Clinical research of peri-implant diseases – quality of reporting, case definitions and methods to study incidence, prevalence and risk factors of peri-implant diseases

Authors


  • Conflict of interest and source of funding statement

  • The authors declare no personal conflict of interest. This workshop was financially supported by the European Federation of Periodontology and by unrestricted grants from Astra, Nobel Biocare and Straumann.

Address:

Cristiano Tomasi

Department of Periodontology

Institute of Odontology

The Sahlgrenska Academy at University of Gothenburg

Box 450

SE 405 30 Gothenburg, Sweden

E-mail: cristiano.tomasi@odontologi.gu.se

Abstract

Aim

To review the quality of reporting and the methodology of clinical research on the incidence, prevalence and risk factors of peri-implant diseases.

Methods

A MEDLINE search was conducted for cross-sectional, case-control and prospective longitudinal studies reporting on peri-implant diseases. To evaluate the quality of reporting of the selected studies the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) checklist was utilized.

Results

The search provided 306 titles and abstracts, out of which 40 were selected for full-text analysis. Finally, 16 studies were included out of which five assessed prevalence and only two the incidence of peri-implant diseases. 13 articles studied risk indicators for peri-implant diseases. None of the scrutinized articles adhered fully to the STROBE criteria. The large majority of articles did not (i) clearly state the applied study design, (ii) describe any effort to address potential sources of bias, (iii) explain how missing data were addressed, (iv) perform any kind of sensitivity analysis, (v) indicate the number of participants with missing data for each variable of interest.

Conclusion

Collectively, the findings of this review indicate a need for improved reporting of epidemiological studies on peri-implant diseases.

Introduction

Chronic conditions, such as peri-implantitis and periodontitis have implications for policy and health care utilization (Fox 2005, Lavis et al. 2005). Accordingly, biological complications following dental implant therapy are receiving increasing interest in the scientific literature. However, limited information is available regarding incidence, prevalence and risk factors of peri-implant diseases (Heitz-Mayfield 2008, Zitzmann & Berglundh 2008).

The medical dictionary defines incidence as “the number of new cases of a specific disease occurring during a certain period” while prevalence is described as “the number of cases of a disease in existence at a certain time point” (Dorland 1994). It is understood that these parameters can only be evaluated in observational studies (Aschengrau & Seage 2003). A risk factor, as identified in a multiple-step process (Beck 1994), is defined as “an environmental, behavioral, or biological factor that, if present directly increases the probability of a disease occurring and, if absent or removed, reduces that probability” (Genco et al. 1996).

The 6th European Workshop on Periodontology suggested the use of cross-sectional studies for the analysis of the prevalence of peri-implant diseases, while longitudinal studies should provide information on disease incidence (Zitzmann & Berglundh 2008). True risk factors for peri-implant diseases can only be identified by prospective longitudinal studies while retrospective and cross-sectional studies can identify risk indicators for disease (Heitz-Mayfield 2008).

When studying prevalence, incidence and risk factors of a disease, a clear definition of pathological condition is required. Thus, sound cut-off points defining presence and absence of the disorder, definitions of disease severity and an appropriate research methodology are of utmost importance. A variety of different parameters and disease criteria have been used when studying incidence, prevalence and risk factors of peri-implant diseases (for review, see Zitzmann & Berglundh 2008). Accordingly, a wide range of different results has been reported (Karoussis et al. 2004, Fransson et al. 2005, 2008, Ferreira et al. 2006, Roos-Jansåker et al. 2006a,b, Koldsland et al. 2010). To make observations comparable, the 6th Workshop on Periodontology suggested two definitions of peri-implant diseases to be adopted in future research. Namely, that the diagnosis “peri-implant mucositis” should be based on the parameter “bleeding upon probing” without loss of supporting bone while the diagnosis “peri-implantitis” should include both “bleeding upon probing” and detectable bone loss (after 1 year in function) (Lindhe & Meyle 2008).

The reporting of observational research, for example, the quality, the content as well as the structure and style of manuscripts, has been criticized (Pocock et al. 2004, Tooth et al. 2005). It has been argued that without sufficient quality and content of the published article, the benefits of research might be achieved more slowly and that generalizability of results might be hampered (Bogardus et al. 1999). It is important, however, to distinguish between quality of reporting and quality of what was actually done in the design, conduct and analysis of a study. A high-quality report ensures that all relevant information about a study is available to the reader, but does not necessarily reflect a low susceptibility to bias (Huwiler-Muntener et al. 2002, Sanderson et al. 2007).

To improve scientific texts describing randomized trials the Consolidated Standards of Reporting Trials (CONSORT) was developed in 1996 and revised 5 years later (Moher et al. 2001). More recently, a similar suggestion was made for observational research: The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement (Vandenbroucke et al. 2007a,b,c). The aim was to ensure clear presentation of what was planned, done, and found in an observational study. The STROBE Statement is a checklist of items that should be addressed in articles reporting on the three main study designs of analytical epidemiology: cohort, case-control and cross-sectional studies (von Elm et al. 2008).

The multitude of different disease criteria, the diagnostic and methodological inconsistencies as well as a varying quality of reports have so far hampered attempts to draw strong conclusions in the field of peri-implant diseases. The aim of the current study was to review the quality of reporting and the methodology of clinical research on the incidence, prevalence and risk factors of peri-implant diseases.

Material and Methods

Case definition

Different forms of biological complications were categorized as peri-implant diseases. These included (i) progressive bone loss with or without information on bleeding on probing, (ii) variously defined peri-implant mucositis and (iii) variously defined peri-implantitis.

Type of studies/inclusion criteria

For articles reporting on the incidence/prevalence of peri-implant diseases, studies of cross-sectional and longitudinal design were considered. For articles reporting on risk factors, cross-sectional, longitudinal, prospective and case-control studies were eligible. Articles reporting site/implant-level analysis only, without considering subject-level prevalence of peri-implant diseases were not considered for inclusion.

No limits regarding minimum follow-up time nor minimum number of subjects were set. Studies reporting only mean values of bone loss over time or survival and success rates with the implant as the statistical unit were excluded.

Search strategy

A MEDLINE search (PubMed) was conducted and work published in the English language until March 2011 was included in the review. The following search terms were used in different combinations: “dental implant”, “mucositis”, “peri-implantitis”, “periimplantitis”, “peri implantitis”, “peri-implant disease”, “bone loss”, “biological complication”, “incidence”, “prevalence”, “risk factor” and “risk indicator”. Search limitations were the terms “english” and “human”.

