Prosthetic Rehabilitation of the Posterior Atrophic Maxilla, Short (≤6 mm) or Long (≥10 mm) Dental Implants? A Systematic Review, Meta-analysis, and Trial Sequential Analysis

Naples Consensus Report Working Group A

Abstract

Purpose: 

To compare the clinical and patient-reported outcomes of ≤6-mm implants with those of ≥10-mm implants placed after both lateral and transcrestal sinus floor elevation.

Materials and Methods: 

Using PubMed (MEDLINE), EMBASE, and Cochrane, a literature search for randomized controlled trials was performed. All the outcome variables were evaluated through a quantitative meta-analysis, and the influence of other clinical covariates were determined with a metaregression. For the survival outcomes, trial sequential analysis (TSA) was performed to adjust results for type I and II errors and to analyze the power of the available evidence.

Results: 

After full-text reading, 12 studies were included in the analyses. No statistically significant difference was found after 3 years between the 2 study groups (P = 0.36). Short implants displayed fewer biological complications (P = 0.05), less marginal bone loss (MBL) from implant placement (P < 0.01), and reduced surgical time and treatment cost. However, long implants showed a statistically significant smaller number of prosthetic complications (P = 0.03). TSA confirmed the results of the meta-analysis, revealing that additional studies are needed due to low statistical power of the available evidence.

Conclusion: 

The placement of short implants is a predictable option in treating patients with maxillary atrophy up to a 3-year follow-up. Studies with a longer observational period are needed to study the long-term performance of these implants.

Following tooth extraction, the alveolar socket results in a series of adaptive, both vertical and horizontal, modifications.¹ These morphological changes in the bone, reducing bone height, often preclude the placement of standard implants while avoiding damage to anatomical structures, such as the Schneiderian membrane.

In the last 3 decades, maxillary sinus floor elevation has been the gold standard procedure for implant-directed arch reconstruction to correct posterior maxillary atrophy.² Schneiderian membrane elevation was introduced by Boyne and James² as a procedure that enables direct access to the lateral sinus wall, detachment of the membrane from the internal bony walls, and insertion of bone graft apical to the atrophied arch. With time, the placement of implants after lateral sinus floor elevation (lSFE) has been adapted as common practice, with reported 15-year survival rates up to 98.3% subsequent to implant loading.³ Unfortunately, despite the well-documented predictability of this technique, overall idealness may still be hindered by drawbacks such as graft failure, postoperative sinusitis, relative costliness, and lengthy intraoperative durations.^4–6 A rate of 1 to 5 incidence of complications pertinent to lSFE and subsequent implant placement was actually reported in 2008.⁷

In an attempt to reduce the number of complications after lSFE, Summers⁸ had introduced an alternative, less invasive technique termed the osteotome technique, now commonly known as transcrestal sinus floor elevation (tSFE). With this technique, membrane elevation is performed through an osteotomy, drilled within the arch past the alveolar crest, by exerting controlled pressure using instruments known as osteotomes.⁹ The principal advantage of using this procedure is avoiding a lateral window to the maxillary sinus; however, its drawback is the minimal bone gain compared to that seen with the lateral window approach.¹⁰

With advancing technology and enhanced knowledge of implant microstructure, short implants have evolved into a clinically viable alternative to sinus elevation and standard implant placement. This may be attributed to their ease of placement, the greater likelihood of avoiding advanced bone grafting procedures, and the positive outcomes demonstrated by clinical trials.^11–13 To the contrary, the high crown-implant ratio,¹⁴ the higher possibility of encountering low-density bone in the posterior maxilla¹⁵ and the expectedly faster disease progression when periimplantitis around a shorter fixture initiates (leading to earlier implant loss) are all critical when opting for this treatment option.

At present, a clear decision for short implant length has not surfaced in the literature, where authors have determined it to be ≤6 mm,¹⁶ ≤8 mm,¹⁷ and even up to ≤10 mm.¹⁸ Thus, the present article is focused on guiding clinicians to the most beneficial choice between 2 different, but equally valid treatment approaches based on a meta-analysis of randomized controlled trials (RCTs). The aim of this article was to compare survival rate, marginal bone level change, biological and prosthetic complication rates, surgical time, and cost effectiveness of short (≤6 mm) implants versus long (≥10 mm) implants placed after either lSFE or tSFE.

