Clinical efficacy of early loading versus conventional loading of dental implants

The aim of this study was to determine the clinical differences between early and conventional loading protocols for dental implants. A comprehensive search of the Medline, Embase, and OVID databases for studies published through January 10, 2015 was conducted. Fourteen studies were included in our analysis. We found that early loading imposed a significantly higher risk of implant failure than did conventional loading (risk ratio = 2.09, 95% confidence interval [CI] [1.18, 3.69], P = 0.01), while no significant differences between the methods were found with regards to the marginal bone loss (weighted mean differences [WMD] = 0.11, 95% CI [−0.07, 0.28], P = 0.23), periotest value (WMD = 0.02, 95% CI [−0.83, 0.87], P = 0.96), or implant stability quotient (WMD = 0.79, 95% CI [−0.03, 1.62], P = 0.06). As for the health status of the peri-implant tissue, conventionally loaded implants demonstrated better performance than did early loaded implants. Subgroup analyses demonstrated that the sample size, time of publication, loading definition, implant position, extent, and restoration type influenced the results. Although early implant loading is convenient and comfortable for patients, this method still cannot achieve the same clinical outcomes as the conventional loading method.

study has reported the clinical differences between early and conventional loading. Therefore, the aim of this study was to determine whether or not early loading resulted in different clinical outcomes than conventional loading in terms of the implant failure rate, marginal bone loss, implant stability, peri-implant parameters, and complications.

Focused question.
We hoped to answer the following question by developing a protocol: does early implant loading produce different results than conventional loading in terms of the implant failure rate, marginal bone loss, implant stability, health of the peri-implant soft tissue, and complications. This question was developed by using the following PICOs definitions: • Population: patients who need at least one implant, older than 18 years, having adequate oral hygiene, having adequate alveolar bone volume to place implants, and in good general health to permit implant surgery. • Intervention: early loading of implants (test group) was defined as implants loaded between 1 week and 2 months after insertion. • Comparison: conventional loading of implants (control group) was defined as implants loaded more than 2 months after insertion; also referred to as delayed loading 10 . • Outcomes: (1) the primary outcome was the implant failure rate, defined as the rate of mobility and removal of implants 11  Search strategies. An electronic search limited to English publications with abstracts was conducted using the Medline, Embase, and OVID databases, and included all studies published through January 10, 2015.
The following search terms and strategies were used: (dental implant OR oral implant) AND (early loading OR early restoration OR early prostheses OR early overdenture) AND (conventional loading OR conventional restoration OR conventional prostheses OR conventional overdenture) AND (delayed loading OR delayed restoration OR delayed prostheses OR delayed overdenture).
The reference lists of identified studies and relevant reviews on the same subject were also searched for other potential studies.
Inclusion and exclusion criteria. The inclusion criteria were as follows: (1) human studies, (2) clinical comparative studies, (3) studies comparing the implant failure rates in any group of patients who received early and conventionally loaded implants, (4) studies providing data regarding the relevant clinical outcomes, (5) studies with follow-up periods of at least 6 months after loading, and (6) studies published in English.
