The direct digital workflow in fixed implant prosthodontics: a narrative review

An electronic search of publications from January 1st 2015 to March 31st 2020 was conducted. The cut-off point (2015) was selected because the rate of advancement in scanner hardware and software [20] and dental CAD/CAM material science [19] has accelerated in the past 5 years. The search strategy used a combination of free-text words. A MEDLINE (PubMed) search was performed and the search terms together with the number of records returned are shown in Table 1.

This review included randomized control clinical trials (RCTs), prospective and retrospective clinical trials, case series and in vitro studies focusing on intraoral digital implant impression accuracy. In vitro and in vivo studies comparing different IOS devices in terms of scanning efficiency were included. Studies comparing intraoral digitization to conventional implant impressions in terms of accuracy were also included. Reports on the accuracy of fit as well as on the success and/or survival of monolithic zirconia and lithium disilicate restorations on implants, produced through an IOS impression procedure were also identified and included in this review. In vitro and in vivo studies looking into the fit accuracy of restorations produced through an indirect digital workflow (laboratory scanning) were excluded. Case reports were also excluded. The search included only English-language articles. To further identify any missed articles, the reference lists of the included papers were screened.

The following questions were formulated and addressed in this review:

Several in vitro studies were identified examining the scanning accuracy of IOS in single-implant edentulous sites (Table 2). IOS scan accuracy has been studied in conjunction with the digital manufacturing of the master model through rapid prototyping techniques. This approach necessitates the milling or 3D-printing of the master model from the IOS scan in order for the restoration to be completed in a semi-digital approach usually employing a porcelain layering step. Alternatively, a complete digital workflow utilizing a monolithic restoration and without necessitating the fabrication of a physical model can be used. Evidence suggests, however, that neither of these approaches is without discrepancies and that the final implant position in the virtual or the physical master model is statistically significantly different compared to the analogue position in the cast model, produced from a conventional impression [21‐23]. Mangano et al. [23] in a comparative study reported discrepancies in the virtual position of a single implant ranging from 15 ± 0.8 to 43 ± 11 μm depending on the scanner tested. Deviations of 7–37 μm in the final vertical position of the single virtual implant were also reported in another in vitro study by Chew et al. [24] and correlated to the implant platform placement depth and the scanner used. In another in vitro study by Chia et al. [25] a 15 N/cm torque, applied during tightening of the polyetheretherketone (PEEK) implant scanbody, was shown to alter the position of the implant as much as 11 (± 4.9) μm in an apical direction due to compression of the scanbody material. The surface matching discrepancies between the scanbody and the implant platform, have also been shown to amount to 9–11 μm [26, 27], further contributing to digital impression total inaccuracies.

Additional implant positional discrepancy can be expected when a physical master model is digitally produced. Revilla-Leon et al. [28] in an in vitro study reported that the design of scanbodies significantly affected the positional accuracy of the implant analogues inside the 3D-printed cast. Mühlemann et al. [22] in an in vivo study reported that the conventional impression and gypsum production procedure exhibited higher positional accuracy (32 ± 11 μm) of the implant analogue compared to IOS and digital model production (57 ± 32–176 ± 120 μm), regardless of the scanner and the rapid prototyping process used for fabrication of the plastic model. The fact that this was an in vivo study and that measurements were made on printed/milled models may have contributed to the un-favorable IOS accuracy results along with operator calibration. Furthermore, Lee et al. [29] attributed the implant positional discrepancy to the inaccuracy of friction-fit placement of the digital implant analogue inside the plastic model. All of the aforementioned factors can lead to the final implant crown being over- or infra-occluded therefore requiring major chairside adjustments [30] or even additional laboratory procedures.