Titles and abstracts were screened for information on variables assessed and full-text analysis was performed in relevant publications. Further manual search included (i) bibliographies of previous reviews and (ii) the following journals: Clinical Implant Dentistry and Related Research, Clinical Oral Implants Research, European Journal of Oral Implants, Implant Dentistry, International Journal of Oral and Maxillofacial Implants, International Journal of Periodontics and Restorative Dentistry, Journal of Oral and Maxillofacial Surgery, Journal of Clinical Periodontology, Journal of Periodontal Research and Journal of Periodontology. A flow-chart of the search is presented in Fig. 1.

Figure 1.

Selection flow-chart.

Assessment of research methodology

To evaluate the research methodology we extracted key components from each of the studies. Data on the aim of the study, study design, sampling procedures, variables assessed, disease definition, follow-up time, validation of measurements, factor assessment and validation, statistical analysis and data reporting were described (Tables 1, 3 and 5).

Table 1. Studies on the incidence of peri-implant diseases – methodology
First author, yearAimStudy design, follow-upSite, setting & fundingSampling, sample sizeMeasurementsPeri-implant disease definitionValidationData reportingIncidence
  1. BoP, bleeding on probing; PPD, probing pocket depth; SUP, suppuration.

Roccuzzo et al. (2010)Incidence of bone loss in population subggroupsProspective cohort study; 10 years

Italy

Private
Not reported

Convenient sample

Initial: 112
Assessed: 110

X-ray

Intra-oral
Clinical
PPD, BoP, plaque

Bone loss ≥ 3 mmExaminer calibration; no data reportedFrequency on subject level22.8%
Zetterqvist et al. (2010)Incidence of peri-implantitis in population subgroups

Prospective RCT,

5 years

Sweden

Private/University
Industry

Convenient sample

Initial: 112
Assessed: 96

X-ray

Intra-oral
Clinical
PPD, BoP, Plaque

Mucositis

BoP+ or SUP+
Peri-implantitis
BoP+ or SUP+ & PPD > 5 mm & bone loss > 5 mm

Not reportedFrequency on subject level1%

Assessment of reporting quality

To evaluate the quality of the reporting of the selected studies, we referred to the STROBE statement on items that should be included in reports of observational studies (Fig. 2). We therefore assessed in each article whether the items named in the checklist pertinent to the given study design were present, absent or not applicable. For detailed explanation and elaboration on the STROBE checklist we referred to (Vandenbroucke et al. 2007a,b,c).

Figure 2.

STROBE statement.

Results

Search results

The combinations of search terms generated a list of 306 titles (MEDLINE until March 2011) and, following screening of the abstracts independently by the two reviewers, resulted in the selection of 40 full-text publications. The level of agreement between reviewers measured by k-score for the first screening was 0.79. The intra-examiner reproducibility was 0.79 and 0.87 respectively for the two reviewers. Following application of the inclusion criteria in the full-text analysis, 24 of the 40 studies were excluded, resulting in 16 articles included in this review (Fig. 1).

Publications were not included for analytical evaluation due to the following reasons:

  • implant-based data analysis only,
  • no frequency analysis of biological complications or peri-implant diseases,
  • retrospective study design.

The 16 included articles reported data from 13 different patient samples.

Case definition

The disease criteria adopted by the various authors differed.

Mucositis was defined as the presence of bleeding after probing (BoP) in six publications (Ferreira et al. 2006, Roos-Jansåker et al. 2006a,b Maximo et al. 2008, Karbach et al. 2009, Koldsland et al. 2010, 2011) of which four added the possibility of pus after probing (Ferreira et al. 2006, Roos-Jansåker et al. 2006a,b, Zetterqvist et al. 2010). Different values of probing pocket depth (PPD) were chosen as thresholds: ≥5 mm (Karbach et al. 2009), <5 mm (Ferreira et al. 2006) and ≥4 mm (Roos-Jansåker et al. 2006a,b). Two studies added the condition of “no bone loss” as part of the mucositis definition (Koldsland et al. 2010, 2011) while others had higher thresholds such as radiographic bone loss up to three threads after the first year of loading (Roos-Jansåker et al. 2006a,b, Maximo et al. 2008).

Peri-implantitis was defined in 12 studies as the presence of BoP and/or pus with concomitant radiographic bone loss (Fransson et al. 2005, 2008, Ferreira et al. 2006, Laine et al. 2006, Roos-Jansåker et al. 2006a,b, Gatti et al. 2008, Maximo et al. 2008, Cury et al. 2009, Koldsland et al. 2010, 2011, Zetterqvist et al. 2010).

Different thresholds of radiographic bone loss were applied as a disease criteria in the different studies:

  • >0.4 mm from time of loading (Koldsland et al. 2010, 2011),
  • detectable bone loss from 1-year examination & bone level ≥1.8 mm (Fransson et al. 2005, 2008)
  • ≥1.8 mm from 1-year examination (Laine et al. 2006, Roos-Jansåker et al. 2006a,b, Maximo et al. 2008),
  • ≥2 mm from time of loading (Koldsland et al. 2011),
  • >2 mm from the last radiographic assessment (Gatti et al. 2008),
  • ≥3 mm from time of abutment connection (Roccuzzo et al. 2010),
  • ≥3 mm from time of loading (Cury et al. 2009),
  • >5 mm (Zetterqvist et al. 2010).

Two articles assessed presence of peri-implantitis as an outcome variable without clearly defining it (Gruica et al. 2004, De Boever et al. 2009), while another study included the presence of PPD ≥ 5 mm without any bone loss threshold (Ferreira et al. 2006). Initially, Fransson et al. (2005) reported on radiographic bone loss only. A sub-sample of subjects identified with progressive bone loss was later examined clinically (Fransson et al. 2008). One study identified subjects with radiographic bone loss of > 3 mm but did not report clinical parameters at the discerned implants (Roccuzzo et al. 2010).

Articles on the incidence of peri-implant diseases

Methodology

Two papers addressing the incidence of peri-implant diseases were identified (Roccuzzo et al. 2010, Zetterqvist et al. 2010). Details are reported in Table 1.

Study design

The two studies were designed as either a prospective randomized controlled trial (Zetterqvist et al. 2010) or a prospective cohort study (Roccuzzo et al. 2010).

Sampling, sample size and observation time

In both studies, subjects were recruited from convenience samples. Both studies initially included 112 subjects out of which, in one study, 101 were still available after 10 years (Roccuzzo et al. 2010), while Zetterqvist et al. (2010) could examine 96 subjects after 5 years.