Furthermore, we used the following exclusion criteria: (1) in vitro or laboratory studies, (2) animal studies, (3) reviews, (4) case reports, (5) craniofacial implant studies, and (6) studies only focusing on the different clinical efficacy between implants loaded after less than 1 week (immediately loaded implants) and implants loaded conventionally. Study selection. Two reviewers (Y. Zhu and Y. Xu) screened the titles and abstracts independently for relevant subjects. Studies that could not be excluded definitively based on the information from the abstracts were searched using the full manuscript to determine whether they should be included. If an agreement could not be reached regarding inclusion/exclusion between the two reviewers, a discussion was held with a third reviewer until an agreement was reached (X. Qu).
Data extraction. Two reviewers (Y. Zhu and Y. Xu) used a specially designed extraction form to extract data independently. Any disagreement about the data extraction was discussed with a third reviewer (X. Qu). The percentage of agreement and Kappa analyses were calculated to evaluate the inter-reviewer reliability of data extraction. The most complete data were extracted when the results of a study were published more than once or presented in different publications. The following information was extracted: first author's name, country, year of publication, study design, follow-up period, number and age of patients, loading protocol, number and placement position of implants (maxilla or mandibular), extent of restoration (full or single implant), type of restoration (fixed or removable), type Scientific RepoRts | 5:15995 | DOi: 10.1038/srep15995 of implant system, and data on dental implant failure, marginal bone loss, ISQs, PTVs, and peri-implant parameters (AL, PD, BI, GI, PI). Quality assessment. The recommendations by Cochrane 12 were used to assess the risk of bias.
Studies were classified as having low risk of bias if they met the following criteria: (1) random allocation, (2) clear selection criteria for subjects, (3) blinding of the patients and examiners, (4) detailed reporting of the outcome data, (5) selection of representative population groups, and (6) use of identical treatments between two groups except for the intervention. If a study missed one of these criteria, it was considered to have a moderate risk of bias. Missing two or more of these criteria led to a high risk of bias 13 . Data analysis. For dichotomous variables (implant failure rates), the risk ratio (RR) and 95% confidence interval [CI] were calculated to compare the selected studies. For continuous variables (marginal bone loss, ISQ, PTV, and peri-implant parameters), the weighted mean difference (WMD) and 95% CI were used.
A fixed-effects model was used as a common measure for study-specific estimates, while a random-effects model was considered if significant heterogeneity was found 15 . The level of statistically significant heterogeneity was set at P < 0.1. In addition, subgroup analyses were conducted whenever possible to identify associations between the risk of implant failure, marginal bone loss, and other relevant elements as possible sources of heterogeneity. The subgroup analyses used the following explanatory variables: (1) length of follow-up, (2) sample size (implants), (3) design of study, (4) geographical location, (5) time of publication, (6) definition of early loading, (7) definition of conventional loading, (8) extent of restoration, (9) implant placement position, and (10) type of restoration.
Visualization of funnel plots, Begg's test, and Egger's regression tests were measured to assess publication biases 16 . Forest plots were drawn to demonstrate the effects in the meta-analyses. Stata version 11 (StataCorp, College Station, TX) and Review Manager 5.2 (The Cochrane Collaboration, Copenhagen, Denmark) were used to conduct the statistical analyses.

