Immediate versus early or conventional loading dental implants with fixed prostheses: A systematic review and meta-analysis of randomized controlled clinical trials

A prospective protocol was developed a priori according to the PRISMA (Preferred Reporting Items for Systematic Review and Meta-Analyses) and the Meta-analysis of Observational Studies in Epidemiology recommendations.^,  The PICOS strategy was used for the search: P (population)=patients requiring at least 1 dental implant; I (intervention)=restoration within 1 week of implant placement; C (comparison)=delayed (also termed “conventional”) loading defined as restorations 8 weeks after insertion, early loading between 1 and 8 weeks; O (outcome)=implant survival rate, marginal bone level changes, peri-implant gingival level, plaque index, probing depth, implant stability, the rate of peri-implantitis or peri-implant mucositis, and subjective feeling of patients; S (study design)=randomized controlled trials (RCTs). The focus question was “Is there a difference in postoperative outcomes when an immediate implant loading protocol is compared with early or conventional loading in fixed restoration(s)?”

From inception until October 2018, a comprehensive electronic search was conducted in CENTRAL (The Cochrane Central Register of Controlled Trials), EMBASE, and MEDLINE via PubMed (The National Library of Medicine). The search strategy is shown in Table 1, and the results filter was set to humans and randomized controlled trials.

There were no restrictions on regions or languages. The computer search was supplemented with a manual search of the reference lists in all retrieved literature. In addition, a search of the online databases in the following journals was performed: British Journal of Oral and Maxillofacial Surgery, Clinical Implant Dentistry and Related Research, Clinical Oral Implants Research, European Journal of Oral Implantology, Implant Dentistry, The International Journal of Oral & Maxillofacial Implants, International Journal of Oral and Maxillofacial Surgery, International Journal of Periodontics and Restorative Dentistry, International Journal of Prosthodontics, Journal of Clinical Periodontology, Journal of Dental Implantology, Journal of Dental Research, Journal of Oral Implantology, Journal of Oral and Maxillofacial Surgery, Journal of Periodontology, Journal of Periodontal Research, Journal of Prosthetic Dentistry.

Two reviewers (J.C., M.C.) selected studies by independently screening the titles and abstracts of search results based on the following inclusion criteria: at least 1 dental implant with a fixed prosthesis; at least 15 participants; studies on immediate loading versus early or conventional loading; at least 1 of those aforementioned outcomes reported; and randomized controlled trials (RCTs). There was no restriction on the follow-up period. Animal studies or studies involving zygomatic implants or implants used for orthodontic anchorage were excluded. Additionally, review studies, case reports, case series, and meeting abstracts were also excluded. The full text of potential articles was reconfirmed and evaluated for data extraction. Any disagreement was resolved by further discussion or an additional author's (Y.W.) evaluation. When multiple articles reported the same trial, the most recent one with completed data was included. Authors of studies were contacted by e-mail when data were found to be incomplete or not reported.

Two authors (T.A.A., J.Y.) independently assessed the risk of bias in the included studies. The quality assessment of the included RCTs was performed by using the Cochrane Collaboration's tool. Seven criteria were assessed: random sequence generation (selection bias), allocation concealment (selection bias), blinding of participants and personnel (performance bias), blinding of outcome assessment (detection bias), incomplete outcome data (attrition bias), selective reporting (reporting bias), and other bias. Studies were classified as low risk if all criteria were met, moderate risk if 1 criterion is missed, and high risk if 2 or more were missed.

For this meta-analysis, the dichotomous variables (such as, implant survival rate) were pooled and analyzed by using risk ratios (RRs) and 95% confidence intervals (95% CIs). As for continuous outcomes (such as, marginal bone level changes), weighted mean differences (WMDs) with 95% CI were used. The Q-test estimated heterogeneity, with significance set at α=.1 and quantified with the I² index (high heterogeneity: I²>75%; low heterogeneity: I²<25%). The random-effect model was used when significant heterogeneity was found between the test and control study. Otherwise, the fixed-effect model was applied. Subgroup analyses were carried out based on items listed in the introduction. All analyses were performed by using a statistical software program (STATA-12; StataCorp LP) (α=.05). Forest plots were used to illustrate the effects of the intervention, and funnel plots were created to screen for publication bias.

