Erythritol versus Glycine Air Polishing in the Non-Surgical Management of Peri-Implantitis

Protocol for a systematic review and meta-analysis. Pre-registered, publicly timestamped, and open to methodological scrutiny before the evidence is collected.

Abstract

Background. Peri-implantitis affects an estimated 10–20% of patients with dental implants over 5–10 years. Air polishing with low-abrasion powders is a proposed non-surgical adjunct to mechanical debridement, but the comparative effectiveness of erythritol-based versus glycine-based powders is unclear. The last systematic review specifically addressing non-surgical peri-implantitis therapy (Schwarz et al., 2015) predates most of the erythritol evidence. European practice is further complicated by the EMS erythritol air-polishing-powder patent, which constrains the comparator landscape on the EU market.

Objectives. To systematically identify, appraise, and synthesize evidence from randomized and non-randomized controlled studies comparing erythritol and glycine air polishing as part of non-surgical peri-implantitis management, in adults diagnosed per the 2017 World Workshop case definition.

Methods. Electronic searches of PubMed/MEDLINE, Cochrane CENTRAL, LILACS, Google Scholar, ClinicalTrials.gov, EudraCT, ISRCTN, DRKS, and ChiCTR from 1 January 2011 to 31 March 2026, with hand-searches of key journals (J Clin Periodontol, Clin Oral Implants Res, J Periodontol) and EFP/Europerio/AAP/IADR conference abstracts. Eligible designs: RCTs (including split-mouth), controlled clinical trials, and prospective cohorts with a comparator arm. Primary outcomes: bleeding on probing, probing pocket depth, clinical attachment level, and adverse events. Risk of bias assessed with Cochrane RoB 2 for randomized studies and ROBINS-I for non-randomized studies. Synthesis: random-effects meta-analysis where at least three studies report comparable outcomes at comparable time horizons and between-study heterogeneity is acceptable (I² ≤ 75%); narrative synthesis otherwise. Pre-specified sensitivity analyses include restriction to low-risk-of-bias studies and restriction to non-industry-funded studies. Certainty of evidence rated with GRADE.

Dissemination. Full synthesis published open-access on preventiohub.com with downloadable PDF, reproducible analysis scripts, and a one-page clinician summary. Quarterly refresh committed. Findings will be communicated to the European Federation of Periodontology and relevant national bodies.

Registration. Preventio Hub Evidence Registry PH-SR-2026-001 — registry index. Timestamped via Internet Archive capture at publication.

Locked editorial decisions

The following decisions have been finalised before the search is executed. Any subsequent change constitutes a protocol amendment and will be logged publicly.

1. Background

1.1 The clinical problem

Peri-implantitis is a biofilm-associated inflammatory condition affecting the tissues around osseointegrated dental implants, characterised by bleeding and/or suppuration on probing, increased probing depths, and progressive marginal bone loss. The 2018 classification from the World Workshop on the Classification of Periodontal and Peri-Implant Diseases and Conditions established the current reference case definition: radiographic bone loss ≥3 mm from the implant shoulder and/or ≥25% of implant length, combined with clinical signs of inflammation.

Estimates of prevalence vary substantially, but cluster around 10–20% of patients and 1–10% of implants over 5–10 years of function. Consequences range from prolonged symptomatic inflammation to implant loss. Non-surgical management is the entry point for most cases — mechanical debridement, sometimes with adjunctive chemical or laser therapy, occasionally with antibiotic adjuncts.

1.2 Air polishing as an adjunct

Air polishing delivers a slurry of powder, water, and compressed air onto tooth and implant surfaces. In peri-implant applications, it is proposed to disrupt subgingival biofilm without mechanically damaging the implant surface — a theoretical advantage over titanium curettes. Powder chemistry matters: the original sodium bicarbonate powders are abrasive to root cementum and not recommended subgingivally; glycine powders (mean particle size ~25 µm) and erythritol powders (~14 µm) are low-abrasion alternatives marketed specifically for peri-implant use.

