Retrospective clinical evaluation of 2- to 6-unit implant-supported fixed partial dentures: Mean follow-up of 9 years.

BACKGROUND
Implant-supported fixed partial dentures (ISFPDs) are one of the most common options to rehabilitate partially edentulous patients.


PURPOSE
To assess the clinical outcomes of ISFPDs.


METHODS
This retrospective study included all patients treated with ISFPDs with 2 to 6 prosthetic units at one specialist clinic. Implant/prosthesis failure and technical complications were the outcomes analyzed.


RESULTS
Six hundred and forty-two patients with 876 ISFPDs (2241 implants) were included, followed up for 108.0 ± 76.2 months. Eighty-eight prostheses and 112 implants (26 before, 86 after prosthesis installation) failed. The estimated CSR of ISFPDs at 30 years was 72.7%. Smokers presented lower implant survival than nonsmokers. Two hundred and ninety-nine ISFPDs (33.2%) presented technical complications. Bruxism was a factor to exert a higher risk of screw and implant fracture, and ceramic chipping. ISFPDs with cantilever presented higher risk of failure, and screw loosening/fracture. Prostheses supported by implants with internal abutment connection or with two pontics had higher risk of presenting ceramic chipping. Extension of the prosthesis did not seem to exert influence on prosthesis failure/complications.


CONCLUSIONS
ISFPDs presented good long-term prognosis. Implant failure was the main reason for ISFPD failure. The results suggest that bruxism and the presence of cantilever may contribute to the increased rate of mechanical complications and prosthesis failure.

more predictable. 1 A review 2 on of the survival and complication rates of implant-supported fixed dental prostheses concluded that these kinds of prostheses are a safe and predictable treatment method with high survival rates, but also observed that technical complications were frequent.
Traditionally, these prostheses were manufactured with a castgold alloy framework covered by acrylic resin teeth. Later, milled titanium frameworks were suggested as an alternative to conventional castings in implant dentistry, as these prostheses could provide a better fit to the implants than conventional castings. 3 When it comes to the acrylic occlusal surfaces, these show complications such as wear and tooth fracture clinically. Porcelain was then proposed as an alternative material for artificial teeth, usually presenting greater wear resistance and more favorable esthetic results than acrylic resin, although presenting chipping as a common clinical complication. 4 Ceramic-veneered zirconia frameworks have been introduced more recently, showing high biocompatibility, low bacterial surface adhesion, high flexural strength, toughness, and esthetic properties. However, as with porcelain, chipping of the veneering material was the most frequently reported complication. 5 In order to minimize these complications, monolithic framework materials were developed and introduced, 6 but the long-term outcome of these restorations remains unknown, due to the lack of sufficient clinical studies. 4 When it comes to dental implants, these were in the early years mainly manufactured with a turned (machined) surface or with an extremely highly roughened surface. The adoption of a moderately rough surface improved the survival rates. 7 The implant abutment connection was another part of the implants that had modifications over time. The external hexagon abutment connection was originally used on the Brånemark implant, but presented some drawbacks. It has been suggested that the connection might allow micromovements under high occlusal loads, resulting in abutment screw loosening or even fatigue fracture. 8 Implants with internal connection were then developed, with a more stable connection permitting a more even stress distribution throughout the body of the implant. 9 Despite the high survival rate in many studies, implant-supported prostheses are not without problems, and their longevity is limited not only by biologic complications but also by prosthetic maintenance requirements, the restoration material, and the implants used. 10,11 Therefore, it should be of interest to investigate the outcome of this prosthetic option in routine practice. In this context, the present retrospective study aimed to assess the clinical outcomes of fixed partial dentures supported by dental implants.

| Materials
This retrospective study included patients treated with dental implants during the period 1980-2018 at one specialist clinic (Clinic for Prosthodontics, Centre of Dental Specialist Care, Malmö, Sweden).
This study was based on data collection from patients' dental records, and the patients were not recalled for further examination. The implants were placed by specialist dentists in oral surgery, and dentists performing the prosthetic treatment were specialists in prosthodontics. The study was approved by the regional Ethical Committee, Lund, Sweden (Dnr 2014/598; Dnr 2015/72).

| Definitions
An implant was considered a failure if presenting signs and symptoms that led to implant removal, that is, a lost implant (due to initial failure to osseointegrate, late loss of osseointegration or fracture of the implant). A prosthetic unit was defined as the replacement of one prosthetic tooth in the prosthesis. For example, an ISFPD replacing teeth in the positions 35-36-37 has three prosthetic units.
For this study, the authors followed the definition of bruxism proposed by Lobbezoo et al. 12 : "Bruxism is a repetitive jaw-muscle activity characterized by clenching or grinding of the teeth and/or by bracing or thrusting of the mandible. Bruxism has two distinct circadian manifestations: it can occur during sleep (indicated as sleep bruxism) or during wakefulness (indicated as awake bruxism)." The sign and symptoms of bruxism were listed according to the International Classification of Sleep Disorders, 13 following the same guidelines used in a recent study 14 (the patients suspected to be bruxers, as diagnosed in the records, were recalled in this previous study in order to be clinically reassessed. Part of these patients-the ones with the type of prosthesis being evaluated here-is included in the present observational study). A diagnostic grading system of "possible," "probable," and "definite" sleep or awake bruxism was used, as suggested for clinical and research purposes. 15 According to an international consensus, 12 "possible" sleep or awake bruxism should be based on self-report, by means of questionnaires and/or the anamnestic part of a clinical examination. "Probable" sleep or awake bruxism should be based on self-report plus the inspection part of a clinical examination. "Definite" sleep bruxism should be based on self-report, a clinical examination, and a polysomnographic recording, preferably along with audio/video recordings. As electromyography and/or polysomnography were not used, the patients of the present study would only fall into the category "probable." Thus, a patient was considered as presenting bruxism based on self-report of clenching/grinding during sleep or during wakefulness, plus the inspection part of a clinical examination. Patients smoking a minimum of one cigarette per day were classified as smokers. Patients who stopped smoking before implant placement, and remained nonsmoking until the patient's last follow-up were classified as former smokers. Any condition or situation that led to the removal and/or replacement of the fixed partial denture was considered as a prosthesis failure.
Technical complications could have happened to the prosthesis or to the implant were the following: • Complications regarding the prosthesis: loss/fracture of suprastructure acrylic teeth or ceramic veneer; fracture of the prosthesis framework (ie, prosthesis with complete transversal buccal-lingual fracture); Implant access hole sealing lost/fractured; mobility of the prosthesis, prosthesis completely loose due to complete unscrew of all prosthetic screws (prosthesis uncoupled); • Complications regarding the implant: failure; fracture of the implant itself; loosening, loss or fracture of connecting/abutment screws; loosening, excessive wear, deformation or fracture of prosthetic abutment.
After completion of the prosthetic treatment, the patients were periodically followed up by a dental hygienist at the clinic for a professional dental hygiene program, with attendance frequency to the professional hygiene recall program based on individual needs.