Materials and Methods

Study Registration

The review protocol was registered with the PROSPERO International Prospective Register of Systematic Reviews under the identification number CRD42018095614.

Reporting Format

At the time of summarizing and describing the search process results, the 27-item Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) statement was used.¹⁹ The Assessment of Multiple Systematic Reviews guidelines (AMSTAR) were followed to meet the predetermined standards of reporting systematic reviews.²⁰

PI(E)CO (Patient, Intervention [Exposure], Comparison, Outcome)

The focused question was elaborated following the PICO format,²¹ where

• P: Any patient receiving one or more short dental implant(s) (≤6 mm) and one or more long implant(s) (≥10 mm) with a tSFE or lSFE procedure evaluated for ≥12 months.
• I(E): Short implant (≤6 mm) placement in the maxilla.
• C: Long implant (≥10 mm) placement after a sinus floor elevation (lateral/transcrestal).
• O: Implant survival rate, prosthetic/biological complications, marginal bone loss, cost, and surgical time.

Focused Question

Do short (≤6 mm) implants perform and long dental implants (≥10 mm) placed after sinus lift procedure in patients with atrophic maxillae?

Information Sources and Search Strategy

Literature searches for publications were conducted electronically and manually by 2 independent reviewers (A.R. and I-C.W.), without language and date restriction. Three databases were used to conduct a computerized and systematic search for published articles until February 2018 using the following search terms: (1) MEDLINE: (Short[All Fields] AND Implants[All Fields]) AND ((“crowns”[MeSH Terms] OR “crowns”[All Fields] OR “crown”[All Fields]) AND Implant[All Fields] OR (“Ratio (Oxf)”[Journal] OR “ratio”[All Fields])), (2) EMBASE: Short Dental Implants AND (“clinical trial”/de OR “controlled clinical trial”/de OR “human”/de OR “longitudinal study”/de OR “multicenter study”/de OR “prospective study”/de OR “randomized controlled trial”/de OR “randomized controlled trial (topic)”/de), and (3) COCHRANE: Short dental implants AND Trials. Furthermore, an electronic screening of Medicine Gray Literature Report was performed to check for ongoing/unpublished trials, in addition to a manual search of periodontics/implantology-related journals from January 2017 to April 2018. The bibliography of all the articles reviewed in full-text was also screened to check for other available publications. Potential articles were examined in full-text by 2 authors (A.R. and S.B.), confirming their eligibility after discussion. All disagreements were resolved through discussion with a third reviewer (I-C.W.). The level of agreement between the reviewers regarding study inclusion was calculated using k statistics. In addition, all the systematic reviews and meta-analyses related to short implants were carefully examined for article identification.^17,22–37

Eligibility Criteria

Articles were included only when they met the following criteria: (1) RCTs involving human subjects receiving one or more short (≤6 mm) implant as the test group and (≥10 mm) implants placed after a sinus floor elevation procedure as the control group; (2) studies with a minimum follow-up period of ≥12 months. The used exclusion criteria were (1) studies with a follow-up of <12 months after implant loading; (2) preclinical animal studies, prospective cohort studies, retrospective studies, case series, case reports, and systematic reviews.