Erythritol powder has a smaller particle size and is frequently formulated with chlorhexidine (0.3% w/w). Proponents argue it produces superior biofilm removal and is less traumatic to peri-implant mucosa. Independent evidence is mixed.

1.3 The commercial context

EMS (Switzerland) holds the European patent on erythritol-based air-polishing powder for oral use. Consequently, on the EU market, non-EMS air polishers (Woodpecker, NSK, Mectron, Acteon) supply glycine or sodium bicarbonate formulations only. EMS's "Super Powder" in the EU is erythritol; Woodpecker's identically-named "Super Powder" in the EU is a glycine formulation. In the United States, the patent landscape differs and erythritol powders from multiple manufacturers are available.

This asymmetric market creates an evidentiary challenge. A large share of the erythritol literature has been conducted with EMS device and powder, with EMS providing in-kind or direct financial support. Reviews that do not distinguish industry-funded from independent trials risk laundering commercial messaging as clinical evidence.

1.4 The evidentiary gap

Prior systematic reviews addressing non-surgical peri-implantitis therapy predate most erythritol evidence (Schwarz et al., 2015) or address air polishing only as one of many adjuncts without head-to-head powder-chemistry comparison. No existing synthesis, to our knowledge, includes (a) a pre-specified industry-sponsorship sensitivity analysis, (b) studies published through Q1 2026, or (c) transparent reporting of individual-study risk of bias for the erythritol-vs-glycine comparison specifically.

1.5 Why Preventio Hub is doing this

Preventio Hub is an independent editorial publication focused on dental prophylaxis. It has no commercial relationship with any air-polishing manufacturer. This synthesis is prepared without industry funding or review. The protocol is published and timestamped in advance of data collection; deviations will be documented in the final report. The analysis code and extraction sheets will be made publicly available.

2. Objectives and research question

2.1 Primary objective

To compare erythritol and glycine air polishing, delivered as adjuncts to standard non-surgical peri-implantitis management, with respect to changes in peri-implant clinical and radiographic parameters.

2.2 Research question (PICO-T)

3. Methods

3.1 Eligibility criteria

Included. Peer-reviewed full-text publications or posted results from registered clinical trials (ClinicalTrials.gov / EudraCT / ISRCTN) reporting randomised controlled trials, controlled clinical trials, or prospective cohort studies with a comparator arm. At least one intervention arm uses erythritol-based air polishing; at least one primary outcome reported with baseline and follow-up values. Adults (≥18 years) diagnosed with peri-implantitis per the 2017 World Workshop criteria or equivalent.

Excluded. Narrative reviews, editorials, letters, commentaries without primary data; studies where erythritol is mixed with non-standard additives (e.g., probiotic blends) that prevent attribution of effect to the powder chemistry; healthy-implant maintenance populations unless a peri-implantitis subgroup is separately reported; studies lacking baseline clinical measurements; conference abstracts without a full paper, except where they are the sole report of an otherwise-eligible RCT (flagged as low-certainty).

Publication window. 1 January 2011 to 31 March 2026, updated once prior to full-synthesis publication.

Language. Reports in any language are eligible. Non-English titles and abstracts are screened after machine translation with human verification; full-text translation is commissioned only for studies surviving abstract screening.

3.2 Information sources

Electronic databases. PubMed/MEDLINE, Cochrane Central Register of Controlled Trials (CENTRAL), LILACS, and Google Scholar (top 200 results per string). Embase (Emtree) and Scopus will be procured for the full-synthesis phase.

Clinical trial registries. ClinicalTrials.gov, EU Clinical Trials Register (EudraCT), ISRCTN, German Clinical Trials Register (DRKS), Chinese Clinical Trial Registry (ChiCTR).

