Research Reviews
Hyaluronic Acid Injections and Skin Aging: Two Significant and Two Nonsignificant Outcomes in a 12-Trial Review
A systematic review included 12 randomized controlled trials of hyaluronic acid (HA) injections, but only 6 supplied enough data for quantitative meta-analysis. Hydration and radiance scores were statistically higher than control scores; elasticity and melanin index did not show statistically significant differences. The hydration finding became nonsignificant when one potential outlier was removed. All four outcomes need equal attention. These findings cannot establish a product ranking or a conclusion about safety.
- Hydration and radiance findings were statistically significant; elasticity and melanin index findings were not.
- The hydration result lost statistical significance after removal of a potential outlier. Radiance was assessed subjectively from photographs, rather than measured by an instrument.
- Each outcome drew on only 2–4 studies. Heterogeneity, or variation among study results, ranged from I² = 50.8% for radiance to 98% for hydration. These estimates cannot support product comparisons or a conclusion about safety.
Although the review covered 12 trials, only 6 contributed to the pooled results. Each endpoint drew on fewer trials still. A significant hydration or radiance estimate cannot stand in for the two nonsignificant results, particularly when the hydration estimate depended on one study.[1]
This is a review of published research for understanding its findings and limits. It does not recommend a particular product or procedure; individual decisions require medical assessment.
Zhou and Yu’s systematic review and meta-analysis appeared in Journal of Cosmetic Dermatology, 2025;24(1):e16760. Both authors list an affiliation with the Department of Dermatology at Sichuan Huamei Zixin Medical Aesthetic Hospital in Chengdu, China. They declared no conflicts of interest and no specific funding for the work.[1]
Key findings
- Twelve randomized controlled trials (RCTs) met the review criteria; only six had data the authors considered sufficient for meta-analysis.
- Searches of four databases through May 2024, plus hand searching, identified 1,526 records. After duplicate removal, 886 remained; 12 trials were ultimately included.
- A standardized mean difference (SMD) puts differences between continuous scores into standard-deviation units. An SMD of 0 means no difference between standardized group means.
- Hydration had an SMD of 1.34 (95% confidence interval [CI] 0.14–2.54); radiance had an SMD of 0.51 (0.22–0.80). Both findings were statistically significant.
- Elasticity had an SMD of 0.25 (95% CI −0.20–0.70); melanin index had an SMD of −1.74 (−4.89–1.41). Neither finding was statistically significant; both intervals included 0.
- Hydration heterogeneity was I² = 98%. Removing a potential outlier gave an SMD of 0.73 (95% CI −0.14–1.59; p = 0.10), a nonsignificant result.
- Each outcome drew on 2–4 trials. This evidence base differs in size from the separate Cochrane review of 65 botulinum toxin RCTs discussed in this research-review series.[2]
- This review did not compare particular products or provide a pooled adverse-event result alongside the four endpoints. Its findings cannot establish the effect or safety of one product.
Why the four endpoints need separate attention
Hydration, radiance, elasticity, and melanin index are different outcomes in research on skin aging. The meta-analysis did not find that all four changed in the same direction with statistical significance. Hydration and radiance were significant; elasticity and melanin index were not. Selecting just the positive findings would expand the conclusion beyond the evidence.[1]
An SMD needs its confidence interval, heterogeneity, and study count beside it. The hydration estimate of 1.34 had a wide interval and substantial variation between studies; removing a potential outlier eliminated statistical significance. Radiance had a narrower interval, but came from physicians’ subjective ratings of standardized photographs. The size, method, and uncertainty belong together.[1]
Scope and methods of the review
Search and selection
The review followed PRISMA reporting guidance and reports registration in PROSPERO as CRD42024540703. Searches covered PubMed, Embase, Scopus, and Web of Science through May 2024, with additional hand searching.[1]
After duplicate removal, 886 of the 1,526 records remained. The authors then screened titles and abstracts, assessed 160 full texts, and included 12 trials.[1]
Eligibility and quantitative synthesis
Eligible RCTs compared HA injection with saline, placebo, or no treatment and reported at least one outcome involving elasticity, hydration, wrinkles, or roughness.[1]
The six trials pooled were Roh 2016, Lim 2014, Fanian 2023, Duteil 2023, Baspeyras 2013, and Alexiades 2023. The paper says they had sufficient data, without providing an individual explanation for why each of the other six could not be pooled.[1]
How the four outcomes were measured
- Hydration was measured using MoistureMeter devices.
- Elasticity was measured using a Cutometer.
- Melanin index was measured using a Mexameter.
- Radiance was rated subjectively from standardized photographs after washing, by a physician or a blinded panel of physicians.