Assessments and validation

Peri-implant conditions were assessed both by intra-oral radiographs and by clinical examination. No information was given in terms of internal validation of the measurements in one study (Zetterqvist et al. 2010) while Roccuzzo et al. (2010) mentioned examiner calibration without reporting further details.

Data analysis

Both studies reported crude values of incidence of peri-implantitis and for sub-groups. The statistical unit was the subject.

Results

Zetterqvist et al. (2010) reported an incidence of peri-implantitis over the 5-year period of 1% while Roccuzzo et al. (2010) found an incidence of 23% during 10 years.

Quality of reporting

Details on the adherence to the STROBE criteria are reported in Table 2. In total, the authors fulfilled approximately 70% of the applicable criteria. No information regarding internal validity or potential sources of bias was given in one paper (Zetterqvist et al. 2010). Handling of missing data was reported in neither of the two studies. Limitations and generalizability of the results were not discussed. Details on funding were not reported in one study (Roccuzzo et al. 2010).

Table 2. Studies on the incidence of peri-implant diseases – STROBE adherence
STROBE criteria adherenceRoccuzzo et al. (2010)Zetterqvist et al. (2010)Adherence
Title and abstract 1 (a)YesYes2/2
1 (b)YesYes2/2
IntroductionBackground2YesYes2/2
Objectives3NoYes1/2
MethodsStudy design4YesYes2/2
Setting5YesYes2/2
Participants6(a)YesYes2/2
6(b)NANANA
Variables7YesYes2/2
Measurements8YesNo1/2
Bias9YesNo1/2
Study Size10YesNo1/2
Quantitative var.11YesNA1/1
Statistical methods12(a)YesYes2/2
12(b)YesNo1/2
12(c)NoNo0/2
12(d)YesYes2/2
12(e)NoNA0/1
ResultsParticipants13(a)YesYes2/2
13(b)YesYes2/2
13(c)NoNo0/2
Descriptive data14(a)NoYes1/2
14(b)NoNo0/2
14(c)YesYes2/2
Outcome data15YesYes2/2
Main results16(a)NoNA0/1
16(b)YesNA1/1
16(c)NANANA
Other analyses17YesNo1/2
DiscussionKey results18YesYes2/2
Limitations19NoNo0/2
Interpretation20YesYes2/2
Generalizability21NoNo0/2
Other infoFunding22NoYes1/2
Adherence22/3218/28Total: 67%

Articles on the prevalence of peri-implant diseases

Methodology

Five articles assessed prevalence of peri-implant diseases (Fransson et al. 2005, Ferreira et al. 2006, Roos-Jansåker et al. 2006a, Maximo et al. 2008, Koldsland et al. 2010). Details are reported in Table 3.

Table 3. Studies on the prevalence of peri-implant diseases – methodology
First author, yearAimStudy design, follow-upSite, setting & fundingSampling, sample sizeMeasurementsPeri-implant disease definitionValidationData reportingPrevalence
  1. PPD, probing pocket depth; BoP, bleeding on probing; CAL, clinical attachment level; CI, confidence interval; GI, gingival index; OGP, orthopantomogram; OR, odds ratio; SUP, suppuration; SPT, supportive periodontal therapy.

Fransson et al. (2005)Prevalence of subjects with progressive bone lossDesign not reported, minimum follow-up of 5 years, range 5 to 20 years

Sweden University

Not reported

Convenient sample

Initial: 662
Assessed: 662

X-ray

Intra-oral

Bone loss ≥ 3 threads & progression in relation to bone level at year 1Inter-examiner agreement regarding bone level thresholds, no data reportedFrequency on subject level27.8%
Ferreira et al. (2006)Prevalence of peri-implant diseaseCross-sectional, minimum follow-up of 6 months, mean 42.5 monthsBrazil University Not reported

Convenient sample

Initial: 229
Assessed: 212

X-ray

Intra-oral Clinical
PPD, BoP, SUP

Mucositis

BoP+ or SUP+ & PPD < 5 mmor no bone loss
Peri-implantitis
BoP+ & PPD ≥ 5 mm

Inter- and intra-examiner agreement; k scores reportedFrequency on subject level

Mucositis

64.6% Peri-implantitis
8.9%

Roos-Jansåker et al. (2006a)Prevalence of peri-implant diseaseDesign not reported, minimum follow-up of 9 years, range 9 to 14 years

Sweden University

Institutional

Convenient sample

Initial: 294
Assessed: 216

X-ray

Intra-oral Clinical
PPD, BoP, SUP

Mucositis

BoP+ or SUP+ & PPD ≥ 4 mm Peri-implantitis
Bone loss ≥ 3 threads after year 1 & BoP+ or SUP+

Inter- and intra-examiner agreement; k scores reportedFrequency on subject level

Mucositis

48% Peri-implantitis
16%

Maximo et al. (2008)Prevalence of peri-implant diseaseDesign not reported, minimum follow-up of 1 year, mean 3.4 yearsBrazil University Not reported

Convenient sample

Initial: 224
Assessed: 113

X-ray

Intra-oral Clinical
PPD, BoP, SUP, GI.

Mucositis

BoP+ and bone loss < 3 threads Peri-imlantitis BoP+ and PPD > 5 mm and bone loss ≥ 3 threads

Intra-examiner calibration, stand error of measurement reportedFrequency on subject level

Mucositis

36.3% Peri-implantitis
12.4%

Koldsland et al. (2010)Prevalence of peri-implant diseaseCross-sectional, minimum follow-up of 2 years, mean 8.4 years

Norway University

Institutional

Convenient sample

Initial: 164
Assessed: 99

X-ray

Intra-oral and OPG
Clinical
PPD, BoP, CAL, Occlusion

Mucositis

BoP+ and no bone loss Peri-imlantitis BoP+ and bone loss > 1 standard deviation of the measurement error

Inter- and intra-examiner agreement; k scores reportedFrequency on subject level

Mucositis

39.4% Peri-implantitis
47.1%

Study design

The recruitment of subjects and sampling methodology in these studies suggest a cross-sectional design. However, only two of the five papers explicitly mentioned the study design adopted (Ferreira et al. 2006, Koldsland et al. 2010).

Sampling, sample size and observation time

All studies included subjects from a convenience sample. None considered a patient sample randomly selected from a wider population. The number of assessed subjects ranged from 99 (Koldsland et al. 2010) to a maximum of 662 (Fransson et al. 2005). The observation periods in the included studies generally started with the loading of the implants. In terms of minimum follow-up time required, one study included subjects with implants loaded for 6 months (Ferreira et al. 2006). One study included subjects with a minimum of 2 years of implant loading (Koldsland et al. 2010). Fransson et al. (2005) reported on subjects with a minimum of five and a maximum of 20 years of loading. A follow-up between 9 and 14 years was reported by Roos-Jansåker et al. (2006a). The mean follow-up time ranged from 3.4 (Maximo et al. 2008) to over 9 years (Fransson et al. 2005, Roos-Jansåker et al. 2006a).