Data Extraction and Statistical Analysis

During the first step of data extraction, after removing duplicates coming from different databases, studies were discarded based on title and abstract screening. The final stage of screening involved full-text reading using a predetermined data extraction form to confirm the eligibility of each study based on the aforementioned inclusion and exclusion criteria. Data such as patient characteristics, treatment covariates, and clinical outcomes were independently extracted by 2 reviewers (A.R. and I-C.W.) and systematically analyzed. The primary outcomes reported in this meta-analysis included survival rate, prosthetic and biological complication rate, and marginal bone level changes after implant placement, which were divided by different follow-up marks (1- and 3-year) with subgroup analyses. To standardize the reporting quality, Mantel-Haenszel weighted risk ratios (RRs) and 95% confidence interval (CI) summarized for dichotomous data were based on the exact numbers of event reported in the RCTs, and mean difference (with 95% CI) of continuous data was calculated between groups. Significance level favoring either group was set for P < 0.05, in which RR >1 indicating a higher event rate of short implants than longer implants in an augmented maxillary arch. A continuity correction 0.5 with reciprocal proportion to the imbalanced group size was used to include the zero-event-in-both-arm studies. Outcome measures were divided into implant-level variables (survival rate) and patient-level variables (marginal bone loss and complication rates). Summary estimates of 3 main outcome variables were acquired with heterogeneity tests (χ² [Cochran Q] and I² statistics). In addition, a Trial Sequential Analysis (TSA) was performed for the outcome survival rate at both 1- and 3-year follow-ups. TSA was performed to adjust results for types I and II errors and analyze the power of the meta-analytic findings. The required information size (RIS) and alpha-spending monitoring boundaries were calculated setting type I error at 5% and type II error at 20% (power of 80%). For RIS calculation, incidence in both intervention and control arms were estimated according to results of the meta-analysis, and no heterogeneity correction was applied. Graphical evaluation was performed to figure out whether the cumulative Z-curve crosses the trial sequential monitoring boundary, the futility boundary, and the RIS threshold. Different clinical parameters reported in the studies were dissected to analyze the possible influence of a sinus floor elevation procedure (tSFE vs lSFE approach), short implant length (4 vs 5 vs 6 mm), and smoking habits (short/long ratio represented the percentage of smokers in the short vs long implant group; heavy smokers being excluded or included). Methods of prosthesis retention (cement- vs screw-retained) and the choice of splinting were inspected; however, they failed to reach a sufficient number of studies feasible for analyses.

Risk of Bias and Qualitative Assessment

The assessment of the risk of bias among the selected articles was performed by 2 independent authors (A.R. and S.B.) using the Cochrane Risk of Bias Tool for RCTs.³⁸ The evaluated articles were classified as having low, moderate, or high risks of bias depending on the quality of their methodology.

Results

Study Selection

During the initial screening, 1590 articles were included: 443 via PubMed (MEDLINE), 184 via Cochrane, and 926 via EMBASE, while an additional 37 articles were obtained through manual screening. Overall, after evaluation of title and abstracts, 47 articles were selected and 35 of these were discarded after full-text reading (Supplementary Table 1, Supplemental Digital Content 1, https://links.lww.com/ID/A5) leaving 12 articles to be included in the final analysis (Fig. 1). The k value for inter-reviewer agreement for potentially relevant articles was 0.87 (titles and abstracts) and 0.85 (full-text articles). The articles published by Toma et al and Schincaglia et al^39,40 reported the 1-year follow-up data from the same 3-year follow-up cohort of Pohl et al.⁴¹ Similarly, the publication by Felice et al⁴² is a 3-year follow-up of the study sample from the article of Pistilli et al.⁴³ Finally, the articles of Gastaldi et al^44,45 are the 3-year follow-up of the articles of Felice et al⁴⁶ and Pistilli et al.⁴⁷

Fig. 1: PRISMA flowchart describing the entire screening process (title, abstract screening, and full-text reading).

Quality Assessment

The results of the risk of bias assessment for the included RCTs following the recommendations of the Cochran Handbook for Systematic Reviews⁴⁸ are reported in Supplementary Figure 1, Supplemental Digital Content 2, https://links.lww.com/ID/A6. Overall, 5 studies^{6,44,45,49,50} were considered as having a moderate risk of bias, and 2 studies^41,42 exhibited a high risk of bias.

Publication Bias

All the publication bias assessed by Egger test is depicted in the funnel plots of Figure 2. In the total survival rate, no evidence of publication bias was observed based on the symmetry in the funnel plot and the results of the Egger linear regression method (intercept = −0.08, 95% CI: −1.18 to 1.0, with t = 0.18, df = 11, and P value = 0.43).

Fig. 2: Begg funnel plot for assessing publication bias. Evidence of publication bias are observed by the asymmetry in the funnel plot and confirmed by the significance result of Egger test.

The funnel plot showed noticeable asymmetry in the publication bias of studies measuring the marginal bone loss starting from implant placement, and the visual impression is confirmed by Egger test that yields a statistically significant P value of 0.004. On the other hand, the funnel plot of marginal bone loss measured starting from implant loading did not show similar publication bias (P = 0.20). Finally, Egger test and the funnel plot showed that prosthetic (P = 0.83) and biological complications (P = 0.48) both demonstrated no presence of publication bias.

