Photo and Plasma Activation of Dental Implant Titanium Surfaces. A Systematic Review with Meta-Analysis of Pre-Clinical Studies

Background: Ultraviolet (UV) and non-thermal plasma functionalization are surface treatment modalities that seem able to improve osseointegration. The aim of this systematic review and meta-analysis is to assess the effect of the two methods and possible differences. Materials and Methods: The systematic research of pre-clinical animal studies was conducted up to May 2020 in the databases PubMed/Medline, Scopus and the Cochrane Lybrary. A meta-analysis was performed by using the DerSimonian–Laird estimator in random-effects models. Results: Through the digital search, 518 articles were identified; after duplicate removal and screening process 10 papers were included. Four studies evaluating UV treatment in rabbits were included in the meta-analysis. The qualitative evaluation of the included studies showed that both UV photofunctionalization and non-thermal plasma argon functionalization of titanium implant surfaces might be effective in vivo to improve the osseointegration. The meta-analysis on four studies evaluating UV treatment in rabbits showed that bone to implant contact values (expressed as standardized mean differences and raw mean differences) were significantly increased in the bio-activated groups when follow-up times were relatively homogeneous, although a high heterogeneity (I2 > 75%) was found in all models. Conclusions: The present systematic review and meta-analysis on pre-clinical studies demonstrated that chair-side treatment of implants with UV or non-thermal plasma appear to be effective for improving osseointegration. This systematic review supports further clinical trials on this topic.


Introduction
Prosthetic rehabilitation with dental implants represents a successful therapy for the replacement of severely compromised or missing teeth, with long-term success rates above 95% [1]. The success of dental implants largely depends on a safe integration into the maxillary bone, or rather achieving stable osseointegration over time [2]. Osseointegration is considered a demarcation response to a foreign body of Ti when the Ti implant is immobile in bone [3]. This demarcation is immune-driven and is classified as a type IV hypersensitivity [4]. Based on the original definition, the modification of a Ti implant surface implies that the surface would be more biocompatible, thereby increasing the bioaffinity of the hard tissue and accelerating the bone response to the surface. In order to improve the biological response to Ti implants, various techniques have been suggested in order to modify Ti surface roughness, chemistry, topography, and electrical charge, focusing on the biological performance of Ti surfaces [5][6][7] Higher bone-implant contact (BIC) values, better bone apposition and peri-implant bone maintenance over time were demonstrated on implants with rougher surfaces compared to machined surfaces, including stimulation of cell migration and proliferation [6][7][8]. BIC values in modern implants normally vary between 65 and 73% but do not reach the ideal 100% [9]. Moreover, titanium surface modifications showed, in short-term evaluations, an enhanced connective fiber attachment and a similar inflammatory response [10,11].
In recent years, various methods (ultraviolet (UV), non-thermal plasma functionalization, blasting, etching, and anodization) were proposed to improve the hydrophilicity of the titanium surface, its functionality and in order to improve its chemistry as well as reduce surface contamination [5,12]. Among them, (UV) and non-thermal plasma functionalization are surface treatment modalities that are able to improve all the biological aspects mentioned above and can be applied chair-side [11,13]. They allow maintaining the micro-and nano-topographical features of titanium surfaces, enhancing the biological potential before implant surgery, without damaging the surface topography. Various in vitro studies about UV and non-thermal plasma functionalization showed that this method is able to increases hydrophilicity, turning the electrostatic charge to positive, and removing hydrocarbons from the surface. Positive effects are induced also on alkaline phosphatase activity, calcium deposition, spreading of human stem cells, protein absorption capacity, osteoblast migration, attachment, spread, and proliferation [13][14][15][16][17].
The aim of this systematic review was to assess in pre-clinical animal studies the effect of UV photofunctionalization or non-thermal plasma functionalization on the osseointegration of dental titanium implants.

Material and Methods
This systematic review fully adhered to the guidelines of the preferred reporting of systematic reviews and meta-analyses (PRISMA) statement [18] and the protocol was registered on Prospero (CRD42020185209). The proposed focused question was: "What is the effect of UV photofunctionalization and non-thermal plasma of argon activation of titanium dental implants on osseointegration in animals?" The focused question was established according to the PICO strategy: • Population: Healthy animals with at least one titanium dental implant. • Intervention: Any surface activation with UV or non-thermal plasma. • Comparison: Any type of "non-activated" titanium dental implant. • Outcomes: Bone-to-implant contact (BIC), implant stability quotient (ISQ) or removal torque.

