The study protocol and informed consent following the Helsinki Declaration of 1975 (version 2008) was approved by the Medical Research Ethics Committee (2016/793), University of Bergen, Norway. The study was conducted as a randomized controlled trial with a split-mouth study design. Participating subjects read and signed the informed consent prior to enrolling in the study. The CONSORT guidelines were followed.
Prestudy calibration and training
Two operators performed the clinical aspects of this study. Author TS, masked to treatment assignments, performed all clinical recordings and sampling, author IU, unaware of previously recorded data, performed all treatments.
A calibration exercise was performed to obtain intra-examiner reproducibility for the primary outcome variables probing depth (PD) and clinical attachment level (CAL). In a sample of 10 patients, PD and CAL were recorded twice, 1 day apart, at six sites per tooth. Intraclass correlation coefficients (ICCs) were calculated separately for each site. ICC for repeated measures ranged between 0.87 and 1.00 for PD and between 0.88 and 1.00 for CAL. The calibration exercise also included the secondary outcome variable bleeding on probing (BoP). The Cohen’s kappa test displayed 83% agreement for the two pairs of recordings with a corresponding kappa value of 0.65.
As part of the research protocol, IU was trained in proper use of the air-polishing device and completed a pilot study in 13 patients.
The sample size estimation was based on change in PD. A difference of 0.5 mm was considered clinically relevant . Standard deviation of the difference between repeated PD measurements from the intra-calibration exercise was 0.5 mm. A power analysis based on 20 subjects and with the level of significance (α) set to 0.05, resulted in 98.9% power to detect a true difference of 0.5 mm.
Study subjects were recruited among patients managed in the Department of Clinical Dentistry, Section of Dental Hygiene and Section of Periodontology, University of Bergen SPT program June 2015 through June 2016. Inclusion criteria mandated 30–80-year old healthy subjects having received SPT every 3–6 months for 2–3 years following periodontal therapy, having bilateral non-mobile, fully erupted mandibular first, second or third molars with degree II furcation defects, and PD ≥ 4 mm with bleeding on probing (BoP) or pus. Following clinical examination for eligibility and medical status, 20 patients were enrolled (Fig. 1). The previous periodontal diagnosis of all included patients was moderate or severe chronic periodontitis .
Exclusion criteria were mobile mandibular molars, molars with clinical or radiographic evidence of supra-/subgingival calculus or apical pathology, use of systemic antibiotics within 6 months or SPT within 3 months of study, any current medical condition affecting periodontal treatment or use of the abrasive air-polishing device. Subjects with diabetes, cancer, HIV/aids, acute infections, disorders that may compromise wound healing, or pregnant were also excluded.
Following baseline examination, mandibular jaw quadrants were randomized (coin toss controlled by the study supervisor) to either receive debridement using the erythritol powder/air-polishing system (test) or conventional ultrasonic/curette instrumentation (control) using a split-mouth study design. Sequence of treatments was randomized in a similar fashion. Treatments were delivered at baseline, and repeated at 3, 6, 9 and 12 months. Test sites thus received root debridement using the low abrasive erythritol powder (Air-flow powder plus®, EMS, Nyon, Switzerland) applied through a Perio-Flow hand piece connected to an airflow unit (Air-Flow Master®, EMS, Nyon, Switzerland). The hand piece was fitted with a nozzle for subgingival delivery directing the power/air jet perpendicular to the root surface at the water exit at the tip of the nozzle. The nozzle was inserted to the apical aspect of furcation sites with PD ≥ 4 mm using striking movements over the furcation area for 5 s . Sites adjoining the test site with PD ≥ 4 mm were similarly treated.
Control sites were debrided using an ultrasonic scaler (Piezon Master 400 Perio Slim Tip®; Electro Medical System, Nyon, Switzerland) with power set at 75% and water as coolant, and root planed with sharp curettes (Gracey SAS, Hu-Friedy, Chicago, IL, USA). Treatment of test and control sites were carried out without anesthesia.
Following treatment of test and control sites, remaining teeth were debrided with ultrasonic and hand instruments and polished using polishing paste delivered in a rotating rubber cup. For non-experimental sites, local anesthesia was used as needed. Based on the percentage of tooth surfaces with visible plaque following staining with disclosing solution, the patients received individualized oral hygiene instruction at each appointment. Patients were returned to their regular SPT upon completion of study in which site-specific adjunctive therapy was continuously considered.
