Ethics and study design
The protocol for this study was approved by the Ethical Committee, Faculty of Dentistry, Chulalongkorn University (IRB NO. 036/2017). The study followed the STROBE statements for reporting observational studies . This study employed a cross-sectional design. Two-hundred patients who received endosseous dental implant at a faculty-based clinic from 1996–2014 were recruited for this study. The inclusion criteria was each patient had at least 1 dental implant restored with a fixed prosthesis that was in function for more than 1 year. The exclusion criteria was patients who had implant-supported removable prosthesis, edentulous patients, and implant-supported prosthesis was in function less than 1 year. All participating patients were placed in a maintenance program, and all gave informed consent to participate in this study.
The patient’s demographics, e.g., age, sex, and dental history, were obtained from history taking, past chart review, and dental examination. The data comprised their medical and dental history, smoking habits, history of periodontal therapy, oral hygiene status, history of implant treatment, and implant prosthesis types. The patients’ clinical and radiographic examinations were performed at one visit as described in a recent publication  and reported thoroughly in the next session. Maintenance care at the implant sites was performed according to the cumulative interceptive supportive therapy (CIST) protocol . All patients received a report of their oral examination and were scheduled for their next maintenance visit.
The clinical evaluation was performed by three calibrated examiners (NS, TT, and KS) who assessed the following clinical parameters:
Modified plaque index (mPLI) : scores were determined at 4 sites per implant (mesiobuccal, distobuccal, mid-buccal, and mid-lingual). Scored from 0 to 3: 0—no plaque detection, 1—plaque recognized by running a probe across the marginal surface of the implant, 2—plaque seen with the naked eye, and 3—abundance of soft matter.
Modified bleeding index (mSBI) : scores were determined at 4 sites per implant (mesiobuccal, distobuccal, mid-buccal, and mid-lingual). Scored from 0–3: 0—no bleeding when periodontal probe is passed along the gingival margin adjacent to the implant, 1—isolated bleeding spots visible, 2—blood forms a confluent red line on margin, and 3—heavy or profuse bleeding.
Probing depth (PD): measurements were taken at 6 sites per implant (mesiobuccal, distobuccal, mid-buccal, mesiolingual, distolingual, and mid-lingual)
Mucosal recession (RE): measured in millimeters from the restorative margin to the mucosal margin. Recession was measured at 6 sites per implant (mesiobuccal, distobuccal, mid-buccal, mesiolingual, distolingual, and mid-lingual).
Width of the keratinized mucosa: the width of the keratinized mucosa was measured in millimeters at the narrowest distance between the mucosal margin and the mucogingival junction at the buccal aspect of the implant using visual and functional methodologies to identify the color, texture, and mobility differences between the keratinized mucosa and non-keratinized oral mucosa.
Tissue phenotype: the tissue phenotype was classified as a thin phenotype if the outline of the underlying periodontal probe could be seen through the buccal mucosa, and as thick phenotype if not .
All measurements were manually performed with a conventional periodontal probe (UNC-15; Hu-friedy, Chicago, IL, USA) for natural teeth and a plastic periodontal probe (12-UNC COLORVUE®; Hu-Friedy, Chicago, IL, USA) for implants. Distances were measured to the nearest millimeter (mm).
Radiographic examination was performed using standardized periapical radiographs. Digital radiographs were then taken and visualized with Infinitt proprietary software v.2 (Infinitt Co., Seoul, Korea). The interproximal bone level (BL) was evaluated by a single calibrated examiner (TS). The interproximal bone level was defined as the distance from the implant shoulder to the alveolar bone crest and was measured at the mesial and distal aspects of each implant. The most severe bone level site was selected to represent the bone level of each implant. Because the patients had been treated with different dental implant systems, we could not define a universal point of reference for all implants. Therefore, a reference point at the abutment-crown or fixture-abutment connection was defined for the respective implant system.
Due to lacking of baseline radiographs at the insertion of prosthesis, the interproximal bone loss cannot be interpreted in this study. Therefore, the interproximal bone level will be used as a parameter for analysis.
The peri-implant health and diseases were assessed based on previously established case definitions:
Healthy peri-implant: without peri-implant soft tissue inflammation and bone loss.
