Eleven participants, nine men and two women between 28 and 42 years of age (mean age 31.3 ± 4.6), were recruited as the study subjects. All participants provided informed consent verbally. Study information was disseminated to participants both verbally and in written form. The inclusion criteria were as follows:

Generally healthy subjects

Dentate subjects without ill-fitting restorations (≥ 24 teeth including both first upper molars)

No dental treatment planned during the study

Non-smoker or no use of tobacco for at least 6 months prior to the study enrolment

No use of systemic antibiotics in the 6 months prior to the study enrolment or daily use of mouthwashes

An impression of the upper jaw from each participant was taken to create a hard resin splint (Palapress® vario, Heraeus Kulzer, Wehrheim, Germany), and six implants were mounted on both buccal aspects of each splint and bonded with resin caps that could hold the top and apex of the implant (Fig. 1). The implants used in this study were commercially available rough surface implants (GC Aadva® implant, GC, Tokyo, Japan, 3.3-mm diameter, 8-mm length, Sa value 2.0–2.3) and machined surface implants that were identical in design other than the surface roughness (not commercially available, Sa value 0.3–0.5) (Fig. 2). When the implants were mounted on the splint, the cover screws were tightened to prevent plaque accumulation inside the implant bodies. The participants were instructed to wear the splints for 24 h/day during the 4-day experimental period except for mealtimes. During eating and drinking, the splint was taken off and kept in a provided splint box to avoid drying. The participants were allowed to brush their teeth twice a day but were not allowed to use any kind of mouthwash during the entire experimental period. At the end of the 4-day experimental period, the implants were carefully removed from the splints by breaking the resin caps that held them in place. Each implant was randomly assigned to a treatment method (Cont, control (no decontamination); G, gauze soaked in saline; US, ultrasonic scaler (SUPRASSON P-MAX, Satelec-Acteon Group, Bordeaux, France; power setting: P5, tip: Implant Protect IP3L/R); Air, air abrasive (AIR-FLOW MASTER PIEZON®, EMS, Nyon, Switzerland; power setting: water flow 100%, air pressure 75%, powder: AIR-FLOW® PERIO POWDER, nozzle: PERIO-FLOW® nozzles, distance from the nozzle to the implant 2 mm); Rot, Rotary stainless steel instrument (iBrush, NeoBiotech©, Los Angeles, CA, USA; rotating speed 1500 rpm); Las, Er:YAG laser (Erwin AdvErL, J.Morita©, Kyoto, Japan; power setting 60 mJ/pulse, 10 pps, tip C600F, distance from the tip to the implant 2 mm) according to the random number table generated by a spreadsheet software (random number table generator: Excel® for Mac 2011, version 14.7.2, Microsoft®) (Fig. 3). All methods, including Cont, were applied to each set of 6 implants. One investigator (M.O) was blinded to which implant was assigned to which method. The investigator was experienced in using each method to treat peri-implantitis in clinical practice. The implant driver was connected to the implants to hold them without touching the implant surface during decontamination. The implants were decontaminated by their assigned method, apart from those assigned to the Cont group, for 1 min. After cleansing, decontaminated implants were immediately stored in a phosphate-buffered saline solution. Qualitative SEM analysis was performed using samples taken from the participant who had shown average plaque accumulation on the implant surface in the preliminary study (data not shown), and quantitative colony-forming unit (CFU) counts were performed using the remaining 10 participants’ samples. This study protocol was approved by the ethical committee of Osaka University (H26.E-36).