Search Strategy
An electronic search in PubMed ® /MEDLINE, Scopus, and the Cochrane Central Register of Controlled Clinical Trials (CENTRAL) databases was performed starting April 2020. No publication year nor language limit was applied, so that the search could include all the available papers until 21 May 2020. The search was complemented by manual searches of the reference lists of all full-text articles selected. The following search terms were used: "photofunctionalization", "photofunctionalized", "ultraviolet(s)", "uv", "plasma(s)", "argon", and "dental implants". More details on research queries and Boolean operators are available in the electronic supplementary material (Table S1).

Eligibility Criteria
All articles on animals that presented a test group and a control group were considered eligible. Studies were required to have recruited a minimum of five healthy animals, have at least one titanium dental implant and have a minimum follow-up of two weeks. The studies were required to have compared implants treated with UV or non-thermal argon plasma (test group) and untreated implants. In addition, the studies to meet the inclusion criteria were required to have assessed the outcomes of interest (BIC value, removal torque and ISQ). Studies that did not met all the above-mentioned inclusion criteria were excluded. Review studies, retrospective studies, report studies based on questionnaires and interviews, studies without a clinical evaluation, case reports and redundant articles, studies on mini-implants and/or for orthodontic anchoring, and genomic and/or epigenomic analysis studies were also excluded.

Selection of Studies
Two reviewers (P.P. and E.D.G.) independently read titles and abstracts of the entries yielded from the initial electronic database search. After this initial assessment, both reviewers read separately the full-text versions of the studies that could be potentially included in this systematic review. The final selection of articles was made on the basis of the eligibility criteria described above. Any disagreement in the final selection was resolved by open discussion between both reviewers. In the case that no agreement could be reached, one of the co-author (L.C.) acted as arbiter.

Data Extraction
Data from the studies included in the final selection were extracted by one of the authors (E.D.G) using the Microsoft Excel spreadsheet software (Excel 16.4, Microsoft CO, Redmond, WA, USA). The accuracy of data was verified independently by another coauthor (P.P.). The following data were extracted were study design, title, author, publication year, follow-up period, number of patients (animals) and implants, implant design and surface characteristics, and the outcomes of interest (BIC, ISQ and removal torque). If data were missing, the authors of the original articles were contacted and asked to provide further details.

Quality Assessment
The overall quality of evidence at the outcome level in animal studies was independently assessed by two authors (E.D.G and P.P.), according to SYRCLE's RoB tool (Systematic Review Centre for Laboratory Animal Experimentation). Risk of bias in individual studies was assessed independently and in duplicate by the two coauthors as part of the data extraction process. This evaluation was conducted using the Cochrane-recommended approach for assessing risk of bias in animal intervention studies [19], including ten quality parameters: sequence generation, baseline characteristics, allocation concealment, random housing, blinding, random outcome assessment, incomplete outcome data, selective outcome reporting, and other sources of bias. Disagreements were discussed in order to aim for consensus. Each parameter was rated as: yes (low risk of bias); no (high risk of bias); or unclear (unclear risk of bias).

Statistical Analysis
The statistical heterogeneity among studies was expressed as τ 2 and estimated by the Cochran's Q test. The I 2 was calculated to assess variability due to heterogeneity rather than chance (I 2 = 25%: low; I 2 > 25% and = 50%: moderate; I 2 > 50% and = 75%: considerable; I 2 > 75%: high heterogeneity). H 2 was calculated as the ratio between total and sampling variability. Maximum likelihood (ML) and restricted maximum likelihood (RML) with Akaike information criteria (AIC) were returned for model fit statistics. The estimates of the effects were expressed as standardized mean difference (SMD) or raw mean difference (RMD). Study estimates were pooled with the random effects model and the DerSimonian-Laird estimator. In the random-effects models, the selected studies and their outcomes are assumed to be a random selection from a larger population of studies. A forest plot was created for each measured outcome to illustrate the effects in the meta-analysis of the different studies and the global estimation. Contour-enhanced funnel plots with Kendall's Tau and Egger's regression were used for publication bias assessment. For further evaluation of residual heterogeneity, a normal quantile-quantile (Q-Q) plot was evaluated. Statistical significance was assumed in each test with P value < 0.05. Statistical analysis was carried out by using the R software (version 3.6.3; R Foundation for Statistical Computing. Vienna, Austria) with the metafor package (version 2.1-0).