Assessments and validation

One study reported only radiographic measurements of progressive bone loss (Fransson et al. 2005) but added clinical data on parts of the the earlier material in a later publication (Fransson et al. 2008). The remaining studies considered both radiographic and clinical assessments. In one study intra-examiner validity of the measurements was performed and described by the value for the error measured (Maximo et al. 2008). Fransson et al. (2005) described the inter-examiner agreement regarding the assessment of radiographic thresholds but no further information was given. Three articles reported in detail on the assessment of intra- and inter-examiner reproducibility (Ferreira et al. 2006, Roos-Jansåker et al. 2006a, Koldsland et al. 2010).

Data analysis

The crude value of prevalence, using the subject as the statistical unit was reported in all of the included articles. No sub-grouping was performed in assessing prevalence of peri-implant diseases. One article calculated the prevalence while applying different thresholds of disease (Koldsland et al. 2010).

Results

The reported prevalence of peri-implant mucositis varied between 36.3% (Maximo et al. 2008) and 64.6% (Ferreira et al. 2006). The results for peri-implantitis ranged from 8.9% (Ferreira et al. 2006) to 47.1% (Koldsland et al. 2010).

Quality of reporting

Details on the adherence to the STROBE criteria are reported in Table 4. In total, the authors fulfilled between 62% and 71% of the applicable criteria. Three of the five studies did not explicitly report on the design of the study (Fransson et al. 2005, Roos-Jansåker et al. 2006a, Maximo et al. 2008). Background and objectives were described in all articles, while setting and location of the study was described in detail in four of five articles (Fransson et al. 2005, Roos-Jansåker et al. 2006a, Maximo et al. 2008, Koldsland et al. 2010). None of the articles reported a description testifying efforts to address potential sources of bias, while a total of four studies described how the study size was determined (Fransson et al. 2005, Roos-Jansåker et al. 2006a, Maximo et al. 2008, Koldsland et al. 2010). Regarding statistical analysis, none of the included studies reported any information about handling of missing data nor did they describe sampling strategy. A thorough description of eligibility of subjects was given in all articles. Only one article published a flow diagram describing participant recruitment and eligibility (Fransson et al. 2005). None of the articles reported details on participants with missing data and two failed to discuss limitations of the given study (Fransson et al. 2005, Roos-Jansåker et al. 2006a). Generalizability of results were discussed in three of the articles (Ferreira et al. 2006, Roos-Jansåker et al. 2006a, Maximo et al. 2008). Three articles did not explicitly mention the source of funding (Fransson et al. 2005, Ferreira et al. 2006, Maximo et al. 2008).

Table 4. Studies on the prevalence of peri-implant diseases – STROBE adherence
STROBE criteria adherenceFransson et al. (2005)Ferreira et al. (2006)Roos-Jansåker et al. (2006a)Maximo et al. (2008)Koldsland et al. (2010)Adherence
Title and abstract 1 (a)NoNoNoNoNo0/5
1 (b)YesYesYesYesYes5/5
IntroductionBackground2YesYesYesYesYes5/5
Objectives3YesYesYesYesYes5/5
MethodsStudy design4NoYesNoNoYes3/5
Setting5YesNoYesYesYes4/5
Participants6(a)YesYesYesYesYes5/5
6(b)NANANANANANA
Variables7YesYesYesYesYes5/5
Measurements8YesYesYesYesYes5/5
Bias9NoNoNoNoNo0/5
Study Size10YesNoYesYesYes1/5
Quantitative var.11YesYesYesYesYes5/5

Statistical

methods

12(a)YesYesNoYesYes1/5
12(b)YesYesNAYesNA3/3
12(c)NoNoNoNoNo0/5
12(d)NANANANoNo0/0
12(e)NoNoNoNoNo0/5
ResultsParticipants13(a)YesYesYesYesYes5/5
13(b)NANoYesYesYes1/4
13(c)YesNoNoNoNo1/5
Descriptive data14(a)YesYesNoYesYes4/5
14(b)NoNoNoNoNo0/5
14(c)NANANANANANA
Outcome data15YesYesYesYesYes5/5
Main results16(a)NoYesNANoNA0/2
16(b)NAYesNAYesNA2/2
16(c)NoNoNANoNA0/3
Other analyses17YesYesYesYesYes5/5
DiscussionKey results18YesYesYesYesYes5/5
Limitations19NoYesNoYesYes3/5
Interpretation20YesYesYesYesYes5/5
Generalizability21NoYesYesYesNo3/5
Other infoFunding22NoNoYesNoYes2/5
Adherence18/2920/3117/2721/3220/28Total: 65.3%

Articles on risk factors of peri-implant diseases

Methodology

Thirteen of the selected articles reported on risk factors for peri-implant diseases (Gruica et al. 2004, Ferreira et al. 2006, Laine et al. 2006, Roos-Jansåker et al. 2006b, Fransson et al. 2008, Gatti et al. 2008, Maximo et al. 2008, Cury et al. 2009, De Boever et al. 2009, Karbach et al. 2009, Roccuzzo et al. 2010, Zetterqvist et al. 2010, Koldsland et al. 2011). Details are reported in Table 5.

Table 5. Studies on risk factors of peri-implant diseases – methodology
First author, yearAimStudy design, follow-upSite, setting & fundingSampling, sample sizeMeasurementsPeri-implant disease definitionValidationFactors investigatedFactor validationStatistical analysisData reportingConclusion
  1. BoP, bleeding on probing; CAL, clinical attachment level; GI, gingival index; OGP, orthopantomogram; PPD, probing pocket depth; SUP, suppuration; SPT, supportive periodontal therapy.