The screening process is depicted in Figure 1. Eighty-nine articles were selected for full-text analysis after the evaluation of titles and abstracts (agreement=87.4%; kappa=0.63). Forty-nine articles met inclusion criteria and were assessed for reliability (Table 2). After evaluation, 49 full-text articles that belonged to 39 trials were identified. Of the 39 studies, 18 belonged to 8 trials, which were divided into the following 8 series: In the first series, 2 articles reported the data at 2 years and 15 years, respectively.^,  In the second series, 3 articles showed results at varying time periods.^{,  ,}  In the third series, 2 articles reported the outcomes at 1 year and 5 years.^,  In the fourth series, 2 articles reported the data at 9 months and 4 years.^,  In the fifth series, 3 articles showed the results at 5 months, 1 year, and 3 years.^{,  ,}  In the sixth series, 2 articles reported the data at 1 and 3 years.^,  In the seventh series, 2 articles showed the results at 1 and 2 years.^,  In the eighth series, 2 articles reported the data at 1 and 3 years.^, 

Table 2 shows the methodological characteristics of the selected studies. Nine of 39 articles were split-mouth trials, and 30 studies were parallel studies. Seven of 30 parallel trials had 2 test groups that met the inclusion criteria; therefore, each comparison was regarded independently.^{,  ,  ,  ,  ,  ,  ,} 

This systematic review pooled data from 1868 participants (914 in a test group and 954 in control), and a total of 3746 implants were inserted (1880 in an experimental group and 1866 in control) at baseline. A total of 1785 participants were followed up (864 in the experimental group and 921 in the control group), and 3486 implants (1749 in the experimental group and 1737 in control group) were reported at the end of the trial. The maximum follow-up period was 180 months, and the minimum was 10 days

Figure 2 depicts the risk of bias for RCTs. Six studies showed a low risk of bias,^{,  ,  ,  ,  ,  ,}  8 trials showed medium risk of bias,^{,  ,  ,  ,  ,  ,  ,}  and the remaining showed high risk of bias. Funnel plots and the Begg test were used to detect publication bias.

According to previous systematic reviews, the patient and implant are regarded as statistical units in the current meta-analysis. The results of the 2 methods are as follows: for implant as a statistical unit, the mean survival rates were 96.8% in the test and 98.6% in the control group. In 9 of 29 included trials, the survival rate of implants was 100%.^{,  ,  ,  ,  ,  ,  ,  ,  ,}  The results from the meta-analyses of the remaining 20 studies are reported in Figure 3. The meta-analysis resulted in a statistically significant lower survival rate for the test (immediate loading) group compared with that for the conventional group (RR=0.974; 95% CI, 0.954, 0.994; P=.012). No publication bias was detected by the Begg test (P=.261; Fig. 4). In the subgroup analyses, a lower survival rate was shown in the immediately loaded implants than in conventional loading in regard to the following: nonsubmerged technique (RR=0.969; 95% CI, 0.946, 0.994; P=.013), delayed implant (RR=0.974; 95% CI, 0.953, 0.996; P=.020), occlusal contact (RR=0.969; 95% CI, 0.947, 0.992; P=.009), single missing tooth (RR=0.958; 95% CI, 0.921, 0.998; P=.038), several missing teeth (RR=0.955; 95% CI, 0.920, 0.991; P=.015), surgical guide stent (RR=0.953; 95% CI, 0.920, 0.988; P=.009), operative area not only restricted in the maxillary nonmolar or mandibular posterior region (RR=0.972; 95% CI, 0.951, 0.995; P=.015), flap operations in both groups (RR=0.972; 95% CI, 0.950, 0.994; P=.015), and interim prostheses used for immediate loading while definitive restorations placed in the control group (RR=0.966; 95% CI, 0.943, 0.991; P=.007) (Table 3).

For patient as a statistical unit, the mean survival rate was 95.0% in the test group and 97.3% in the control group. Of the 27 included studies, there was no implant failure in 8 studies.^{,  ,  ,  ,  ,  ,  ,  ,  ,}  The overall effect of the meta-analyses showed a higher rate of implant failure in the test group but without a statistically significant difference (RR=0.963; 95% CI, 0.927, 1.001; P=.059) (Fig. 5). No publication bias was detected by the Begg test (P=.780; Fig. 6). The subgroup analyses resulted in a higher rate of failure for immediate loading implants than for conventional loading implants in regard to the following conditions: a nonsubmerged technique was used (RR=0.951; 95% CI, 0.907, 0.997; P=.037), several missing teeth (RR=0.903; 95% CI, 0.820, 0.993; P=.036), surgical guide used (RR=0.921; 95% CI, 0.864, 0.983; P=.014), and interim prostheses used for immediate loading while definitive restorations were placed in the control group (RR=0.949; 95% CI, 0.905, 0.995; P=.030). In the immediately loaded group, a relatively higher failure rate was identified for delayed implant (RR=0.958; 95% CI, 0.913, 1.005; P=.081), occlusal contact (RR=0.948; 95% CI, 0.897, 1.003; P=.064), single missing tooth (RR=0.957; 95% CI, 0.911, 1.006; P=.087), operative area not only restricted in the maxillary nonmolar or mandibular posterior region (RR=0.949; 95% CI, 0.896, 1.004; P=.070), and flap operations in both groups (RR=0.961; 95% CI, 0.922, 1.001; P=.058), without statistically significant differences (P>.05) (Table 4).

