1322 participants, 18 RCTs), and WOMAC stiffness subscale (high certainty, SMD = −0.53; 95% CI, −0.83 to −0.23; 595 participants, 7 RCTs). However, WOMAC pain (SMD = −0.74; 95% CI, −1.61 to 0.14; 1086 participants, 10 RCTs) and functional limitation (SMD = −1.96; 95% CI, −5.87 to 1.94; 969 participants, 8 RCTs) outcomes were not statistically significant. Evidence from meta-regressions revealed that longer intervention durations significantly predicted improvements in VAS scores (β = 2.26, P < .001), WOMAC index (β = .45, P = .012), and functional outcomes (β = .29, P = .016; Table 4).

Table 4.

Meta-regression Analyses for Osteoarthritis-related Outcomes

			95% CI
Outcome	Factor	β	Lower	Upper	P-value
Pain reduction	Publication year (per 1 year)	−.004	−0.02	0.01	.631
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	.01	−0.02	0.04	.697
	Duration (per 1 week)	.0004	−0.01	0.01	.956
VAS (score)	Publication year (per 1 year)	1.04	−1.31	3.40	.385
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	−1.57	−4.91	1.78	.358
	Duration (per 1 week)	2.26*	1.76*	2.76*	<.001*
WOMAC (index)	Publication year (per 1 year)	−.33	−1.05	0.40	.374
	Control	NA	NA	NA	NA
	Dosage (per 1 g/day)	−.11	−1.07	0.84	.816
	Duration (per 1 week)	.45*	0.10*	0.80*	.012*
WOMAC (pain)	Publication year (per 1 year)	−.03	−0.18	0.13	.748
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	−.06	−0.27	0.14	.543
	Duration (per 1 week)	−.01	−0.08	0.06	.716
WOMAC (stiffness)	Publication year (per 1 year)	−.08*	−0.14*	−0.03*	.003*
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	.05	−0.002	0.10	.059
	Duration (per 1 week)	.006	−0.03	0.04	.696
WOMAC (functional limitation)	Publication year (per 1 year)	−.36	−1.02	0.30	.279
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	.55	−0.31	1.41	.212
	Duration (per 1 week)	.29*	0.05*	0.53*	.016*
Lesquene index	Publication year (per 1 year)	NA	NA	NA	NA
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	NA	NA	NA	NA
	Duration (per 1 week)	NA	NA	NA	NA
Adverse events	Publication year (per 1 year)	−.08	−0.55	0.39	.727
	Product used in control group	NA	NA	NA	NA
	Dosage (per 1 g/day)	.24	−1.20	1.68	.741
	Duration (per 1 week)	.003	−0.17	0.18	.970

β, unstandardized beta coefficient; CI, confidence interval; NA, not applicable; P, statistical significance; VAS, visual analog scale; WOMAC, Western Ontario and McMaster Universities Arthritis Index. *Indicates significant findings.

Collagen Supplementation and Skin-Related Outcomes

Lastly, in the domain of dermatological health, collagen supplementation consistently yielded favorable outcomes. Substantial improvements were noted in skin elasticity (high certainty, SMD = 1.01; 95% CI, 0.44-1.59; 1217 participants, 20 RCTs) and hydration (high certainty, SMD = 0.71; 95% CI, 0.44-0.99; 954 participants, 19 RCTs). However, collagen did not significantly affect skin roughness (high certainty, SMD = −0.73; 95% CI, −1.55 to 0.09; 513 participants, 8 RCTs). Meta-regression analyses highlighted that cutaneous elasticity was significantly modulated by intervention duration (β = −.26, P = .001), and both elasticity and hydration outcomes were influenced by the year of publication (β = −.30 and β = .25, respectively, both P < .001; Table 5).

Assessment of Risk of Bias

Using the criteria suggested by the AMSTAR-2, among the 16 meta-analyses included, 1 reported a high quality, 4 low quality, and the others were rated as critically low (Supplemental Table 2). The most common potential sources of bias were the lack of protocol registration before the commencement of the review (Item 2), lack of justification for excluding individual studies (Item 7), and lack of proper discussion of publication bias (Item 15).

DISCUSSION

This is the most comprehensive umbrella review performed to date, underscoring that collagen supplementation exerts a broad spectrum of effects across multiple health domains. In synthesizing evidence from 16 meta-analyses (113 RCTs, ∼8000 participants), we found collagen supplementation showed notable benefits in several domains, although the magnitude and certainty varied.