Characteristics of the Included Articles

All included studies adopted the conventional loading protocol (loading after 4–5 months of healing) regardless of the long implant/augmentation or short implant group. One study used tSFE approaches in the SFE procedure, whereas the remaining 6 studies used the lateral window approach.^{6,41,42,45,49,50} Among 9 studies, 3^6,41,42 exclusively used 6-mm short implants, 2 studies only used 5-mm short implants,^45,50 1 study used 4-mm short implants,⁴⁹ and the remaining study used both 5- and 6-mm short implants.⁴⁴ The characteristics of the included studies are shown in Table 1.

Survival Rates

The total 1-year survival rate throughout the 9 RCTs was 98.1% and 95.1% in the short and long implant group, respectively. The 3-year survival rate, according to the results of 6 RCTs, was 97.8% for short implants and 98.8% for long implants. The results of the meta-analyses lacked statistical significance in both the 1- and 3-year subgroups between the short and long implants (RR = 1.01; 95% CI: 0.98–1.04 and RR = 0.99; 95% CI: 0.96–1.02, respectively in Mantel-Haenszel fixed model, I² = 0%) (Fig. 3, A and B). Results of the TSA confirm the results of the meta-analysis because the z-curve did not cross the trial sequential monitoring boundaries. In addition, an estimate RIS of 1146 and 1475 patients for 1- and 3-year follow-ups, respectively, was calculated (Supplementary Figure 2, Supplemental Digital Content 3, https://links.lww.com/ID/A7).

Fig. 3: Forest plots (RR) for survival rate comparing the short versus long implant study groups in (A) 1-year, (B) 3-year results. Mantel-Haenszel weighted RR <1 revealed a lower survival rate of short implants than long implants.

The surgical approach had no influence on the 1- and 3-year survival rates (P = 0.99 and 0.89) (Fig. 4, A). The difference of length in the short implant group did not impact the 1- and 3-year survival performance (P = 0.53 and 0.58). Metaregression for smoking short/long ratio and heavy smoker excluded/included also failed to show a significant difference in 1-year survival rate (P = 0.53 and 0.54). A similar nonsignificant result of smoking short/long ratio was also shown in the 3-year results (P = 0.39).

Fig. 4: Subgroup analyses for the effects of the surgical approach on (A) survival rate, (B) complication rate, and (C) marginal bone loss measured from implant placement, at 1 and 3 years. P < 0.05: Significant difference between augmentation and nonaugmentation groups.

Marginal Bone Loss

The short implant group showed less changes in marginal bone level at 1 year after implant placement in 7 of the included studies (difference in mean values = −0.14 mm, P < 0.001, I² = 51.6%, Q = 12.4) (Fig. 5, A). A similar result was also demonstrated in the 3-year follow-up studies, which showed the difference in mean values = −0.21 mm, favoring short implants with less MBL (P = 0.002, I² = 68.0%, Q = 15.6) (Fig. 5, B). When prosthetic loading was considered as a baseline (1 year follow‐up), short implants demonstrated statistically significant less marginal bone loss (P < 0.01) (Fig. 5, C). The difference in sinus floor elevation approaches did not reveal a significant influence in 1- and 3-year MBL difference (P = 0.77 and 0.82) (Fig. 4, C). MBL measured at 1-year recall showed significant difference within different short implant lengths (5 or 6mm) (P = 0.01). Conversely, 3-year results showed a difference between 5- and 6-mm lengths, with a difference in mean values of −0.40 and −0.07 mm, respectively (P < 0.01). The results of a metaregression showed that the smoking ratio between the short and long implant groups seemed to be affecting the MBL difference at 3-year results (coefficient = −0.06, Q = 4.00, P = 0.045); on the contrary, the 1-year did not show the significance (coefficient = −0.04, Q = 2.25, P = 0.13). Other smoking-related covariates showed no significant impacts on the MBL differences.

Fig. 5: Forest plots comparing marginal bone loss in short versus long implant study groups at 1 and 3 years, considering as baseline implant placement (5A,5B) and implant prosthesis loading (5C). Negative value in difference in mean values indicates more marginal bone loss in the long group.