The change of crestal bone level was reported in all except 5 trials.^{,  ,  ,  ,}  Most investigations used periapical radiographs except 1 using panoramic radiographs. To avoid high heterogeneity among studies, studies that used periapical radiographs to evaluate the crestal bone were combined in the independent meta-analyses. The loss of marginal bone level ranged from −1.32 mm (loss) to 0 mm in the test group and from −1.25 mm to −0.10 mm in the control group. The result shows no statistically significant differences in the crestal bone loss between the test and control groups (WMD=0.016; 95% CI, −0.052, 0.084; P=.645) when the data at all sites of implants were combined (Fig. 7). For any of the subgroup analyses, no statistically significant differences were found between groups, except in trials with occlusal contact (WMD=0.083; 95% CI, 0.003, 0.163; P=.043) and flapless operations in both groups (WMD=−0.3; 95% CI, −0.489, −0.111; P=.002), despite the high heterogeneity within some items (Table 5). For the study evaluating the change of marginal bone with panoramic radiography,^,  the differences in crestal bone loss between the 2 groups were not statistically significant (P>.05).

Implant stability was assessed using 2 methods: the implant stability quotient (ISQ) measured by resonance frequency analysis (RFA) with the Osstell device (Integration Diagnostics Ltd) and the implant Periotest (Siemens AG) value (PTV) with the periotest device. The Osstell was used in 7 trials. The data were reported as Figures in 2 studies,^,  whereas 1 reported the minimum and maximum values.^,  The remaining 4 studies presented the mean and standard deviation. However, 1 of them did not show the sample size. Based on the last 3 studies,^{,  ,}  the ISQ ranged between 69.4 and 77.1 in the test group and between 69.8 and 78.6 in the control group. Regarding the result of the meta-analysis, there was no statistically significant difference (WMD=−0.436; 95% CI, −1.469, 0.598; P=.409).

In the 3 studies reporting PTV,^{,  ,  ,}  the mean ranged between −1.8 and 4.07 in the test group and between −1.3 and 4.0 in control. The meta-analysis found insufficient evidence to determine whether there was a difference between immediate and delayed loading (WMD=−0.233; 95% CI, −0.707, 0.241; P=.335).

Gingival inflammation was reported in 12 studies, and 5 indexes were used. The percentage of sites with positive bleeding on probing (BOP [%]) was reported in 5 studies. However, in 1 study,^,  the author without exact data stated there was no significant statistical difference between the 2 groups. In the remaining 4 trials,^{,  ,  ,}  BOP (%) varied from 0% to 40% in the test group and 0% to 36% in the control group. Modified sulcus bleeding index (mBI) was presented in 4 trials. One^,  reported the change of mBI, and no significant statistical difference was shown in the Student t test (P>.05). In the other 3 studies,^{,  ,}  the mean of mBI varied from 0.5 to 1.3 and from 0.67 to 1.4 in the test and control groups, respectively. Gingival index (GI) was used in 2 studies.^,  The GI ranged from 0.29 to 0.32 in the test group and from 0.25 to 0.29 in the other group. For these 3 indexes, the result of the meta-analyses showed insufficient evidence to determine whether statistically significant differences existed.

For 2 studies reporting gingival bleeding time index (GBTI), no meta-analysis was performed because the mean of 1 study was 0 in both groups. However, the other study presented GBTI with a decreasing trend, reaching their lowest values at 12 months in both groups. Sulcus bleeding index (SBI) was used in only 1 trial. No statistically significant difference was found between the test and control subjects (P>.05).

For peri-implant gingival level, change in papilla, free gingiva, and keratinized mucosa were reported. For the height of papilla, 5 trials use the papilla index (PPI). Three of them reported the mean, standard deviation, and sample size in both groups.^{,  ,  ,}  The meta-analysis, with low heterogeneity (P=.751, I²=0.0%), presents insufficient evidence of statistically significant differences between the 2 groups (WMD=0.061; 95% CI, −0.169, 0.292; P=.602). One publication recorded the Jemt-index frequency, and there was no statistically significant difference (P>.05) for the 2 procedures. One^,  reported that papilla index at all sites, both groups combined, either remained unchanged (28.5%) or improved (63%). The other 3 studies use the height of interproximal papillae. However, 1 investigation^,  did not report the sample size, and the meta-analysis result of the remaining 2^,  showed insufficient evidence to determine the difference between the test and control procedure (WMD=0.078; 95% CI, −0.115, 0.271; P=.429).

Regarding free gingiva change, which was reported in 5 included studies, 3 investigations showed the recession of the mid-buccal site,^{,  ,  ,}  the change ranging from −0.67 mm to 0.06 mm in the test group and from −1.16 mm to −0.09 mm in the control. The meta-analysis, with high heterogeneity among trials, showed insufficient evidence with a statistically significant difference (WMD=0.204; 95% CI, −0.297, 0.704; P=.425). The other 2 studies measured the gingival recession from the crown margin to the gingival margin,^,  with a reference line connecting the highest free gingival margins of adjacent dentition. Evidence of a statistically significant difference was lacking between the test and control participants regarding gingival recession (P>.05).