In individuals with osteoarthritis, collagen supplementation was consistently associated with symptom relief. Our umbrella review identified significant improvements in self-reported pain, VAS scores, total WOMAC score, and WOMAC stiffness, all supported by high-certainty evidence. Although WOMAC pain and functional limitation subscales were not statistically significant, meta-regression demonstrated that longer intervention durations were positively associated with improvements in VAS, WOMAC index, and function. These findings suggest that collagen supplementation exerts a potentially clinically relevant, symptom-modifying effect in osteoarthritis, likely through stimulation of extracellular matrix synthesis and inhibition of inflammatory cartilage degradation.¹⁸ The consistency of benefit across multiple validated endpoints, coupled with an excellent safety profile, positions collagen as a compelling adjunctive strategy in the nonpharmacological management of joint disease.

The findings observed suggest that collagen supplements may promote a measurable restructuring of the extracellular matrix. This supports the “inside-out” model of skin rejuvenation, in which nutraceuticals act not as superficial cosmetic aids but as agents of deeper tissue regeneration.¹⁹ The duration-dependent improvements in elasticity and hydration reinforce the notion that collagen's effects accumulate gradually, requiring ongoing intake. Notably, these findings come from high-certainty data and reflect modern study quality improvements, as shown by the year-dependent modulation. That said, the absence of statistically significant changes in skin roughness highlights an important boundary: collagen likely improves turgor, tone, and moisture rather than resolving textural surface features. From a public health perspective, this distinction matters, collagen may be best positioned not as an anti-wrinkle “quick fix,” but as a foundational dermal support for individuals seeking holistic skin maintenance. Given its safety profile and noninvasive nature, clinicians may increasingly consider collagen a legitimate adjunct for skin aging, particularly among postmenopausal or photodamaged patients for whom traditional interventions are less suitable or cost prohibitive.

Although collagen is not considered a direct modulator of cardiometabolic pathways, our findings suggest it may contribute indirectly through favorable shifts in body composition—specifically, reductions in fat mass (moderate certainty) and increases in fat-free mass (low certainty). These changes may reflect improved musculoskeletal integrity rather than primary metabolic effects.²⁰ Meta-regression identified significant dose-dependent associations with improved body fat percentage, fasting glucose, and TGs, although traditional cardiometabolic markers—lipids, glucose, and blood pressure—showed limited or inconsistent effects. Year of publication significantly predicted changes in BMI and HDL-C levels, suggesting a positive evolution in study design, population selection, or the robustness of collagen formulations over time—each of which may have contributed to more pronounced or reliable cardiometabolic findings in recent trials. Overall, collagen may support metabolic health indirectly through improvements in lean mass and adiposity, particularly in sarcopenic or sedentary populations, but should not yet be positioned as a core intervention for cardiometabolic disease without further trial data on long-term outcomes.

Collagen supplementation was associated with modest yet statistically significant improvements in muscle health. Notable gains were observed in fat-free mass and muscle architecture, with a consistent effect on maximal strength. Improvements in tendon morphology were also observed, suggesting possible benefit to the extracellular matrix supporting muscle function. Meta-regression analysis showed a positive association between higher daily doses and improved tendon mechanical outcomes, whereas longer durations were paradoxically associated with reduced mechanical effect, warranting further investigation. However, no significant effects were identified for tendon mechanical properties, nor for strength recovery or postexercise soreness at 24 to 48 h (all low certainty). These findings suggest collagen's impact is likely chronic and structural rather than acutely ergogenic. Taken together, these results indicate that collagen may contribute meaningfully to musculoskeletal morphology, particularly in settings of aging or exercise, by enhancing connective tissue integrity and lean mass preservation, although not necessarily acute performance recovery.

Collagen supplementation showed limited and inconsistent benefits in oral health. A modest reduction in gingival thickness was observed, but no significant improvements were seen in keratinized mucosa width, probing depth, or aesthetic satisfaction. Meta-regression suggested that longer durations were associated with reductions in keratinized mucosa width, although the clinical relevance of this remains uncertain. Overall, current evidence does not support a robust role for collagen in oral or periodontal health.