Grey literature and supplementary sources. Conference abstract books from EFP, Europerio, AAP, and IADR (2012–2026); preprint servers (medRxiv, bioRxiv, SSRN); manual hand-search of the last five years of Journal of Clinical Periodontology, Clinical Oral Implants Research, Journal of Periodontology, and the International Journal of Oral & Maxillofacial Implants; backward and forward citation tracking on every included study. Industry-hosted white papers (EMS, Acteon, NSK, Mectron) are listed in a transparency appendix but excluded from the synthesis proper.

3.3 Search strategy

A draft PubMed search string follows. Each database's strategy will be adapted per local syntax with MeSH/Emtree mapping and reviewed per the PRESS 2015 checklist (McGowan et al., 2016) prior to execution. Full per-database search strings will be published with the synthesis.

(
  "Peri-Implantitis"[MeSH] OR "peri-implantitis"[tiab]
  OR "peri implantitis"[tiab] OR "periimplantitis"[tiab]
  OR "peri-implant mucositis"[tiab] OR "peri-implant disease"[tiab]
)
AND
(
  "Air Abrasion, Dental"[MeSH] OR "air polishing"[tiab]
  OR "air-polishing"[tiab] OR "air abrasion"[tiab]
  OR "air powder"[tiab] OR "air-flow"[tiab] OR "airflow"[tiab]
  OR "perio-flow"[tiab] OR "perioflow"[tiab] OR "GBT"[tiab]
  OR "guided biofilm therapy"[tiab]
)
AND
(
  "erythritol"[MeSH] OR "erythritol"[tiab]
  OR "glycine"[MeSH] OR "glycine"[tiab]
  OR "sodium bicarbonate"[tiab]
  OR "low-abrasive"[tiab] OR "low abrasive"[tiab]
)
AND ("2011/01/01"[PDat] : "2026/03/31"[PDat])

The Cochrane Highly Sensitive Search Strategy will be applied for RCT identification in PubMed. The final search will be re-run 3–5 days before publication to capture recent indexation.

3.4 Study selection

Records from all sources are imported to Zotero and deduplicated. Titles and abstracts are independently screened by two reviewers in Rayyan. Full texts of potentially eligible records are retrieved and independently assessed against eligibility criteria. Disagreements are resolved by discussion; unresolved disagreements are arbitrated by a third reviewer. A PRISMA 2020 flow diagram documents numbers at every stage (identified, deduplicated, screened, assessed for eligibility, included, excluded with reasons).

Reviewer calibration is achieved through a pilot screening of 50 randomly sampled records before live screening. Target inter-rater agreement κ ≥ 0.8; if below, eligibility criteria are clarified and the pilot is re-run.

3.5 Data extraction

A structured extraction form (piloted on three studies before live use) captures: bibliographic details; study design, randomisation and blinding methods, sample size, country, setting, funding source, conflicts of interest; participant characteristics; intervention and comparator protocol details; per-outcome per-timepoint data with missing-data handling; adverse event reporting; and notes on protocol deviations, language, and translation source. Two reviewers extract data independently for a 20% random sample of included studies for quality control; extraction across remaining studies is single-reviewer with structured validation checks.

3.6 Risk of bias

Randomised studies are assessed with the Cochrane Risk of Bias tool version 2 (RoB 2), at the outcome level for each primary outcome, across five domains: randomisation, deviations from intended intervention, missing outcome data, measurement of outcome, and selection of reported result. Non-randomised studies are assessed with ROBINS-I across its seven domains. In vitro and animal studies (narrative appendix only) are appraised with a modified SYRCLE tool, narrative use only. Risk-of-bias judgments are summarised in traffic-light plots and weighted summary bars.

3.7 Data synthesis

Narrative synthesis is performed for all outcomes regardless of meta-analysis feasibility, grouped by comparator type, follow-up time horizon, and risk-of-bias category. Summary-of-findings tables per outcome are produced.

Meta-analysis is performed where all of the following hold: at least three studies report the same outcome at a comparable time horizon; between-study heterogeneity is acceptable (I² ≤ 75%, assessed via Cochran's Q and tau²); and consistent effect metrics are extractable or convertible.