Bibliographic Search and Study Selection
The initial database search yielded a total of 518 entries; of which, 220 were found in PubMed ® /MEDLINE, 285 in Scopus, and 13 in Cochrane Library. A flow chart that depicts the screening process is displayed in Figure 1. After excluding all duplicates, the total number of entries was reduced to 407. A total of 368 articles were excluded after review of title and abstract. Hence, full-text examination was conducted for 39 articles. A total of 28 additional articles were excluded after full-text review and application of the eligibility criteria. The final selection consisted of 10 articles, of which 4 were included in the meta-analyses. Detailed data for the 10 included studies are listed in Tables 1  and 2. Four authors were contacted to obtain missing information, and only one of them [20] answered. J. Clin. Med. 2020, 9, x FOR PEER REVIEW 4 of 19

Bibliographic Search and Study Selection
The initial database search yielded a total of 518 entries; of which, 220 were found in PubMed ® /MEDLINE, 285 in Scopus, and 13 in Cochrane Library. A flow chart that depicts the screening process is displayed in Figure 1. After excluding all duplicates, the total number of entries was reduced to 407. A total of 368 articles were excluded after review of title and abstract. Hence, full-text examination was conducted for 39 articles. A total of 28 additional articles were excluded after full-text review and application of the eligibility criteria. The final selection consisted of 10 articles, of which 4 were included in the meta-analyses. Detailed data for the 10 included studies are listed in Tables 1 and 2. Four authors were contacted to obtain missing information, and only one of them [20] answered.

Description of Included Studies
Detailed data for the 10 included studies are listed in Tables 1 and 2. All studies included in the present review are studies performed on test case animals, with a study group and a control group. Four studies were performed on rabbits [20,26,27,29] and one on rats [28]. Five studies were performed on beagle dogs [21][22][23][24][25] The studies included in the review showed wide variations regarding length of the follow-up. The majority of them had a follow-up of a few weeks only and only two had a follow-up up to 12 weeks [25,27].

Excluded Studies
Out of the 39 papers for which the full-text was analyzed, 29 articles [13, were excluded from the systematic review (Table A1). The main reasons for exclusion were the following: small sample size, (articles that recruited less than five animals); outcome, (articles that did not evaluated BIC value, removal torque, or ISQ); and specimens, (articles that did not evaluated titanium dental implants).

Quality Assessment
According to SYRCLE's RoB tool (Systematic Review Centre for Laboratory Animal Experimentation), risk of bias of animal studies is assessed and displayed in Figure 2 [19]. No article showed low risk of bias for all domains.

Description of Included Studies
Detailed data for the 10 included studies are listed in Tables 1 and 2. All studies included in the present review are studies performed on test case animals, with a study group and a control group. Four studies were performed on rabbits [20,26,27,29] and one on rats [28]. Five studies were performed on beagle dogs [21][22][23][24][25] The studies included in the review showed wide variations regarding length of the follow-up. The majority of them had a follow-up of a few weeks only and only two had a follow-up up to 12 weeks [25,27].

Excluded Studies
Out of the 39 papers for which the full-text was analyzed, 29 articles [13, were excluded from the systematic review (Table A1). The main reasons for exclusion were the following: small sample size, (articles that recruited less than five animals); outcome, (articles that did not evaluated BIC value, removal torque, or ISQ); and specimens, (articles that did not evaluated titanium dental implants).

Quality Assessment
According to SYRCLE's RoB tool (Systematic Review Centre for Laboratory Animal Experimentation), risk of bias of animal studies is assessed and displayed in Figure 2 [19]. No article showed low risk of bias for all domains.

Qualitative Assessment of Outcomes
Differences in bone-to-implant contact (BIC) values between test and control groups were evaluated in eight studies [20,[22][23][24][25][26][27][28]. In these studies, the functionalization treatment of the implant surfaces, both by UV and by non-thermal plasma, led to better BIC results than the control group. Only one study in rabbits reported lower BIC values in the test groups (UV-treated) than in the control group [29]. In fact, in the study by Sanchez-Perez et al., at 8 weeks, a minimal variation in the BIC values between the test group and the control group was noted, with BIC values of 26.835% for the control group and 24.225% for the test group. Differences in ISQ values between test group and control group were evaluated in one study [25], in which ISQ values were higher in the test group at 8 and 12 weeks of follow-up compared to the group control. Differences in removal torque values

Qualitative Assessment of Outcomes
Differences in bone-to-implant contact (BIC) values between test and control groups were evaluated in eight studies [20,[22][23][24][25][26][27][28]. In these studies, the functionalization treatment of the implant surfaces, both by UV and by non-thermal plasma, led to better BIC results than the control group. Only one study in rabbits reported lower BIC values in the test groups (UV-treated) than in the control group [29]. In fact, in the study by Sanchez-Perez et al., at 8 weeks, a minimal variation in the BIC values between the test group and the control group was noted, with BIC values of 26.835% for the control group and 24.225% for the test group. Differences in ISQ values between test group and control group were evaluated in one study [25], in which ISQ values were higher in the test group at 8 and 12 weeks of follow-up compared to the group control. Differences in removal torque values between the test group and control group were evaluated in two studies [20,22]. In all studies the removal torque (Ncm) values were higher in the test groups than in the control groups.