Results
Literature Search. Figure 1 provides a flow chart of the literature search process. The search yielded a total of 1474 primary papers. Of these, 1360 papers were discarded by the two reviewers, leaving 114 papers after the titles and abstracts were evaluated (inter-rater agreement = 86%, kappa = 0.87). An additional 17 papers were included after searching the references of relevant studies, resulting in 137 papers that required full-text evaluation. After the full-text evaluation, 23 papers that reported data from 14 investigations remained. The data from four papers reporting on three investigations were repeated 17-23 and five papers from two investigations reported different results on the same material in different years [24][25][26][27][28][29][30] . Therefore, nine papers were excluded, leaving 14 papers in the final analyses. One of the included papers was a three-armed study that utilized two experimental groups based on the different definitions of early loading 31 . Another of the included papers performed two controlled experiments depending on the different types of implant systems 32 . These two papers were divided into two parts when calculated.
Study characteristics. The detailed study characteristics are presented in Table 1. In the 14 studies, 1619 implants were completely inserted, while 1265 implants remained after years of follow-up (681 implants in the test group, 584 implants in the control group). The number of implants in each study ranged from 20 33 to 490 20 . The earliest study was published in 2000 34 , whereas the latest was published in 2012 30 . The follow-up period of the studies ranged from a minimum of 6 months 33 to a maximum of 10 years 31 . As for the study design, six studies enrolled patients prospectively 18,33-37 , seven were randomized controlled trials (RCTs) 23,26,[30][31][32]38,39 , and one was a retrospective study 20 . In terms of the geographic locations, four studies were conducted in Asia 23,30,33,37 , seven in Europe 18,20,31,[34][35][36]39 , and three in Oceania 26,32,38 .
The group definitions and prosthetic characteristics are presented in Supplementary Table S1. In both groups, different definitions were used. Ten studies defined loading within 3 weeks as early loading 18,20,23,26,30,31,[33][34][35][36] , while the other five studies placed the prostheses within 6 weeks 31,32,37-39 . One study had two experimental groups: one group had the implants loaded within 3 weeks and the other group had the implants loaded within 6 weeks 31 . The majority of studies in the conventional group loaded implants 3-4 months after insertion, except one, which loaded the implant after 6 months 37 .
Marginal bone loss. The marginal bone level in the early loading group ranged from 0.17 to 2.55 mm at baseline, from 0.97 to 2.87 mm at 1 year after loading, from 1.25 to 2.86 mm at 3 years, and from 1.61 to 2.98 mm at 5 years after loading; in the conventional loading group, the marginal bone level ranged from 0.11 to 3.5 mm at baseline, from 0.90 to 3.30 mm at 1 year after loading, from 1.26 to 3.5 mm at 3 years, and from 1.57 to 3.70 mm at 5 years after loading.
The results of the meta-analysis regarding the overall marginal bone loss are presented in Fig. 3. Eleven studies that compared early and conventionally loaded implants reported marginal bone loss. The results illustrated that no significant difference was found in the overall marginal bone loss between these two loading protocols (WMD = 0.11, 95% CI [− 0.07, 0.28], P = 0.23). Seven studies included marginal bone loss at 1 year after loading 18,20,23,31,32,36,38 . The results illustrated that early loaded implants had more marginal bone loss than did conventionally loaded implants after 1 year of loading (WMD = 0.18, 95% CI [0.00, 0.35], P = 0.05). Five studies included marginal bone loss at 2-3 years after loading 30,31,34,36,39 . No significant difference was found between the two loading types after 2-3 years of loading (WMD = 0.26, 95% CI [− 0.15, 0.68], P = 0.21). Three studies reported the marginal bone loss at over 5 years after loading 26,30,31 . Again, no significant difference was found between the two groups after over 5 years of loading (WMD = 0.10, 95% CI [− 0.10, 0.29], P = 0.33).
Subgroup analyses showed that the early loaded implants in the maxilla had a higher risk of marginal bone loss than did the implants in the mandibular region, while other elements had little significant influence on the marginal bone loss. The results from the subgroup analyses and meta-analysis on cumulative marginal bone loss are presented in Table 3.  Tawse Implant stability. Clinical implant stability was measured by two methods: the Ostellt device, which measured the resonance frequency of a small transductor that was attached to the implant abutment and that was stimulated using a wide range of frequencies (ISQ); and the Periotest device, which measured the temporal contact of the tip of the instrument during repetitive percussions on the implant (PTV) 40 . Two studies were excluded since they only reported medians and ranges 35,38 . Therefore, PTVs were collected from four studies [32][33][34]36 . The mean values varied from − 4.9 to − 1.1 in the early loading group and from − 4.52 to − 3 in the conventional group. No statistically significant difference was found between the two groups (WMD = 0.02, 95% CI [− 0.83, 0.87], P = 0.96).
In the four studies that reported ISQs 23,26,30,37 , the mean values varied from 66.6 to 81.9 in the test group and from 65 to 81.2 in the control group. Two studies were excluded because they lacked standard deviations 38,39 . The results showed a greater change in ISQ in the test group compared to in the control group, but this difference was not statistically significant (WMD Peri-implant tissue health status. The peri-implant parameters were obtained from three studies 30,32,38 . Five different indexes were recorded to estimate the peri-implant tissue health status: AL, PD, BI, GI, and PI. The results showed that conventionally loaded implants had a better peri-implant tissue health status than did the early loaded implants in terms of their AL (WMD = 0. 43 Table 2. Subgroup analysis to investigate differences between studies included in meta-analysis for failure rate.   Table 3. Subgroup analysis to investigate differences between studies included in meta-analysis for MBL (mm). MBL = marginal bone loss.  However, there was an increase in these outcomes in the test group. The forest plots of these results are presented in Supplementary Figures S1-S5.