Table 1: Characteristics of the Included Articles

Prosthetic and Biologic Complication Rates

At 1-year follow-up, prosthetic complication rates did not reveal a significant difference between short and long implant groups (RR = 0.58, 95% CI: 0.23–1.43, P = 0.24) (Fig. 6, A). However, at the 3-year follow-up, the difference reached statistical significance with P = 0.03 and RR = 2.66 (95% CI = 1.12–6.34), favoring the short implant group (more prosthetic complications) (Fig. 6, B). Conversely, the 1- and 3-year results of biological complication rates showed significantly higher incidences in the long implant group (RR = 0.26, 95% CI = 0.11–0.54, P = 0.003; and RR = 0.22, 95% CI = 0.05–1.02, P = 0.05) (Fig. 6, C and D).

Fig. 6: Forest plots (RR) for prosthetic (A and B) and biological (C and D) complication rates comparing the short implant with the long implant group at 1 and 3 years. RR >1 represent a higher complication rate of short implants compared with longer.

Surgical intervention style (tSFE vs lSFE) did not impact the prosthetic complication rates at 1 and 3 years (P = 0.78 and 0.93) (Fig. 4, B). In terms of biological complications, it remained a statistically nonsignificant influence on the complication rates at the 1- and 3-year follow-ups (P = 0.54 and 0.78). Prosthetic complication rates showed no difference in the stratified analyses in different lengths (P = 0.63 and 0.99). Among the 1-year studies reported biological complication rates, implant length displayed no impact on the incidence of biological complications, with only 6-mm implant reaching significance and with RR = 0.19 (P = 0.01) in the individual subgroup analysis. Similar results, short of statistical significance, were detected in the 3-year results of biological complication rates, with only the 6-mm implant group displaying significance and RR = 0.08 (P < 0.001).

Prosthetic complication rates were not related to smoking short/long ratio at the 1- or 3-year follow-ups (P = 0.95 and 0.90, respectively), and other smoking-related covariates cannot be analyzed because of insufficient data. Similarly, smoking short/long ratio failed to reach significance in the impact on the biological complication rates at 1 and 3 years (P = 0.38 and 0.66, respectively).

Treatment Cost and Surgical Time

A discrepancy, between the 2 study groups, in the duration and cost of surgery was only mentioned in 2 of the included studies.^6,40 With regard to the cost of treatment, the first study⁴⁰ demonstrated the mean price for extra-short implant placement as 941 EUR (Euro) versus the 1946 EUR for the mean SFE and long implant placement price. The second study⁶ reported a close difference between the control (short implant) and test (SFE and long implant) groups with mean prices of 700 EUR versus 1322 EUR, respectively. Both studies have shown a nearly 2-fold increase in price for the procedures including SFE (P < 0.01).

The mean duration of the implant placement procedure in the former study⁴⁰ was observed to be 52.6 minutes for the extra-short implants versus 74.6 minutes for the long implant group, with a control to test group ratio of 50%. In the latter study,⁶ the mean time was 19.1 minutes for extra-short implant placement versus 32.2 minutes for SFE and long implant placement, with a comparable ratio of 59%. The differences in both the aforementioned studies were statistically significant (P < 0.05 and 0.01, respectively).

Discussion

Evidence-based health care guides the operator to clinical decision-making based on the results of scientific research. Among the various types of study designs, a meta-analysis of RCTs is considered the highest level of scientific evidence. This is due to its ability to increase the sample size derived from different studies and access subgroup-level queries that may not be otherwise investigated in other study designs.

In the present meta-analysis, by only including RCTs that compare short implants versus long implants inserted after an SFE (tSFE or lSFE), 2 treatment approaches to restore partial edentulism in the posterior maxilla have been compared. The selection of ≤6- and ≥10-mm implants is to emphasize the length difference between the 2 groups to a minimum of 4 mm. This, in turn, reproduces a common clinical scenario of partial edentulism and limited bone height in the posterior maxilla. In fact, previous meta-analyses comparing either ≤8 versus ≥9.3³⁶ or ≤8 versus >8 mm^29,51 may be unable to detect a clinically significant discrepancy within the included sample of investigations.

After the 1- and 3-year follow-up marks, short implants demonstrated comparable performance to long implants. This result is not a novel finding, confirming those reported by previous meta-analyses that have included single implants only,⁵² as well as single and splinted implants together.^29,36,51 The general short observational period present in most of the included investigations resulted in a high consistency of findings across the study sample. This was primarily assumed because of the lack of heterogeneity in survival rate (I² = 0%) and the forest plot's visibly limited dispersion. However, despite the great deal of design similarity between the studies and their comparable results, conclusions can only be drawn as a short-term comparison of the 2 treatment approaches. In addition, as revealed by TSA, the number of patients included in the studies remains low, yielding low power for the evidence obtained.