Gruica et al. (2004)Relation between genetic polymorphisms of IL-1 and biological implant complicationsDesign not reported, minimum follow-up of 8 years, range 8 to 15 years

Switzerland University

Industry

Convenient sample

Initial: 223
Assessed: 180

X-ray

Intra-oral
Clinical
PPD, BoP, SUP, CAL, DNA sample from cheek cells
Interview

Peri-implantitis

Not defined

Not reportedSmoking, IL-1 genotypeNot reportedMultivariate logistic regressionOR and 95% CIHigher risk of peri-implantitis in heavy smokers with IL-1 positive genotype.
Ferreira et al. (2006)Peri-implant disease association with demographic, behavioural and biological risk variablesCross-sectional, minimum follow-up of 6 months, mean 42.5 months

Brazil University

Not reported

Convenient sample

Initial: 229
Assessed: 212

X-ray

Intra-oral
Clinical
PPD, BoP, SUP

Mucositis

BoP+ or SUP+ & PPD < 5 mm or no bone loss Peri-implantitis
BoP+ & PPD ≥ 5 mm

Inter- and intra-examiner agreement; k scores reportedAge, gender, plaque, periodontal status, diabetes, time follow-up, maintenance frequencyPlaque and periodontal status: Inter- and intra-examiner agreement; k scores reportedPearson's Chi-square test Multinomial Logistic Regression modelOR and 95% CIHigher risk of peri-implantitis in subjects with periodontitis, diabetes and poor oral hygiene.
Laine et al. (2006)Association between IL-1 polymorphisms and peri-implantitisCase-control, minimum loading time of 2 years

Sweden University

Institutional

Convenient sample

Total: 120
Case: 71 Control: 49

X-ray

Not specified
Clinical
DNA samples collected by rinsing
Interview

Peri-implantitis

Bone loss ≥ 3 threads and BoP+

Not reportedIL-1 genotype, smoking, gender, ageNot reportedChi-square test and Fisher's exact test Logistic regression analysisOR and 95% CIIL-1RNgene polymorphism is associated with peri-implantitis.
Roos-Jansåker et al. (2006b)Associated factors related to peri-implant lesionsDesign not reported, minimum follow-up of 9 years, range 9 to 14 years

Sweden University

Institutional

Convenient sample

Initial: 294
Assessed: 216

X-ray

Intra-oral
Clinical
PPD, BoP, SUP Interview

Mucositis

BoP+ or SUP+ & PPD ≥ 4 mm
Peri-implantitis
Bone loss ≥ 3 threads after year 1 & BoP+ or SUP+

Inter- and intra-examiner agreement; k scores reportedKeratinized mucosa, plaque, % teeth with bone loss, SPT, gender, age, education, smoking, systemic diseaseNot reportedUnivariate and multivariate logistic regression analysisOR and 95% CIHigher risk of peri-implantitis in periodontitis patients and in smokers.
Fransson et al. (2008)

Clinical

characteristics of previously identified individuals exhibiting radiographic bone loss

Design not reported, minimum follow-up of 5 years, range 5 to 20 years

Sweden University

Not reported

Convenient sample

Initial: 482
Assessed: 82

X-ray

Intral-oral
Clinical
PPD, BoP, SUP
Interview

Bone loss ≥ 3 threads & progression in relation to bone level at year 1Inter-examiner agreement regarding bone level thresholds, no data reportedAge, number of implants, function time, jaw, gender, smokingNot reportedMultivariate logistic regressionp-valuesLarger numbers of implants affected in smokers.
Gatti et al. (2008)Prevalence of peri-implantitis in population subgroupsProspective cohort study, follow-up of 5 yearsItaly Private Not reported

Convenient sample

Initial: 62
Assessed: 56

X-ray

Intra-oral
Clinical
PPD, BoP, SUP

Peri-implantitis

Bone loss > 2 mm from last x-ray & SUP+/BoP+ & PPD > 5 mm

Not reportedPeriodontitis (severe, moderate, none)Not reportedChi-square testNo calculations carried out due to too few eventsTrend towards higher risk of peri-implantitis in periodontitis patients.
Maximo et al. (2008)Relation of peri-implant diseases with periodontal bone loss, systemic conditions and demographic profileDesign not reported, minimum follow-up of 1 year, mean 3.4 years

Brazil University

Not reported

Convenient sample

Initial: 224
Assessed: 113

X-ray

Intra-oral
Clinical
PPD, BoP, SUP, GI Interview

Mucositis

BoP+ and bone loss < 3 threads Peri-imlantitis BoP+ and PPD > 5 mm and bone loss ≥ 3 threads

Intra-examiner calibration, stand error of measurement reportedPeriodontal bone loss, age, gender, socioeconomic status, smoking, BMI, medical historyNot reportedCorrelation analysis with Pearson Chi-square or Spearman testp-valuesPositive correlations between mucositis/peri-implantitis and time of loading and between peri-implantitis and periodontal bone loss.
Cury et al. (2009)Association between a specific gene polymorphism for TNFα and peri-implantitisCase-control, minimum loading time of 6 months, mean 7.4 years

Brazil University

Institutional

Convenient sample

Total: 90
Case: 49 Control: 41

X-ray

intra-orals
Clinical
PPD, BoP, SUP Blood test

Peri-implantitis

BoP+/SUP+ and bone loss ≥ 3 mm

Not reportedGenetic polymorphismNot reportedChi-square testp-valuesPolymorphism in the TNFα-308 gene is not associated with peri- implantitis
De Boever et al. (2009)Outcome in population subgroups and comparing two implant surfacesProspective cohort study, mean follow-up of 48 monthsBelgium University Industry

Convenient sample

Initial: 221
Assessed: 194

X-ray

Intra-oral
Clinical
PPD, BoP, SUP, Plaque Interview

Peri-implantitis not definedNot reportedImplant surface, periodontal history, age, health status, smoking, plaque, follow-upNot reportedMultiple regression analysisp-valuesMore implant pathology in subjects with aggressive periodontitis.
Karbach et al. (2009)Identify risk factors for peri-implant mucositisDesign not reported, minimum follow-up of 6 months, range 1–19 yearsGermany University Not reported

Convenient sample

Initial: 100
Assessed: 100

Clinical

PPD, BoP, Plaque
Interview
Clinical charts

Mucositis

BoP+ & PPD ≥ 5 mm

Not reportedSurface roughness, smoking, augmentation, type of dentition, radiation therapyNot reported

1 implant per subject assessed

Multivariate regression analysis

OR and 95%

Confidence interval

Smoking is a risk factor for the formation of peri-implant mucositis.
Roccuzzo et al. (2010)Incidence of bone loss in population subggroupsProspective cohort study; 10 yearsItaly Private Not reported

Convenient sample

Initial: 112
Assessed: 110

X-ray

Intra-oral Clinical
PPD, BoP, Plaque Clinical
charts

Bone loss ≥ 3 mmExaminer calibration; no data reportedHistory of periodontitis, adhesion to SPTNot reportedKruskal-Wallis rank anova, Mann-Whitney U-test, Fisher exact testFrequency on subject level, p-valuesHistory of periodontitis and non-adhesion to SPT are risk factors for peri-implant disease.
Zetterqvist et al. (2010)Incidence of peri-implantitis in population subgroupsProspective RCT; 5 years