Complications.
Four studies reported technical complications 20,23,26,37 . The most common complications were fractures of the prostheses and loosening of the abutment screws. Technical complications occurred more often in the test groups according to the records from the included studies. All of these complications were solved by adjusting the implants or prostheses without affecting the outcomes. One study noted that denture contouring was the most common adjustment. The number of patients needing denture contouring adjustments was higher in the test group(15 people) than it was in the control group (12 people). This could be explained by the secondary gingival healing after surgery in the early loading group, which may result in space around the abutments, while relining impressions performed in the conventional loading group after a period of healing avoided this space 23 . Biological complications were reported in three studies 20,26,33 . The most common complication was peri-implant inflammation, which eventually returned to normal.
Two studies reported no complications in either group 32,34 . None of the remaining studies mentioned any complications.
Quality assessments. The results of the quality assessments of the included studies are presented in Supplementary Table S2. Of the 14 studies, one was considered to have a low risk of bias 39 , while three had a moderate risk of bias 31,32,38

Discussion
Immediate loading, early loading and conventional loading are the three main loading protocols in implantology. A recent meta-analysis has already comprehensively compared immediately loaded implants to conventionally loaded implants 9 . The results from that study demonstrated that immediate loading may impose a higher risk for implant failure compared to conventional loading. While early loading protocols (implants loaded 1 week to 2 months after insertion) permit a longer healing time than immediate loading protocols (implants loaded less than 1 week after insertion) and the efficacy difference between early loading and conventional loading protocols is uncertain, so the present meta-analysis, which utilized 14 different studies that compared early vs. conventional loading protocols with data from 1619 implants, aimed to establish whether early loading can achieve the same clinical outcomes as conventional loading with shorter time.
The results of this meta-analysis indicate that early loading has a higher risk of failure compared with conventional loading, which is not totally in agreement with the conclusions published recently in similar systematic reviews. The results from two recent meta-analyses demonstrated that the data were insufficient to determine whether or not there was a clinically important difference in implant failure or marginal bone loss between the two groups based on the limited number of available studies 10,41 . The results from two other meta-analyses did not show any significant differences between early and delayed loading 42,43 .
These discrepancies can be explained by the relatively small number of available studies together with their small sample sizes, which are not sufficient to establish definitive conclusions. Moreover, studies with a follow-up period of 4-12 months were included in a recent Cochrane systematic review, which was considered as not long enough to assess implant failure 10 . Therefore, the present meta-analysis included studies with a follow-up period of 6 months to 7 years.
Although recent systematic reviews have also studied the different failure rates between early and conventional loading 10,41-43 , the limitation of these studies was that insufficient data were available to perform comparative analyses of prosthodontic and peri-implant outcomes with reliable clinical results. In the present study, we enlarged the search scale and updated the search time frame. Thus, to our knowledge, the present study is the most comprehensive and newest meta-analysis estimating the clinical difference between early and conventional loading protocols in terms of implant failure, marginal bone loss, ISQ, PTV, and peri-implant parameters. The possible influence of the outcomes was also assessed through subgroup analyses involving the length of follow-up, sample size (implants), design of the study, geographical location, time of publication, definition of early loading, definition of conventional loading, extent of restoration, implant placement position, and type of restoration. The current meta-analysis also calculated the cumulative marginal bone loss in years, which gave a continuous view of the differences Scientific RepoRts | 5:15995 | DOi: 10.1038/srep15995 these two loading protocols caused over time, while other recent systematic reviews only presented the results of implant failure and marginal bone loss 10,[41][42][43] .
Despite the strengths above, this meta-analysis still has several limitations. First, it was difficult to completely eliminate the confounding factors inherent in the included studies that may have resulted in a bias to the outcomes 20 . Second, the I 2 estimates of marginal bone loss (I 2 = 97%; P < 0.001) and PTV (I 2 = 93%, P < 0.001) were evaluated as high. Heterogeneity introduced by methodological differences between the included studies could not be avoided. Therefore, subgroup analyses were conducted on studies of marginal bone loss, which indicated that the selected elements had no influence on the outcomes. The data on the PTV, however, were insufficient for a subgroup analysis. Third, the literature search was limited to studies in English and only covered three electronic databases, which might have resulted in an election bias to the outcomes. Considering the limitations mentioned above, the results of this meta-analysis should be interpreted with caution.

Conclusions
The primary outcome of this study was that early loading had a higher risk of implant failure than did conventional loading. The secondary outcomes revealed that the two loading protocols were not significantly different in terms of the associated marginal bone loss, changes in implant stability, and health status of the peri-implant tissues (except for the AL and PD), which indicated that these two loading protocols behaved similarly once osseointegration occurred. Although early implant loading is advantageous with regards to convenience and comfort for patients, this method still cannot achieve the same clinical outcomes as the conventional loading. Careful selection of patients and surgery indications is still required. These findings may provide dentists with additional insights when developing treatment strategies and making prognoses. Meanwhile, based on this review, more high-quality and well-designed RCTs with large sample sizes are needed.