Instruments measured hydration, elasticity, and melanin index. Radiance came from subjective photograph ratings. The four endpoints should not be treated as the same kind of instrument reading.[1]
How results were combined
The authors calculated SMD using Cohen’s d and combined results with a random-effects model using maximum-likelihood estimation. They set statistical significance at p ≤ 0.05. For heterogeneity, they used p = 0.10 in Cochran’s Q test and also calculated I².[1]
The paper reports using a random-effects model without separately explaining why it was selected. Variation among the included studies should not be presented as the authors’ stated reason for that choice.[1]
How to read SMD
A standardized difference centered on zero
SMD expresses a difference between continuous group means in standard-deviation units. Zero means no difference between standardized means. When a 95% CI crosses 0, the analysis does not establish a statistically significant group difference.[1]
The endpoint determines how to read the SMD’s direction. The authors interpreted positive hydration and radiance SMDs as higher scores than controls and a negative melanin SMD as a direction toward lower values. The melanin interval crossed 0 widely, however, so it could not establish that direction.[1]
Rough benchmarks are not clinical thresholds
SMD values of 0.2, 0.5, and 0.8 are often used as rough small, medium, and large benchmarks. They are not this paper’s success thresholds or clinical standards applicable to every skin measurement.
SMD alone also cannot tell how much change an individual would notice. An estimate of 1.34 is not an assured large personal change, and 0.51 is not a change that everyone would perceive.
Four outcomes, given equal attention
Hydration: significant, but sensitive to a potential outlier
- Evidence: Four studies. The paper’s total and group participant counts are arithmetically inconsistent, so no hydration participant total is given here.
- Estimate: SMD 1.34; 95% CI 0.14–2.54; p < 0.05.
- Heterogeneity: I² = 98%; between-study variance τ² = 1.45.
- Interpretation: The interval was wide, with its lower bound of 0.14 close to 0. Removing the potential outlier gave SMD 0.73, 95% CI −0.14–1.59, and p = 0.10, which was not statistically significant.[1]
Radiance: significant, based on subjective ratings
- Evidence: Three studies, with 307 participants: 188 in intervention groups and 119 in control groups.
- Estimate: SMD 0.51; 95% CI 0.22–0.80; p < 0.05.
- Heterogeneity: I² = 50.8%; τ² = 0.02.
- Interpretation: The finding was significant and its interval relatively narrow among the four endpoints. It came from physicians’ subjective ratings of standardized photographs after washing, rather than an instrument measurement.[1]
Elasticity: no statistically significant difference
- Evidence: Three studies, with 299 participants: 184 in intervention groups and 115 in control groups.
- Estimate: SMD 0.25; 95% CI −0.20–0.70; p = 0.27.
- Heterogeneity: I² = 76%; τ² = 0.11.
- Interpretation: The interval included 0. This cannot be described as an established improvement, but it also does not prove that the groups have identical outcomes.[1]
Melanin index: a nonsignificant result with uncertain direction
- Evidence: Two studies, with 68 participants: 34 in each group.
- Estimate: SMD −1.74; 95% CI −4.89–1.41; p = 0.28.
- Heterogeneity: I² = 94.1%; τ² = 4.87.
- Interpretation: The very wide interval crossed 0. It cannot establish either a decrease or increase in melanin index, and does not prove the absence of pigment change.[1]
Adverse events reported by individual studies
The review provides no pooled adverse-event estimate alongside the four endpoints. Individual studies described their own events; some were among the six studies contributing to the endpoint analyses. Their reports are not a combined incidence and cannot be calculated across all participants or transferred to another study.[1]
- Baspeyras 2013, one of the six pooled trials, reported at least one adverse event after injection in 50 people (87.7%). Most were expected reactions of mild or moderate intensity, including hematoma, edema, papules, erythema, and other transient inflammation. Four people had severe hematomas.
- Alexiades 2023, the hydration study whose removal changed that result to nonsignificant, reported treatment-related adverse events in six participants. None led to discontinuation.
- Fanian 2023, one of the six pooled trials, described mild events that mostly resolved spontaneously. An accidental death was reported separately and judged unrelated to treatment; it was a report from this trial, not the other included studies.
- Li 2022, among the 12 reviewed trials but outside the six pooled trials, was listed in Table 1 as reporting one treatment-related serious adverse event, arterial ischemia, that resolved in 16 days.
- Jones 2013 was listed as reporting two people who needed treatment for adverse events and one person with firmness that remained unresolved.
- In their discussion, the review authors list redness, pain, tenderness, firmness, swelling, itching, bruising, bumps, and rare discoloration among possible HA-filler reactions, and stress careful administration and follow-up.[1]
These are separate study reports, not rates representing all six or all twelve trials. They cannot establish that risk is negligible or that an event is common across the review. Seek medical assessment for unexpected discomfort that persists after an injection.