Sweden Private/University

Industry

Convenient sample

Initial: 112
Assessed: 96

X-ray

Intra-oral
Clinical
PPD, BoP, Plaque

Mucositis

BoP+ or SUP+ Peri-implantitis BoP+ or SUP+ & PPD > 5 mm & Bone loss > 5 mm

Not reportedImplant surface characteristicsNot applicableChi-square testp-valuesThe studied implant surface characteristics are no risk factors for peri-implantitis.
Koldsland et al. (2011)Assess possible risk indicators for peri-implantitisCross-sectional, minimum follow-up of 2 years, mean 8.4 years

Norway University

Institutional

Convenient sample

Initial: 164
Assessed: 109

X-ray

Intra-oral and OPG Clinical
PPD, BoP, CAL, Occlusion
Interview

Mucositis

BoP+ and no bone loss
Peri-implantitis
BoP+ and bone loss > 1 standard deviation of the measurement error

Inter- and intra-examiner agreement; k scores reportedLoading time, keratinized mucosa, plaque, location, gender, periodonal history, smokingNot reportedMultilevel anlysisOR and 95% Confidence intervalSignificantly higher risk of peri-implantitis for implants placed in the maxilla.

Study design

Five of the 13 articles did not explicitly report on the design of the study (Gruica et al. 2004, Roos-Jansåker et al. 2006b, Fransson et al. 2008, Maximo et al. 2008, Karbach et al. 2009). Subject recruitment and methodology, however, suggest a cross-sectional design. Two other articles reported on their cross-sectional approach (Ferreira et al. 2006, Koldsland et al. 2011). One study was a prospective randomized controlled trial (Zetterqvist et al. 2010) while three publications were designed as prospective cohort studies (Gatti et al. 2008, De Boever et al. 2009, Roccuzzo et al. 2010). Two more were described as case-control studies (Laine et al. 2006, Cury et al. 2009).

Sampling, sample size and observation time

Again, all 13 studies were based on convenience samples. The number of assessed subjects ranged from 56 (Gatti et al. 2008) to a maximum of 216 (Roos-Jansåker et al. 2006b). Regarding minimum follow-up time required, one study included subjects with implants loaded for 6 months (Ferreira et al. 2006). Maximum time of follow-up for some subjects was 19 years (Karbach et al. 2009). The prospective studies had observation periods between 5 (Gatti et al. 2008, Zetterqvist et al. 2010) and 10 years (Roccuzzo et al. 2010). The two case-control studies (Laine et al. 2006, Cury et al. 2009) had varying minimum loading times of 6 and 24 months, respectively.

Assessments and validation

Several different potential risk factors were studied in the articles. Subject variables such as gender, age, history of periodontitis and smoking were examined in virtually all of the studies. Local factors such as the presence of keratinized tissues as well as molecular variables such as genetic polymorphisms were assessed by some of the articles (Table 5).

Seven of the 13 studies did not report any data on the internal validity of performed measurements (Gruica et al. 2004, Laine et al. 2006, Fransson et al. 2008, Gatti et al. 2008, Cury et al. 2009, De Boever et al. 2009, Zetterqvist et al. 2010), while one study stated to have performed examiner calibration, but did not report any details (Roccuzzo et al. 2010). Another study reported an assessment of intra-examiner validity of the measurements giving the value for the error measured without adding further details (Maximo et al. 2008). Three articles reported in detail on the assessment of intra- and inter-examiner reproducibility (Ferreira et al. 2006, Roos-Jansåker et al. 2006b, Koldsland et al. 2011).

Regarding the internal validity of the assessment of potential risk factors, only one article reported data on reproducibility of factor measurements (Ferreira et al. 2006).

Data analysis

The type of statistical or data analysis used varied greatly. Half of the articles adopted multivariate logistic regression models (Gruica et al. 2004, Ferreira et al. 2006, Laine et al. 2006, Roos-Jansåker et al. 2006b, Fransson et al. 2008, De Boever et al. 2009, Karbach et al. 2009). Only one article compensated for intra-individual dependency of the variables assessed by using a multilevel regression model with the implant as the lower and the subject as the highest level (Koldsland et al. 2011). One of these articles analysed data from one implant within each subject, without specifying the selection criteria (Karbach et al. 2009). Chi-square tests were used in seven studies to evaluate differences between groups in terms of prevalence (Ferreira et al. 2006, Laine et al. 2006, Gatti et al. 2008, Maximo et al. 2008, Cury et al. 2009, Roccuzzo et al. 2010, Zetterqvist et al. 2010).

Six articles reported odds ratio values with 95% confidence intervals (Gruica et al. 2004, Ferreira et al. 2006, Laine et al. 2006, Roos-Jansåker et al. 2006b, Karbach et al. 2009, Koldsland et al. 2011). Gatti et al. (2008) did not perform any statistical calculations to determine risk due to too few events occurring. Six articles reported only p-values of the investigated factors (Fransson et al. 2008, Maximo et al. 2008, Cury et al. 2009, De Boever et al. 2009, Roccuzzo et al. 2010, Zetterqvist et al. 2010).

Results

Several articles were able to correlate a history of periodontitis with an increased risk of peri-implant diseases (Ferreira et al. 2006, Roos-Jansåker et al. 2006b, Gatti et al. 2008, Maximo et al. 2008, De Boever et al. 2009, Roccuzzo et al. 2010). Similar findings were described for smoking (Gruica et al. 2004, Roos-Jansåker et al. 2006b, Fransson et al. 2008, Karbach et al. 2009). One study, however, failed to demonstrate such a correlation (Koldsland et al. 2011). Several other potential risk factors were identified (Table 5).

Quality of reporting

Details on the adherence to the STROBE criteria are reported in Table 6. In total, the authors fulfilled between 48% and 78% of the applicable criteria. The study design was explicitly present in the title or the abstract in five (Gatti et al. 2008, Cury et al. 2009, De Boever et al. 2009, Roccuzzo et al. 2010, Zetterqvist et al. 2010) of the 13 studies, and was described in the material and methods section in three articles (Ferreira et al. 2006, Laine et al. 2006, Koldsland et al. 2011). Five papers (Gruica et al. 2004, Roos-Jansåker et al. 2006b, Fransson et al. 2008, Maximo et al. 2008, Karbach et al. 2009) did not describe or mention the design of the study at all (see Tables 5 and 6). Background and objectives were described in all of the articles. Neither of the two case-control studies (Laine et al. 2006, Cury et al. 2009) elaborated on matching criteria and on why the particular number of controls per case were selected. As mentioned above, only one article reported a description testifying efforts to address potential sources of bias (Roccuzzo et al. 2010). Six of the 13 studies explained how the study size was determined (Gruica et al. 2004, Roos-Jansåker et al. 2006b, Fransson et al. 2008, Maximo et al. 2008, Roccuzzo et al. 2010, Koldsland et al. 2011).