Regarding implant-rehabilitated short-span edentulous sites, several studies have compared the IOS accuracy of various scanner devices to the conventional impression accuracy (Table 3). Digital implant impression for short span prostheses, supported by up to 2–4 implants located within the same quadrant, has been mainly compared for in vitro accuracy to an elastomeric impression technique, utilizing either addition-cure silicone materials in a single or dual mix technique, or a polyether mono-phase technique using custom trays [21, 24, 25, 31]. Implant impression posts, in the conventional impression approach, were usually not splinted. Statistical superiority of the conventional method was reported in the majority of studies but the accuracy deviation of the IOS devices ranged from 27 to 66 μm depending on the scanner, whereas for the conventional method the deviation ranged from 26 to 49 μm [24, 25, 31]. To what extend this statistical significance translates into clinical significance is not known. In the study by Basaki et al. [21], the IOS deviation was reported to be 116 (± 94) μm as compared to 56 (± 29) μm for the conventional impression procedure but the calculation was performed on the polyurethane milled casts that were produced from the digital impressions. Therefore, additional deviations in the milling process may have aggravated this discrepancy. In a recent in vivo study by Alsharbaty et al. [32] the authors reported statistically significant differences in accuracy between the conventional and the digital impression of partially edentulous sites with 2 adjacent implants, although clinical significance could not be concluded according to the authors.

Complete-arch IOS accuracy of multiple implant impressions has been studied intensively in the past 5 years. The conventional method, utilizing elastomeric impression materials and multiple implant impression post splinting, has been the gold standard against which the accuracy of various scanners was tested (Table 4). Current evidence on the superiority of one technique over the other is inconclusive. There is available research postulating that IOS of complete edentulous arches with 5 or 6 implants is either equally or statistically significantly more accurate than conventional elastomeric impressions taken using impression post splinting and an open custom tray approach [33‐35].This finding is independent of the scanner used, as different IOS technologies such as confocal microscopy and active triangulation have been tested in the aforementioned studies. Impression material type is also non-contributory, as both polyether (PE) and polyvinylsiloxane (PVS) high accuracy elastomeric materials were used. In contrast, there is also available evidence supporting the significant statistical superior accuracy of the splinted, open-tray, conventional elastomeric impression technique over the IOS impression for complete-arch implant rehabilitation [36‐38]. Again, this finding was irrespective of IOS device and impression material used. This lack of consensus can be attributed to factors such as the study design, the different IOS device software and hardware used or the statistical analysis employed but it is unclear whether statistical significance translates into clinical significance.

There appears to be scientific evidence in the available literature, however, regarding the superiority of the digital intraoral scanning method in relation to the conventional, non-splinted elastomeric impression technique using either an open or a closed custom tray for complete-arch impressions [39, 40]. In a study by Rech-Ortega et al. [40], the authors stated that despite the higher accuracy of the digital scanning method, both techniques exhibited a deterioration when more than 4 implants were involved in the scanning scope. Alikhasi et al. [39] reported that the digital impression technique was statistically more accurate than both the direct (open tray) and the indirect (closed tray) conventional elastomeric impression method.

Besides the conventional and the complete digital workflow for edentulous arches with multiple implants, there is always the option of producing a 3D-printed or milled cast from the IOS impression and using this as the master model. Research on the accuracy of such models produced using rapid prototyping techniques is scarce (Table 5). Papaspyridakos et al. [41] in an in vitro study on the accuracy of 4 implant analogue positions in SLA (Stereolithography) casts produced through IOS scans, concluded that the mean deviation of the printed casts was 59 (± 16) μm. The implant analogue 3d deviations were statistically significantly different from the master model, but still within a clinically acceptable range according to the authors [41]. In another in vitro study, Revilla-Leon et al. [42] tested several 3D-printing technologies for the production of a completely edentulous maxillary cast with 7 implants. The authors reported that not all production methods led to results comparable to the conventional gypsum master model in terms of accuracy. Digital light processing (DLP) and Polyjet 3D printing technologies with specific 3D printers showed comparable accuracy to the stone model. Implant analogue deviations ranged from 21 (± 16) μm (Polyjet) to 27 (± 20) μm (DLP).

Several in vitro studies have been conducted comparing different scanner devices regarding both partial [43‐47] and complete-arch [23, 44, 48‐54] accuracy. Digital scans from the various IOS devices were compared for trueness and precision against the scans from a highly accurate reference laboratory scanner. Evidence suggests that scanner type and generation can influence scanning accuracy as some scanner devices exhibited higher precision (low standard deviation) and higher complete-arch scanning accuracy compared to others. Nevertheless, the majority of newer generation scanners produce complete-arch accuracy values less than the maximum 150 μm threshold, currently accepted in clinical practice [45, 55‐58].