Before clinical examination, radiographs of the test and control teeth were used to assess vertical bone loss and to rule out apical pathology and supra- or subgingival calculus. The following clinical parameters were recorded at baseline, and at 3, 6, 9 and 12 months of study: PD as the distance from the gingival margin to the probable base of the pocket in mm; CAL as the distance in mm from the cemento-enamel junction or the margin of a dental restoration to the probable base of the pocket. Local probing at test and control sites and full mouth PD and CAL at six sites per tooth were recorded to the closest mm using a periodontal probe (PCP, UNC 15, Hu-Friedy, Chicago, IL, USA). BoP was recorded as present upon gentle probing to the base of the pocket . Full mouth gingival bleeding was recorded as percentage of sites showing BoP assessed at four sites per tooth, including local bleeding at test and control sites. Full mouth dental plaque scores were recorded as the percentage of tooth surfaces with visible plaque following staining with disclosing solution assessed at four sites per tooth . Presence or absence of plaque at control and test sites were also recorded . Furcations were classified at baseline and at follow-up examinations using a curved scaled Nabers furcation probe marked at 3 mm intervals (PQ2N; HU-Friedy) according to horizontal classification criteria . Furcation defects, featuring horizontal loss of periodontal support > 3 mm into the furcation but not encompassing the total width of the furcation area, were classified as degree II . Vertical attachment loss at furcation site was assessed using a periapical radiograph and clinical probing depths/CALs .
Gingival crevicular fluid assessments
Gingival crevicular fluid (GCF) was recorded at baseline, and at 6 and 12 months . Briefly, furcation sites were isolated with cottons rolls, cleaned for supragingival plaque, and air-dried. A perio paper strip was then placed 1–2 mm into the orifice of the site and left in place for 30 s. Next, the perio strip was inserted into the Periotron 8000® (Oraflow, Smithtown, NY, USA) calibrated to estimate the volume of GCF collected.
At baseline, and at 6 and 12 months the supragingival area above the furcation site was wiped clean using sterile cotton pellets. Three sterile paper points were then inserted into the pocket of the furcation site. The paper points were kept in place 20 s  removed and immersed into pre-reduced, anaerobic transport medium (PRAS; Dental Transport Medium, Morgan Hill, CA, USA). Sample tubes, separately pooled by treatment, were sent to Microbiological Diagnostic Service, Institute of Oral Biology, University Oslo, Norway for analysis using checkerboard DNA-DNA hybridization [30, 31]. Bacterial samples were analyzed for qualitative and quantitative detection of”red complex” species P. gingivalis, Treponema denticola and Tannerella forsythia  as well as Aggregatibacter actinomycetemcomitans, Prevotella intermedia, Fusobacterium nucleatum subsp. polymorphum, Fusobacterium nucleatum subsp. vincentii, Fusobacterium nucleatum subsp. nucleatum, Parvimonas micra and Prevotella nigrescens.
Pain experience assessments
Visual analogue scale (VAS) scores were used to estimate patient discomfort experienced during test and control treatment . Scoring was performed at 12 months following completion of the debridement with 0 = “no pain” and 100 = “worst pain I can imagine”.
Data were entered into MS-Excel (Microsoft, Redmond, WA, USA) proofed for errors and then imported into Stata, version 15 (StataCorp, College Station, TX, USA). All analyses were performed by a statistician (SAL) who had not taken part in data collection or treatments. Primary clinical outcome variables were changes in PD and CAL. BoP, GCF, total number bacteria and VAS scores were defined as secondary outcome variables.
Summary statistics (means ± SEM) for the clinical variables were calculated for the test and control at baseline, and at 6 and 12 months. Due to the repeated nature of data, multilevel analysis (mixed effect models) taking into consideration incomplete data at 12 months was applied to analyze the data at patient and tooth level. Time and treatment were considered fixed factors. Mixed models were applied for both primary and secondary outcome variables.
For testing differences in microbial composition at test and control sites harboring different proportions of bacteria > 105 at baseline, and at 6 and 12 months, logistic regression models with robust standard error were applied. VAS scores were analyzed using ordinary linear regression models with robust standard error. The level of significance was set at 0.05.