The authors’ limitations and how to assess the review
Limitations stated by the authors
The authors said few studies qualified because eligibility required a control arm, including saline or placebo. They left long-term efficacy for future research. That is different from a finding that every included study had short follow-up.[1]
Bias assessment and variation among studies
The authors described overall risk of bias as less than 50% in every study, with none at 0%. This assessment does not mean bias was absent.[1]
The studies differed in design, HA molecular weight and density, cross-linking, lidocaine content, and saline dilution. These differences prevent reducing the pooled findings to one material or injection approach.[1]
Why these are not product comparisons
The review combined studies of different HA injections rather than directly comparing particular products. It cannot establish superiority, equivalence, or preference, or assign the pooled result to one product.[1]
The abstract reports the significant hydration and radiance findings alongside the nonsignificant elasticity and melanin findings and calls for further research. Its conclusion also uses the phrase “reversing the effects of skin aging.”[1] The final conclusion repeats the nonsignificant elasticity finding but omits melanin index while suggesting a contribution to skin quality in appropriately selected patients, doses, and timing. The four-outcome pattern is a more accurate guide to the evidence than adopting that broad conclusion as a treatment claim.
Affiliation and disclosure are different information
Both authors list a medical aesthetic hospital in Chengdu, China, as their affiliation. They declare no conflicts of interest or specific funding. Affiliation describes context; the disclosures are the authors’ formal statements. Neither establishes that the work is unreliable nor removes its other limitations.[1]
Limits to keep in view
- Twelve trials were reviewed, six were pooled, and individual outcomes used only 2–4 studies. The scale cannot be directly equated with the separate Cochrane review of 65 RCTs.[2]
- Hydration was significant at SMD 1.34, with I² = 98% and a wide interval. After removal of the potential outlier, it was nonsignificant: SMD 0.73; 95% CI −0.14–1.59; p = 0.10.
- Heterogeneity differed across hydration (I² = 98%), radiance (50.8%), elasticity (76%), and melanin index (94.1%). Each estimate needs that variation beside it.
- Radiance and elasticity each used three trials; melanin index used two and only 68 participants, with a very wide CI. Radiance was based on subjective photograph ratings.
- Study design, material properties, cross-linking, and dilution varied. These results cannot establish the performance or relative ranking of specific products.
- With no pooled adverse-event estimate alongside the endpoints, the review cannot establish safety. Long-term efficacy also remained a question for future research.[1]
HA fillers are based on hyaluronic acid. Polynucleotide (PN) injectables belong to a different material class. “Skin booster” is an umbrella search and marketing term for various injectable products; “filler” broadly describes injectable tissue-filling materials. These definitions do not recommend a product or turn this review into a product comparison.
References
- Zhou R, Yu M. “The Effect of Local Hyaluronic Acid Injection on Skin Aging: A Systematic Review and Meta-Analysis.” Journal of Cosmetic Dermatology. 2025;24(1):e16760. DOI: 10.1111/jocd.16760. PMID: 39807700. PMCID: PMC11731322. https://pmc.ncbi.nlm.nih.gov/articles/PMC11731322/
- Camargo CP, Xia J, Costa CS, et al. “Botulinum toxin type A for facial wrinkles.” Cochrane Database of Systematic Reviews. 2021;2021(7):CD011301. DOI: 10.1002/14651858.CD011301.pub2. Cited here only for the evidence-base size comparison. https://pubmed.ncbi.nlm.nih.gov/34224576/
Frequently asked questions
How do a systematic review and a meta-analysis differ?
A systematic review searches for, selects, and assesses studies using predefined criteria. A meta-analysis quantitatively combines results from studies with data that can be pooled. This paper included 12 trials in its review but used 6 in the quantitative synthesis.
Why were only 6 of the 12 trials pooled?
The paper says only that 6 studies had sufficient data. It does not give a separate reason for excluding each of the other 6 from quantitative synthesis. Missing means or standard deviations should not be assumed as explanations.
What does SMD measure?
A standardized mean difference (SMD) expresses the difference between continuous scores in standard-deviation units. Zero means no difference between standardized group means. The often-used 0.2, 0.5, and 0.8 benchmarks are rough reference points, not established clinical thresholds. Confidence intervals, study count, and measurement methods also matter.
Does a nonsignificant result prove that there is no effect?
No. The 95% confidence intervals for elasticity and melanin index included 0, so this analysis did not clearly establish a group difference. The melanin interval was especially wide and could not establish a direction. A nonsignificant result is not proof that no effect exists.
What questions does this review raise for a consultation?
The trials used different materials and injection methods without comparing specific products. Clarify the proposed material and method, whether the intended endpoint is hydration, radiance, elasticity, or melanin index, how it would be assessed, and how uncertainty and adverse events would be explained. Group averages are not personal forecasts or Beyond Clinic's treatment results.