Several clinical factors contributing to the global deviations in complete-arch intraoral scanning have been identified and studied in the literature. Operator experience is one clinical parameter that has been reported to influence scanning accuracy in a study utilizing an Active Wavefront Sampling (AWS) technology scanner (Lava COS) [59] but more recent studies with newer generation scanners using both AWS and Confocal Microscopy technology failed to verify this finding, or identify the clinically relevant level of operator experience [48, 60, 61].

Implant angulation is another clinical factor that has been extensively studied for its effect on both partial and complete-arch implant digital impression accuracy. In the vast majority of studies, where single-part all-PEEK scanbodies were used as scanning posts for both partial [21, 24, 25] and complete-arch [33, 39, 48, 59, 62, 63] digital impressions, scanbody angulation did not affect scan accuracy. In the contrary, in a study by Arcuri et al. [61], the authors reported that complete-arch scan accuracy was, indeed, influenced by scanbody angulation. This finding may be attributed to the material of the scanbodies themselves. The PEEK-titanium scanbodies that were used in the study, presented the worst overall accuracy results compared to the all-PEEK and the titanium scanbodies they were compared against, possibly due to the interlocking between the two parts. In another in vitro study by Lin et al. [64], accuracy of the 2 implant analogues position in partial, milled polyurethane casts fabricated digitally following IOS, was found to be influenced by minor implant angulation (0–15 degrees) but not by major implant angulation (30–45 degrees). The authors reported that it is unclear whether the design of the two-piece scanbodies used in the study attributed to this result.

Further-on, regarding the design characteristics of scanbodies, their influence in scan accuracy has been tested in both partial and complete-arch digital implant impressions. The refractory and reflective indexes of all-PEEK scanbodies have been reported to be beneficial for complete-arch scan accuracy [44, 61]. Additionally, implant placement depth has not been reported to play a detrimental role in complete-arch IOS accuracy [48, 59] assuming the visible part of the scanbody can provide adequate reference points for IOS registration [62]. Therefore, using scanbodies of adequate length for optimum scan accuracy is indicated [57]. Cylindrical scanbodies with smoother surfaces have also been reported to facilitate IOS digitization by producing less noise as opposed to scanbodies with irregular shape [65‐67]. Recent research has also highlighted the importance of certain features related to scanbody manufacturing tolerances and their effect on the accuracy of the digital IOS impression. Schmidt et al. [68] have reported significant differences in design characteristics such as length and diameter between implant scanbodies of the same manufacturer. These tolerances may affect the accurate transfer of implant position and therefore contribute to the final prosthetic misfit. In addition, Mangano et al. [69] have reported on the congruence between the IOS mesh file and the CAD library file of scanbodies when scanned with different IOS devices. Certain scanners seem to digitize the shape of the scanbody more closely to the actual CAD library file compared to others. Finally, implant scanbody reusability is another important parameter that needs to be consider. Limited evidence regarding all-PEEK scanbodies suggest that using them up to ten consecutive times does not impact on transfer accuracy [70].

Lighting conditions during a scanning session have recently been reported to influence global scanning accuracy. Research has shown that each IOS device scans more accurately in specific lighting conditions [71, 72] that correlate to its inherent image acquisition technology. Regardless of this technology however, precise superimposition or stitching of successive images is imperative for accurate scan results. This process is known to produce dimensional discrepancies that are directly related to both the scanning scope and the interimplant distance. Its effect is multiplied in complete-arch edentulous jaws with limited reference points and landmarks among multiple implants as well as in the mandibular posterior area where scanner tip access is compromised due to tongue movement and limited space [73, 74]. Studies on partial-arch digital scan accuracy have shown that when the range of scan and interimplant distance increased, the scanning accuracy decreased [43, 45, 46, 54]. Moreover, increasing scanning range and interimplant distance have also been reported to influence complete-arch scan trueness and precision [35, 40, 50, 60, 62, 75], although the minimum number of installed implants for an accurate digital impression has not yet been investigated. The main issue with multiple implant scanning in fully edentulous arches remains the difficulty in predictable scanning of the soft tissue between the fixtures themselves. Mizumoto et al. [75] in a recent in vitro study have reported that in the completely edentulous maxilla with 4 installed implants, including scanning data from the palate did not result in statistically significant higher accuracy. In an effort to minimize discrepancies when scanning edentulous sites among multiple implants, Iturrate et al. [76, 77] have investigated the in vitro effectiveness of using an auxiliary geometric device (AGD) firmly attached onto the implant scanbodies. The authors reported statistically significantly higher accuracy when the AGD was used regardless of the IOS scanner tested. Huang et al. [38] in an in vitro study have also reported improved complete-arch accuracy when modified, interconnected scanbodies on 4 implants were used. Motel et al. [65] in a recent in vitro study reported that scanning for a partial edentulous site with 3 adjacent implants in a single step (implant position scan only) led to more accurate results compared to scanning in two steps (emergence profile scan and implant position scan). The authors attributed this to the superimposition discrepancies that occur when the two scans are aligned in the scanner software. Finally, Alikhasi et al. [39] reported that the type of implant connection (internal or external) did not influence complete-arch scan accuracy using a confocal microscopy scanner in a maxillary edentulous jaw with 4 implants.