Quantitative Assessment of Outcomes
The meta-analysis was conducted only if at least three articles presented the same treatment method for functionalization of the implant surface, and the same kind of sample and the same outcome of interest (BIC value) was evaluated. Following these criteria, four articles were included in the quantitative evaluation [20,26,27,29], as reported in the PRISMA flow chart (Figure 1). Based on the peculiarities of the follow-up recorded in some of the studies meeting the enrollment criteria (follow-up at different time-points in two studies, with slightly different numbers of implants), the analysis was conducted on multiple datasets identified on the basis of the different follow-up periods in order to even the follow-up time-points, applying the criterion of n = 1 study with unique follow-up for a single dataset. Following this approach, four datasets were assessed. The first dataset (Table S2)

First Dataset
The first dataset included 33 animals and 84 implants (Table S2) In the normal Q-Q plot, no study was outside the confidence area ( Figure S5A). The forest plot is presented in Figure 3. Publication bias assessment returned no statistical significance for Kendall's Tau (p = 0.333) or Egger's regression (p = 0.076) ( Figure  S9A). The results of the same model with RMD as effect size are presented in the Supplementary  Material (Figures S1, S5B and S9B).
between the test group and control group were evaluated in two studies [20,22]. In all studies the removal torque (Ncm) values were higher in the test groups than in the control groups.

Quantitative Assessment of Outcomes
The meta-analysis was conducted only if at least three articles presented the same treatment method for functionalization of the implant surface, and the same kind of sample and the same outcome of interest (BIC value) was evaluated. Following these criteria, four articles were included in the quantitative evaluation [20,26,27,29], as reported in the PRISMA flow chart (Figure 1). Based on the peculiarities of the follow-up recorded in some of the studies meeting the enrollment criteria (follow-up at different time-points in two studies, with slightly different numbers of implants), the analysis was conducted on multiple datasets identified on the basis of the different follow-up periods in order to even the follow-up time-points, applying the criterion of n = 1 study with unique followup for a single dataset. Following this approach, four datasets were assessed. The first dataset (

First Dataset
The first dataset included 33 animals and 84 implants (Table S2)   In the normal Q-Q plot, one study resulted outside the confidence area ( Figure S6A). The forest plot is presented in Figure 4. Publication bias assessment returned no statistical significance for Kendall's Tau (p = 0.750), differently from Egger's regression (p < 0.001) ( Figure S10A). The results of the same model with RMD as effect size are presented in the Supplementary Material (Figures S2, S6B and S10B one study resulted outside the confidence area ( Figure S6A). The forest plot is presented in Figure 4. Publication bias assessment returned no statistical significance for Kendall's Tau (p = 0.750), differently from Egger's regression (p < 0.001) ( Figure S10A). The results of the same model with RMD as effect size are presented in the Supplementary Material (Figures S2, S6B and S10B).

Third Dataset
The third dataset included 33 animals and 84 implants (Table S4) In the normal Q-Q plot, no study was outside the confidence area ( Figure S7A (Supplementary Materials)). The forest plot is presented in Figure 5. Publication bias assessment returned no statistical significance for Kendall's Tau (p = 0.333) or Egger's regression (p = 0.181) ( Figure S11A). The results of the same model with RMD as effect size are presented in the Supplementary Material (Figures S3, S7B and S11B).

Third Dataset
The third dataset included 33 animals and 84 implants (Table S4) one study resulted outside the confidence area ( Figure S6A). The forest plot is presented in Figure 4. Publication bias assessment returned no statistical significance for Kendall's Tau (p = 0.750), differently from Egger's regression (p < 0.001) ( Figure S10A). The results of the same model with RMD as effect size are presented in the Supplementary Material (Figures S2, S6B and S10B).