Table 6. Studies on risk factors of peri-implant diseases – STROBE adherence
STROBE criteria adherenceGruica et al. (2004)Ferreira et al. (2006)Laine et al. (2006)Roos-Jansåker et al. (2006b)Fransson et al. (2008)Gatti et al. (2008)Maximo et al. (2008)Cury et al. (2009)De Boever et al. (2009)Karbach et al. (2009)Roccuzzo et al. (2010)Zetterqvist et al. (2010)Koldsland et al. (2011)Adherence
Title and abstract 1 (a)NoNoNoNoNoYesYesNoYesYesNo5/13YesNo
1 (b)YesYesYesYesYesYesYesYesYesYesYes13/13YesYes
IntroductionBackground2YesYesYesYesYesYesYesYesYesYesYes13/13YesYes
Objectives3YesYesYesYesYesYesYesYesYesYesYes13/13YesYes
MethodsStudy design4NoYesYesNoNoYesYesNoYesYesYes8/13YesNo
Setting5YesNoNoYesYesYesNoNoYesYesYes9/13YesYes
Participants6(a)YesYesNoYesYesYesNoYesYesYesYes11/13YesYes
6(b)NANANoNANANoNANANANANA0/2NANA
Variables7YesYesYesYesYesYesYesYesYesYesYes13/13YesYes
Measurements8YesYesYesYesYesYesYesYesYesNoYes12/13YesYes
Bias9NoNoNoNoNoNoNoNoYesNoNo1/13NoNo
Study Size10YesNoNoYesYesNoNoNoYesNoYes6/13NoYes
Quantitative var.11YesYesYesYesYesNAYesNAYesNAYes10/10YesYes
Statistical methods12(a)YesYesYesYesYesYesYesYesYesYesYes13/13YesYes
12(b)YesYesYesYesYesYesYesYesYesNoYes12/13YesYes
12(c)NoNoNoNoNoNoNoNoNoNoNo0/13NoNo
12(d)NoNANANANANANoNAYesYesNo2/7NoNo
12(e)NoNoNoNoNoNoNoYesNoNANo2/11NANo
Results Participants13(a)YesYesNoYesYesNoYesNoYesYesYes9/13NoYes
13(b)NoNoNoYesYesNoNoNoYesYesYes7/13YesYes
13(c)NoNoNoNoNoNoNoNoNoNoNo0/13NoNo
Descriptive data14(a)NoYesYesYesYesYesYesYesNoYesYes11/13YesYes
14(b)NoNoNoNoNoNoNoNoNoNoYes1/13NoNo
14(c)NoNANAYesNANANoNAYesYesNA4/6YesNA
Outcome data15YesYesYesYesYesYesNoYesYesYesYes12/13YesYes
Main results16(a)YesYesYesYesYesNoNoYesNoNAYes7/10NoNo
16(b)NAYesNAYesNANANoNAYesNAYes5/6NAYes
16(c)NoNoNoNoNoNANANoNANANo0/9NoNo
Other analyses17YesYesNoNoYesNoYesYesYesNoYes9/13YesYes
DiscussionKey results18YesYesYesYesYesYesYesYesYesYesYes13/13YesYes
Limitations19NoYesNoYesNoYesYesYesNoNoYes8/13YesYes
Interpretation20YesYesYesYesYesYesYesYesYesYesYes13/13YesYes
Generaliz-ability21YesYesNoYesYesYesNoYesNoNoYes9/13YesYes
Other infoFunding22YesNoYesYesNoYesYesNoNoYesYes7/13NoNo
Adherence19/3220/3115/3123/3220/3021/3121/3218/2917/3217/2922/3218/2825/32Total: 63.8%

A thorough description of eligibility of subjects was given in 11 articles, while seven gave reason for non-participation at each stage (see Table 6). One study reported details on participants with missing data (Koldsland et al. 2011), two studies described the sampling strategy (Roccuzzo et al. 2010, Zetterqvist et al. 2010) and only one publication performed and described a sensitivity analysis (Karbach et al. 2009). All of the studies but one (De Boever et al. 2009) reported the outcome in terms of summary measures.

A discussion of limitations of the study was present in eight articles (Ferreira et al. 2006, Roos-Jansåker et al. 2006b, Gatti et al. 2008, Maximo et al. 2008, Cury et al. 2009, De Boever et al. 2009, Karbach et al. 2009, Koldsland et al. 2011). Generalizability was discussed in nine of the papers (see Table 6). Six articles did not explicitly mention the source of funding (Ferreira et al. 2006, Fransson et al. 2008, Gatti et al. 2008, Maximo et al. 2008, Karbach et al. 2009, Roccuzzo et al. 2010) while all discussed possible interpretations of the findings.

Discussion

A limited number of articles aiming at assessing prevalence, incidence or risk factors for peri-implant diseases could be identified and included in the present review. The aim was to assess the quality of reporting in the existing literature by applying the STROBE criteria. None of the 16 included articles fully adhered to the STROBE criteria.

The applied terms for the electronic search were chosen to potentially include all relevant literature. In the current review, only studies applying appropriate research design regarding the questions addressed were considered. For the evaluation of research on the incidence/prevalence of peri-implant diseases, studies of cross-sectional and longitudinal design were considered. For the evaluation of research on risk factors, cross-sectional, longitudinal, prospective and case-control studies were eligible. Furthermore, data analysis on subject-level rather than implant-level was a prerequisite for a study to be included. These criteria are in line with the consensus reached at the 6th EWOP (Lindhe & Meyle 2008). We did, however, not set any limits regarding minimum follow-up time nor minimum number of subjects included, which is in contrast to what has been done in previous reviews on biological complications around implants (Berglundh et al. 2002, Zitzmann & Berglundh 2008). Here, a minimum follow-up period of 5 years and the inclusion of more than 50 subjects were the requirements for inclusion. Despite the comparably less strict inclusion criteria we only included 16 studies in our review after having first found 306 in the initial search and studying 40 titles in full text.