Clinical studies regarding implementation of the complete digital workflow have been reported in the literature (Table 6) with the focus being mainly on the rehabilitation of single posterior implants following a digital intraoral impression procedure with or without the fabrication of a 3D-printed or milled master cast. Monolithic zirconia crowns and monolithic lithium disilicate crowns have been studied individually or compared to each other or to a metal-ceramic counterpart for success and survival.

In the available literature on monolithic zirconia crowns, success ranged from 92 to 100% and survival ranged from 97.5 to 100% for a follow-up of 1–3 years [78‐81]. Technical complications such as ill-fitting crown on a prefabricated abutment, fracture of a cusp, infra-occlusion, inferior aesthetics and crown de-cementation were reported. Biological complications were minimal. A recent in vivo study by Lerner et al. [82] also reported very promising results for monolithic zirconia crowns fabricated on hybrid zirconia abutments following an IOS procedure. Success and survival rates after a mean of 3 years of follow-up were 91.3% and 99% respectively with a 1.9% biologic and 5.7% prosthetic incidence rate.

Available literature on monolithic lithium disilicate CAD-CAM crowns also shows encouraging results. Short-term in vivo studies exhibited a success rate of 89–100% and a survival rate of 100% for a follow-up of 2–3 years [5, 81]. Technical complications such as minor chippings were observed. Biological complications were again minimal.

Joda and coworkers reported on a cohort of patients rehabilitated with single-implant, porcelain-layered zirconia crowns following an IOS impression and a digital model fabrication process [83, 84]. At 3 years follow-up, both the success and survival rates were 100% and the patients reported high levels of satisfaction [84]. After 5 years of function however, one implant was lost leading to a success/survival rate of 95%. Mean bone loss around the implants increased significantly by 0.23 mm mesially and 0.17 mm distally compared to baseline [83].

A recent in vivo study reported on the fit of 2,3 and 4-units zirconia fixed partial dentures on 2 implants following a complete digital workflow [85] (Table 7.) The authors claimed that interimplant angulation exceeding 10 degrees could negatively influence the passive fit of the restorations as opposed to their counterparts, fabricated through a conventional workflow. Interimplant distance, on the other hand, exhibited no significant effect on passive fit of either group.

With regard to time efficiency of the digital workflow, several studies have reported statistically significant shorter clinical and laboratory working times for the complete digital, compared to a semi-digital or conventional workflow for single-implant rehabilitation [79, 86‐88]. Regarding patient satisfaction, evidence also supported the significantly superior acceptance of the digital workflow in terms of comfort and ease of the IOS impression procedure compared to the conventional elastomeric impression [88, 89], although the final aesthetic outcome of monolithic zirconia restorations has been reported to be inferior to their porcelain-layered zirconia counterparts [78].

Within the scope of this review, no studies on the complete direct digital workflow for rehabilitation of multiple implants in edentulous arches were identified. Systematic reviews on the success and survival of implant-supported, zirconia complete fixed dentures fabricated through a conventional impression workflow, suggest that the use of monolithic or minimally veneered zirconia frameworks may help eliminate frequent complications encountered with veneering porcelain chipping [90, 91]. Minimal buccal veneering can also aid in solving the aesthetic problem often encountered with such designs but this restorative option has yet to be tested within the concept of the direct digital workflow.