Peri-implant mucositis: peri-implant soft tissue inflammation present with bleeding during probing at ≥ 1 aspects of the implant (recorded from the mSBI > 2) and no evidence of supporting bone loss after bone remodeling .
Peri-implantitis: presence of soft tissue inflammation with bleeding and/or suppuration on probing at least 1 aspect of the dental implant (recorded from the mSBI > 2) and bone loss around an osseointegrated implant beyond functional remodeling ≥ 3 mm from time of loading . When there was no baseline radiograph, a threshold vertical distance of 3 mm from the expected marginal bone level was diagnosed as peri-implantitis .
Implant survival: the implant with restoration was present at the follow-up examination regardless of its condition .
Keratinized mucosa: the subjects were dichotomized into two groups. The non-keratinized mucosa group (NKM) comprised patients where there was no band of keratinized mucosa present and only alveolar mucosa was detected, and the keratinized mucosa group (KM) comprised patients where the width of the keratinized mucosa was > 1 mm.
Past periodontal status: the diagnosis of periodontal disease was classified using the American Academy of Periodontology (AAP) criteria . Patients with chronic periodontitis were those with bleeding on probing and pocket depth ≥ 4 mm in at least 30% of the total sites before implant placement.
Oral hygiene status (OHS): oral hygiene was categorized as good (mPLI < 1), fair (mPLI = 1–2), or poor (mPLI ≥ 2) .
Examiner calibration was completed before the start of the study. The intra- and inter-examiner reliability of the three clinical examiners (NS, TT, and KS) were assessed using 5 volunteer non-study subjects with ≥ 1 implant restoration. Cohen’s Kappa coefficient was used as a measure of intra- and inter-examiner reliability. The mean intra- and inter-examiner Cohen’s Kappa coefficients were 0.88 and 0.86, respectively, which indicated a high degree of reliability in the measurement.
A single calibrated examiner (TS) measured the implant bone level for 30 cases randomly drawn from the database to assess the intra-examiner reliability for radiographic bone level measurement. The reassessment was performed 7 days later to determine the measurement reproducibility. The mean bone level at the first and second measurement was 1.23 ± 1.2 mm and 1.19 ± 1.18 mm, respectively, resulting in an intra-class correlation coefficient of 0.86.
Sample size calculation
A sample size calculation was analyzed using G*Power software version 3.0.10© 1992-2008. (Universitat Kiel, Universitat Dusseldorf, Universitat Mannheim, Germany). A required sample size of 200 subjects was determined by assuming the following: (1) 90% power, (2) alpha level of 5%, and (3) a constant proportion of 0.22 was calculated from prevalence of peri-implantitis . As a result of calculation, a minimum of 180 subjects was required to provide a 90 statistical power with α = 0.05.
The data were analyzed using Statistical software SPSS version 22.0 (SPSS Inc, Chicago, IL, USA). The Kolmogorov-Smirnov test was performed to determine if the data for each parameter/variable was normally distributed. Descriptive statistics were reported as the prevalence of peri-implant disease at the implant level and subject level.
This study analyzed the data on the implant-based level of 412 implants. Differences in the mean clinical parameters between the NKM and KM groups were evaluated using the t-test or the Mann-Whitney U test. The chi-square test was used to evaluate the correlation between the keratinized mucosa groups (independent variables: NKM and KM) and categorical clinical parameters (mPLI, mSBI, PD, RE, BL, and implant status).
For regression model analysis, the unit of analysis was presence or absence of keratinized mucosa and peri-implant clinical parameters (presence of plaque, rececession > 1 mm, interproximal bone level > 3 mm). Univariate and multiple logistic regression analyses were performed to determine whether the absence of keratinized mucosa was associated with peri-implant clinical parameters and peri-implantitis after controlling for known confounding factors : smoking, diabetes, history of periodontitis, oral hygiene status, implant location, cement or screw-retained restoration, plaque accumulation, bleeding upon probing, and probing depth > 4 mm. Statistical significance was defined as P < 0.05. The risk analysis is shown in terms of odds ratio (OR) with 95% confidence intervals (95% CI).