Discussion
This systematic review focused on the evaluation of the effects of the functionalization treatment of the implant surfaces by means of UV or non-thermal Plasma on the peri-implant bone, and, more specifically, it investigated the effects on the biological process of osseointegration, as evaluated by BIC, ISQ (implant stability quotient), and removal torque. The eligibility criteria of the studies were first determined so as to include in the systematic review studies with the following: a minimum number of five animals, which had at least one titanium dental implant; a test group subjected to treatment with UV or non-thermal plasma and a control group; analysis the BIC value with a minimum follow-up of two weeks; and ISQ and/or removal torque. A qualitative analysis of the results was performed on all included studies, and a quantitative analysis was performed on 4 of the 10 included studies.

Summary of the Results and Possible Limitations
The qualitative analysis of the results shows for the majority of the studies, higher BIC (bone-toimplant contact) values in the test groups subjected to the functionalization treatment of the implant surfaces by UV or by non-thermal plasma. With regard to studies that used UV treatment [20,[26][27][28], BIC values were better in the test group when compared to the control group at all the time points. Two studies that used treatment with non-thermal plasma [21,23] did not find significant differences in BIC between the test and the control group at the first follow-up (1 week), but they found significant differences in BIC values between the test and the control group at the second follow-up (3 weeks) with better values for the test group.

Discussion
This systematic review focused on the evaluation of the effects of the functionalization treatment of the implant surfaces by means of UV or non-thermal Plasma on the peri-implant bone, and, more specifically, it investigated the effects on the biological process of osseointegration, as evaluated by BIC, ISQ (implant stability quotient), and removal torque. The eligibility criteria of the studies were first determined so as to include in the systematic review studies with the following: a minimum number of five animals, which had at least one titanium dental implant; a test group subjected to treatment with UV or non-thermal plasma and a control group; analysis the BIC value with a minimum follow-up of two weeks; and ISQ and/or removal torque. A qualitative analysis of the results was performed on all included studies, and a quantitative analysis was performed on 4 of the 10 included studies.

Summary of the Results and Possible Limitations
The qualitative analysis of the results shows for the majority of the studies, higher BIC (bone-to-implant contact) values in the test groups subjected to the functionalization treatment of the implant surfaces by UV or by non-thermal plasma. With regard to studies that used UV treatment [20,[26][27][28], BIC values were better in the test group when compared to the control group at all the time points. Two studies that used treatment with non-thermal plasma [21,23] did not find significant differences in BIC between the test and the control group at the first follow-up (1 week), but they found significant differences in BIC values between the test and the control group at the second follow-up (3 weeks) with better values for the test group.
The third study evaluating non-thermal plasma treatment [24] showed higher BIC values in the test group than in the control group at both 4 and 8 weeks of healing. Only one study [29] reported slightly higher BIC values for the control group than for the UV-treated test group. Despite this, there is a greater homogeneity of the BIC in the test group compared to the control group.
ISQ values were evaluated in a single animal study [25], demonstrating higher ISQ values in the test group subjected to functionalization treatment with non-thermal argon plasma compared to the control group.
Removal values were evaluated in two studies: one [20] evaluating UV functionalization treatment, and one [22] using functionalization treatment with non-thermal argon plasma. In all these three studies, the removal torque (Ncm) values were higher in the test groups than in the control groups.
The meta-analysis was conducted for four studies [20,26,27,29], in which the test groups were subjected to UV treatment. The assessed outcome was the BIC value. A quantitative analysis of the studies in which the test groups underwent treatment with non-thermal plasma, presenting the same outcome of interest, was not possible due to the lack of data. For the studies included in the quantitative evaluation, following the differences in the follow-up, an analysis was conducted on several datasets identified on the basis of the different follow-up periods without prejudice to the criterion of n = 1 study, and with unique follow-up for each individual dataset [26]. The meta-analysis showed statistically significant difference in favor of the test groups in the first and third dataset, for both SMD and RMD as effect size. No significant differences occurred in the second and fourth dataset. The relatively more homogeneous follow-up of the first and third dataset may have contributed to the model outcomes. However, in all models, a high heterogeneity was found, although the highest I 2 and Q values occurred for the second and fourth dataset. Normal Q-Q plots showed overall satisfactory profiles, with all or most of the studies falling within the confidence region in each model. Notably, no significance for publication bias was returned by the models evaluated by entering the first and third dataset. Data from the present meta-analysis confirmed the importance to increase surface energy to stimulate bone formation through the exposition to UV. However, a similar effect was obtained also through the activation of the electronic mantel following plasma of argon bio-activation [49]. Additionally, it must be noted that despite the increase in surface energy, these procedures were proven to not affect the bacterial contamination of the implant surface (or bone augmentation material), confirming to finally result positive for bone growth [58][59][60][61][62].
Limitations can be attributed to the high heterogeneity of the studies, the applied methodology, (including numerosity of the samples, especially in studies on UV treatment), and type of samples tested (dental implants or disks), etc. Different types of devices and time of application were used to functionalize the surfaces of the implants. Three studies [20,26,27] used a source of UV radiation for 24 h; two studies [28,29] used a UV-emitting device for 15 min. However, even if a different application time was employed, all studies obtained better results in the UV-treated groups. The minimum irradiation time required to obtain a clinically appreciable effect is yet to be determined [29]. Three studies [21][22][23] used a plasma device (KinPen Device) for 60 s or for 20 s. The third study [24] used a plasma reactor (Diener electronic) for 12 min. The fourth study [25] used a Plasma reactor (Line through ISO 9001) for 60 s. Further studies should evaluate the better application time of UV and Plasma in order to standardize the technique.
Another limitation is the definition of osseointegration itself, which is a clinical outcome and not a histological one. Surrogates for osseointegration are used to conduct research, for example, BIC and ISQ. These are measures of specific elements of the bone-implant interface; however, there is little evidence that BIC correlates with long-term clinical outcomes of dental implant therapy, and the use of ISQ, although widely used, remains enigmatic [63].
Further limitations are secondary to the incomplete publication of data. BIC and standard deviation values are not reported in three studies [20,21,23] The authors of the aforementioned articles were contacted to supplement the missing data; however, only the BIC values of a single article were obtained [20].
From the clinical point of view, both methods can be used chair-side by the clinician before implant insertion.
Both the investigated techniques of functionalization are easy to apply, cost effective and are devoid of contraindications.
This study and data from animal experiments presented promising outcomes for UV and non-thermal plasma functionalization. An improvement in osseointegration might be expected following biofunctionalization of dental implants. However, it must be emphasized that these results must be taken with caution, as data from animal studies cannot be directly extrapolated to the clinical practice, and the clinical efficacy of these treatments is yet to be established. Human studies are needed to confirm if biofunctionalization of dental implants might affect the bone-implant interface in the short and in the long term.