Taken collectively, these findings position collagen supplementation as a promising adjunctive intervention across several domains of age-related health, with particularly strong evidence in skin elasticity and osteoarthritis symptom relief. The mechanistic plausibility, favorable safety profile, and growing consistency of trial outcomes underscore its translational potential. Although the effects on cardiometabolic outcomes are less definitive, they suggest a trajectory worthy of further exploration, particularly in older or at-risk populations. These benefits appear duration-dependent across several domains, reinforcing the importance of sustained compliance and long-term adherence to realize meaningful, cumulative effects of collagen supplementation. Given the scale and scope of this umbrella review, the most comprehensive to date, there is now a compelling rationale for clinical guideline bodies, research councils, and public health institutions to reappraise the role of collagen in preventive health strategies. Specifically, we recommend that (1) national dietary supplement frameworks begin formal evaluation of collagen's place in age-related care; (2) funding bodies support long-term, morbidity- and mortality-focused trials of collagen in joint, skin, and metabolic domains; and (3) regulatory clarity be established regarding collagen product quality, dosing, and formulation. Collagen is not a panacea, but its reproducible benefits, affordability, and tolerability mark it as a credible contributor to future models of integrative healthy aging.

Strengths and Limitations

This umbrella review represents the most comprehensive synthesis to date of collagen supplementation across multiple health domains, incorporating data from 113 RCTs and nearly 8000 participants. Its principal strength lies in scope and methodological rigor. By aggregating evidence across musculoskeletal, dermatological, metabolic, oral, and joint outcomes, we offer an integrated perspective for clinical and public health application. The use of AMSTAR-2 for methodological appraisal, meta-regression for evaluating dose and duration effects, and GRADE for grading evidence strength ensures a robust analytical framework. Crucially, this review distinguishes areas where evidence is compelling, such as improvements in skin hydration and osteoarthritic pain, from those where enthusiasm currently outpaces data, including glycemic control and oral health.

However, important limitations remain. The certainty of evidence across domains was frequently moderate to low, reflecting heterogeneity in study designs, dosages, and outcome measures. Many included meta-analyses were rated low or critically low quality because of unregistered protocols, insufficient bias reporting, and risk of publication bias. Sample sizes in subdomains were often small, and most trials were short in duration. We thus encourage future studies on collagen supplementation and health outcomes to both increase sample size and extend follow-ups.

Notably, data on clinical endpoints such as morbidity, hospitalization, or mortality were absent, limiting conclusions about long-term or disease-modifying potential. Future research should investigate whether improvements in intermediate outcomes (such as joint pain, skin elasticity, or fat mass) translate into meaningful reductions in health risk or healthcare burden over time.

Additionally, further work is required to delineate different collagen sources (eg, bovine vs marine) or administration formats (eg, liquid vs powder), and these questions remain unanswered but are highly relevant questions for translational practice. Although meta-regression suggested that higher doses and longer durations may confer greater benefit, these findings require confirmation in adequately powered, head-to-head trials. Body composition may influence the efficacy of collagen supplementation, and future reviews should aim to explore this further. An additional limitation is that key lifestyle variables known to affect skin aging and collagen metabolism, such as ultraviolet exposure, smoking status, hydration, diet, and sleep quality, were not consistently reported across the included meta-analyses and could not therefore be controlled for. This constraint is inherent to umbrella reviews that synthesize secondary data rather than individual participant information. Although such factors may contribute to interindividual variability in response, their omission is unlikely to materially influence the overall direction or robustness of the pooled findings. Finally, it was not possible to investigate outcomes by age groups or explore age as a modifier, as well as menopause status among females, in the meta-regression, owing to insufficient data reported in the included studies. Future research should aim to better report such data for its inclusion in future reviews of this kind. Together, these limitations underscore the need for continued, high-quality research to clarify optimal dosing strategies, formulation choice, and clinical impact beyond symptomatic improvement.

CONCLUSIONS

This umbrella review consolidates the current scientific understanding of collagen supplementation and its role across multiple health domains. The present study found that collagen supplementation demonstrates consistent and clinically meaningful benefits for dermal, bone, and muscular health. However, the breadth of benefits is not universal. The absence of long-term outcomes, such as morbidity, healthcare utilization, or mortality, constrains our ability to fully assess the broader clinical impact. Nonetheless, the signal is clear: collagen, once relegated to cosmetic marketing, now shows potential as a legitimate adjunct in the prevention or management of age-related decline in connective tissue integrity.

Supplemental Material

This article contains supplemental material located online at doi.org/10.1093/asjof/ojag018.

Disclosures

The authors declared no potential conflicts of interest with respect to the research, authorship, and publication of this article.

Funding

The authors received no financial support for the research, authorship, and publication of this article.