Primary model: random-effects DerSimonian-Laird with Knapp-Hartung adjustment. Sensitivity analysis: fixed-effect. Effect metrics: mean difference (MD) in mm for continuous outcomes, with standardised mean difference (SMD) where instruments differ; risk ratio (RR) or odds ratio (OR) for dichotomous outcomes.

Pre-specified subgroup analyses: sub- versus supra-gingival application; smokers versus non-smokers; industry-funded versus independently funded studies.

Pre-specified sensitivity analyses: exclusion of high-risk-of-bias studies; fixed-effect model; exclusion of EMS-funded studies.

Publication bias: assessed via funnel plot and Egger's test when ≥10 studies contribute to a pooled outcome.

Analyses will be conducted in R using meta, metafor, robvis, and dmetar packages. All analysis scripts will be published as an appendix (public repository) to enable reproducibility.

3.8 Certainty of evidence

Certainty of evidence for each outcome is rated using GRADE across five domains: risk of bias, inconsistency, indirectness, imprecision, and publication bias. Summary-of-findings tables are produced using GRADEpro or equivalent. For each primary outcome, the clinician-facing summary reports the certainty grade (⨁⨁⨁⨁ / ⨁⨁⨁◯ / ⨁⨁◯◯ / ⨁◯◯◯) with plain-language interpretation.

3.9 Industry-sponsorship analysis

A distinguishing methodological feature is a structured analysis of commercial entanglement per study. Each included study is classified into one of four categories:

• Category A — Independent: no industry funding; no author conflicts of interest.
• Category B — In-kind support: powders or equipment provided at no cost; no cash funding.
• Category C — Industry-funded: direct cash grant from a manufacturer.
• Category D — Author on industry payroll: one or more authors employed by or consulting for a relevant manufacturer.

A pre-specified sensitivity analysis restricts the primary meta-analyses to Category A + B studies and reports the effect-size change versus the full dataset. Findings are presented without editorial commentary on direction; readers can draw their own inferences about the degree to which commercial entanglement explains observed effects.

4. Ethical considerations

This is a systematic review of previously published human research and does not involve new primary data collection; institutional review board approval is not required. All included studies' original ethical approvals will be noted in the data extraction.

5. Funding and conflicts of interest

No external funding has been received. The synthesis is conducted by the Preventio Hub editorial team at no cost.

The Preventio Hub editorial team declares no commercial relationships with any air-polishing device or consumables manufacturer. The editorial team has not received funding, materials, or consideration from EMS, Acteon, NSK, Mectron, Hu-Friedy, Woodpecker, Dentsply Sirona, or any other manufacturer of relevant devices or consumables. No manufacturer has reviewed this protocol in advance of publication; no manufacturer will receive advance copy of the synthesis findings.

6. Dissemination plan

The full synthesis will be published on preventiohub.com with open access to the main article, a downloadable print-designed PDF, a one-page A4 clinician summary, a public repository of extraction sheets and analysis scripts, Schema.org markup (ScholarlyArticle + ClaimReview), and notification to the European Federation of Periodontology and relevant national society communications channels.

A quarterly refresh is committed: every Q1, the search strategy is re-run, new eligible studies are added, and GRADE assessments are updated. Refresh dates are logged in the amendment section below.

7. Protocol amendments

None at the time of publication. All future amendments will be recorded here, with date, reason, and anticipated impact on the synthesis. The full edit history will be preserved as a public audit trail.

8. Registration

This protocol is published on preventiohub.com and timestamped via Internet Archive capture at the moment of publication. The Preventio Hub Evidence Registry identifier is PH-SR-2026-001. See the registry index for the full list of registered Preventio Hub evidence-synthesis projects.

A PROSPERO registration is not pursued at this initial stage because the registry's named-contact requirement conflicts with the editorial team's anonymity commitment. This remains an open option: if an independent external methodologist joins the project in a named capacity prior to the full synthesis, PROSPERO co-registration will be added and noted here as an amendment without invalidating the existing Preventio Hub Evidence Registry record.

9. References