Conclusions
Based on the qualitative and quantitative assessment conducted as part of the present systematic review, it can be concluded that the treatment of titanium dental implant surfaces using UV or non-thermal plasma may represent an effective method for improving the osseointegration process. Randomized human studies are needed to validate the obtained results.
Supplementary Materials: The following are available online at http://www.mdpi.com/2077-0383/9/9/2817/s1, Table S1: Research query and their outputs; Table S2: First dataset entered in the meta-analysis; Table S3: Second dataset entered in the meta-analysis; Table S4: Third dataset entered in the meta-analysis; Table S5: Fourth dataset entered in the meta-analysis; Figure S1: Forest plot for the random-effects model with the DerSimonian-Laird estimator in the first dataset (effects size: raw mean difference of the bone-implant contact); Figure S2: Forest plot for the random-effects model with the DerSimonian-Laird estimator in the second dataset (effects size: raw mean difference of the bone-implant contact); Figure S3: Forest plot for the random-effects model with the DerSimonian-Laird estimator in the third dataset (effects size: raw mean difference of the bone-implant contact); Figure S4: Forest plot for the random-effects model with the DerSimonian-Laird estimator in the fourth dataset (effects size: raw mean difference of the bone-implant contact); Figure S5: Normal Q-Q plots related to the random-effects models with the DerSimonian-Laird estimator in the first dataset; Figure S6: Normal Q-Q plots related to the random-effects models with the DerSimonian-Laird estimator in the second dataset; Figure  S7: Normal Q-Q plots related to the random-effects models with the DerSimonian-Laird estimator in the third dataset; Figure S8: Normal Q-Q plots related to the random-effects models with the DerSimonian-Laird estimator in the fourth dataset; Figure S9: Funnel plots for the publication bias assessment related to the random-effects models with the DerSimonian-Laird estimator in the first dataset; Figure S10: Funnel plots for the publication bias assessment related to the random-effects models with the DerSimonian-Laird estimator in the second dataset; Figure S11: Funnel plots for the publication bias assessment related to the random-effects models with the DerSimonian-Laird estimator in the third dataset; Figure S12: Funnel plots for the publication bias assessment related to the random-effects models with the DerSimonian-Laird estimator in the fourth dataset.
Funding: This research received no external funding.