Long-term results (5 years) derived from RCTs do exist and do not favor short implant placement, reporting lower survival rates than long implants at 5 years. However, these were excluded from this article because of either long implants being placed in native bone^53,54 or SFE only performed when necessary and not uniformly throughout the study group.⁵⁵ Hence, a long-term comparison between short implant and long implant placed after sinus lift until now remains unfeasible, and it is crucial in constructing an absolute verdict. Despite this, it must be noted that even if not directly compared with short implants, excellent long-term outcomes have been widely reported for implants placed subsequent to SFE.^3,56

With regards to biological complications, long implants exhibited a higher number of adverse events than short implants. In fact, the entire set of included studies experienced a greater number of complications during the SFE, excluding one⁵⁰ that reported a higher incidence of sinus perforation in the short implant group. However, it should be noted that most of these complications were transient and did not lead to the loss of the implants.

For example, in the article by Bechara et al,⁶ only a single complication (sinus infection), of 19, led to implant loss since the remaining 18 were minor complications such as intraoperative bleeding (3) and postoperative pain and swelling (15). Furthermore, Schneiderian membrane perforation was another common complication, being reported by Gastaldi et al 5 times in 20 patients and by Felice et al 4 times in 10 patients.^42,45 These results reflect the prevalence of membrane perforation during SFE reported in the literature, which ranged from 7% to 44%,^57–59 most likely caused by sharp edges and ridges often encountered as septa or bony spines in the sinus.⁶⁰ As seen in a study by Froum et al,⁶¹ membrane perforation is not seemingly an adverse complication because it does not directly influence implant survival rate.

Contrarily, although rare in both groups, a significantly greater number of prosthetic complications, such as crown chipping⁴⁴ and screw loosening,^41,45 were associated with short implants. Screw loosening in implants with single crowns is not uncommon finding, reaching 28.3% on 6-mm soft-tissue level implants.¹² However, this is a much higher rate than that reported by most studies. For instance, Mezzomo et al,³² similar to the observations of the present review, reported a prosthetic complication rate of only 2.8% in short implants.

Finally, marginal bone loss from implant placement was lower in the short implant groups of nearly all the included articles. This difference between implants placed in augmented sinus floors and implants placed in pristine bone has been observed in previous investigations.⁶² This observed difference, discussed by Galindo-Moreno et al, can be explained by the contrasting biomechanical properties of the native bone versus the introduced bone graft. With less stiffness and a higher modulus of elasticity in grafted bone sites, a higher level of stress is accentuated around the periimplant bone crest, resulting in increased bone loss. This finding concurred with a previous meta-analysis comparing short and long implants, reporting more marginal bone loss in the latter group.³⁶

As previously discussed, despite the presence of 12 well-conducted studies, none reached a 5-year observational period, setting this as the major limitation of this article. Thus, additional articles with increased follow-up are necessary to compare the long-term results of these 2 treatment approaches and to validate the reliability of short implants as an alternative to more complex and operator-sensitive procedures.

Conclusion

The placement of short implants is a viable option in treating patients with maxillary atrophy. Up to a 3-year follow-up, short implants demonstrate comparable survival rates, as well as reduced biological complications, marginal bone loss, cost, and surgical time when compared with long implants. However, prosthetic complications are more frequent with extra-short implants. Study designs with longer observational periods are needed to study the long-term performance of these implants.

Disclosure

The authors claim to have no financial interest, either directly or indirectly, in the products or information listed in the article.

Roles/Contributions by Authors

A. Ravidà: manuscript preparation and data interpretation. I-C. Wang: statistic (meta-analysis) and manuscript preparation. G. Sammartino: article conception, study supervisor, and advisor. S. Barootchi: manuscript preparation. M. Tattan: manuscript preparation. G. Troiano: statistic (trial sequential analysis), study supervisor, and advisor. L. Laino: study supervisor and advisor. G. Marenzi: study supervisor and advisor. U. Covani: study supervisor and advisor. H.-L. Wang: study supervisor and advisor.