Along with the evaluation of quality of reporting, we included an evaluation of the research methodology described in the selected articles (Tables 1, 3 and 5). Despite an abundance of existing critical appraisal tools of research reports none has so far been accepted as a “gold standard”, regardless of the study design (Katrak et al. 2004). Therefore, we did not select a specific evaluation tool to describe the methodologies used in the different studies. Instead, we assessed the main methodological issues common to all evaluation tools.

Out of the 16 articles that fulfilled the inclusion criteria, only two reported on incidence, while five reported on prevalence and 13 on risk factors of peri-implant diseases. Although the number of studies initially identified was high, the large majority had to be excluded. Exclusion was mostly due to (i) the type of analysis applied, for example, implant-based data as opposed to subject-level data, (ii) a lack of frequency analysis of peri-implant diseases and (iii) inappropriate study design.

Dental implant therapy has traditionally been evaluated using implant-level data. It has since been suggested that, from an epidemiological standpoint, peri-implant diseases should be assessed applying subject-level data (Zitzmann & Berglundh 2008). At the 6th EWOP it was stated that the prevalence should ideally be assessed through cross-sectional studies of appropriate size and follow-up time. Both, the assessment of incidence and risk factors, require longitudinal, prospective or case-control studies (Lindhe & Meyle 2008). Clearly, these requirements have so far not been met by the majority of articles addressing the issue.

An obvious shortcoming in the current literature on biological complications is the inconsistency of peri-implant disease criteria applied as we identified a total of nine different definitions of peri-implantitis throughout the selected studies. The variety of disease definition is similar to what has been found in the periodontal literature (Savage et al. 2009). Only two research groups (Fransson et al. 2005, 2008, Koldsland et al. 2010, 2011) applied the definition suggested at the 6th EWOP (Zitzmann & Berglundh 2008). It is evident that this lack of uniformity in the definition used in epidemiological and risk factor studies must lead to different results that are not easily comparable. In fact, it has been demonstrated that outcomes in terms of prevalence do vary significantly if different thresholds of disease are chosen (Koldsland et al. 2010).

All studies selected for this review are based on convenience samples. Such sampling methods are sensitive to selection bias, particularly in studies of cross-sectional and case-control design (Kleinbaum et al. 1981, Patten 2000). Subjects should ideally be included based on random selection (Fowkes & Fulton 1991), preferably having being treated in different environments. Thus, the requirement of random selection has so far not been met in the scientific literature.

It is reasonable to assume that incidence, prevalence and severity of peri-implant diseases are influenced by different factors that are related to the subject, that is, smoking, plaque control, systemic diseases, or to the implant, that is, implant surface, surface topography. Function time, however, seems to have limited influence on disease development (Fransson et al. 2010, Koldsland et al. 2011). To explore the relation between covariates and outcome, advanced statistical techniques, such as multi-level modelling, are required (Goldstein et al. 2002).

It is important to separate what was done in a study from how the study is reported. Good reporting is crucial but does not necessarily reflect a low susceptibility to bias (Huwiler-Muntener et al. 2002, Sanderson et al. 2007). It does, however, allow the reader to assess all relevant details of a performed study. Our assessment of the quality of reporting in the selected articles was based on the STROBE criteria. The STROBE Statement is a checklist of items that should be addressed in articles reporting on the three main study designs of analytical epidemiology: cohort, case-control, and cross-sectional studies. It was primarily developed to guide authors but may also serve to critically appraise published articles (von Elm et al. 2008).

While some criteria were fulfilled by virtually all studies, there were several which only a minority or none of the articles lived up to, suggesting significant room for improvement in terms of quality of reporting. One of the most obvious shortcomings was the description of the study design as only five studies clearly and early (title or abstract) included a description, as suggested by the STROBE criteria. Background and objectives of the studies were usually well described.

One of the keys to research in general and observational studies in particular is the selection of participants. Thus, tools of study quality assessment apply most of their items on methods of selecting study participants (Sanderson et al. 2007), as does the STROBE checklist. The majority of studies in this review did describe eligibility and selection of participants. The findings were somewhat different for the two case-controls studies included (Laine et al. 2006, Cury et al. 2009). The choice of cases and controls is particularly crucial, and the method of their selection has major implications for study validity (Vandenbroucke et al. 2007a,b,c). Surprisingly, neither of the two clearly elaborated on the selection of their controls, thus not addressing potential sources of bias. In fact, risk of bias was rarely addressed as only one article reported a description testifying efforts to address potential sources of such (Roccuzzo et al. 2010).

Concerning description of statistical analysis, investigators should predetermine techniques at least for the primary study objectives in a study protocol. Procedures for variable selection should be fully described and not only presented as results in the final model (Altman et al. 1983, Clayton & Hill 1993). Statistical methods should be described “with enough detail to enable a knowledgeable reader with access to the original data to verify the reported results” (International Committee of Medical Journal Editors 2009). These demands have not been consistently met in the selected literature as, for instance, the handling of missing data is not addressed by a single author. Only one article reported details on participants with missing data regarding the studied variables (Koldsland et al. 2011). In addition, it is only one article that describes a sensitivity analysis (Karbach et al. 2009).

One of the main objectives of research is to control for and discuss the potential of systematic bias. Hence, it is crucial to assess the validity of measurements made in a research study (Fowkes & Fulton 1991). In this respect, it is noteworthy that only 8 of the included studies performed and reported internal validation while all 16 articles discuss the external validity of the study results (Tables 1-6). Hence, there exists a discrepancy between the reporting of internal and external validity of results. Contrary to the limited reporting on internal validity and discussion of potential bias, limitations and generalizability of the results were presented by the majority of studies.

It must be kept in mind, however, that the STROBE guidelines were only recently published. The most recent published articles showed efforts to adhere to consensus disease definition and reporting guidelines.

Collectively, the findings of this review indicate a need for improved reporting of epidemiological studies on peri-implant diseases. Other authors have suggested that the quality of reporting in observational studies is poor and that there has been little improvement during the last couple of years (Groenwold et al. 2008). Such studies need to be appropriately designed regarding the research question and then need to be reported in an equally appropriate way, potentially using the STROBE criteria as a guideline.

Conclusions

Two studies aimed at assessing incidence, while five aimed at assessing prevalence.

Thirteen studies evaluated potential risk factors of peri-implant diseases.

All studies evaluated in this review employed convenience samples.

Sample sizes were generally limited and observation times were in most instances below 10 years.

The criteria used to characterize peri-implant diseases varied greatly between studies.

Internal validation was reported on in half of the studies included.

None of the scrutinized articles adhered fully to the STROBE criteria but virtually all studies met more than half of the criteria.

The most recently published articles showed efforts to adhere to consensus disease definition and reporting guidelines.

Ancillary