| Inclusion and exclusion criteria
Only ISFPDs with 2 to 6 prosthetic units and supported by modern threaded cylindrical-or conical-design implants were included. The patients needed to have at least 6 months of follow-up after installation of the prosthesis.

| Data collection
The data were directly entered into a SPSS file (SPSS software, version 25, SPSS Inc., Chicago, Illinois) as the dental records of the patients were being read, and it consisted of several implant-, site-, and patient-related factors.

| Statistical analyses
The mean, SD, and percentage were calculated for several variables.
The test performed were the following: Kolmogorov-Smirnov  loosening, screw fracture, and ceramic chipping/fracture. Hazard ratio (HR) and 95% confidence intervals (95% CI) were estimated from Cox proportional hazard models. In order to verify multicollinearity, a correlation matrix of all of the predictor variables with a significant odds ratio (P value cut-off point of 0.1) identified in the univariate models was scanned, to see whether there were some high correlations among the predictors. Collinearity statistics obtaining variance inflation factor (VIF) and tolerance statistic were also performed to detect more subtle forms of multicollinearity.
For final multivariable Cox regression models, all variables that were moderately associated (P < .10) with prosthesis failure were included. For this prosthesis-level analysis, clustering of multiple prostheses within each patient was accounted for in the Cox models using the methods outlined by Lee et al. 16

| RESULTS
A number of 738 patients were initially planned to receive 1035 ISFPDs in the clinic. A total of 159 prostheses installed in 127 patients were excluded from the study. Thirty-one of these patients had other prostheses that were included in the study. Thus, 96 out of these 127 patients were excluded. These were the reasons for exclusion: • 72 ISFPDs were connected to at least one tooth; • 48 prostheses were placed in patients with a very short follow-up (<6 months); • 10 patients received implants and were planned to receive 12 ISFPDs prostheses, but received the restorations in other clinics; • 11 patients were planned to receive 13 ISFPDs prostheses, but received other type of prosthetic restoration instead (of which for 10 cases the choice was for single-crowns); • 5 patients that received 7 prostheses had one or more implants lost within 6 months after prosthesis installation; • 5 patients planned to receive 6 prostheses, but had one or more implants lost before prosthesis could have been installed; • 1 patient died a few months after the installation of the prosthesis. There was available information about the opposing jaw status for 89.8% of the cases (n = 787). Most of opposing jaws were consisted of natural dentition with/without fixed partial prosthesis or of implant-supported full-arch prosthesis (92.8%; n = 730), followed by partially dentate arches with/without removable partial denture (5.3%; n = 42), and by rehabilitation by removable complete denture or overdenture (1.9%; n = 145).
A number of 39 implants were lost before the installation of the prosthesis and were not replaced. The cases comprising these 39 nonreplaced lost implants were excluded from the study for two reasons: either the patients lost all implants planned to support the ISFPD or only one implant remained, eventually supporting a single crown.
From the total of 2241 implants installed (already excluding the cases of these 39 implants mentioned above), 26 were lost before the installation of the prosthesis, a mean ± SD of 5.8 ± 2.3 months (minmax, 1.7-11.6) after implant installation, of which 20 implants were Eighty-eight prostheses failed (10.0%), with no significant differences in failure rate between those installed in maxillae and in mandibles (P = .147, log-rank test), but with statistically significant higher failure rate in bruxers in comparison to nonbruxers (P = .004, logrank test) (Table 1) (Table 6).

| DISCUSSION
The use of implant therapy in special populations requires consideration of potential benefits to be gained from the therapy. To better appreciate this potential, the present study aimed to assess the clinical outcome of ISFPDs. We present an analysis of 876 ISFPDs, of which 88 failed, with a mean follow-up of 9 years. Some clinicians might be tempted to calculate the general "failure rate" of the study. However, this would not be appropriate, as this outcome was observed over time and not all participants were observed for the same time therefore censoring has occurred. Therefore, all statistics should include time to event methods, 18 22,25 Other studies suggested that bruxism may be a risk factor for fractures of ceramics 26 and, in general, for the need for technical interventions on implant-supported restorations. [27][28][29] Loss of implants was also observed to have a higher prevalence in smokers/former smokers than in nonsmokers. Smoking is a factor that has the potential to negatively affect healing and the outcome of implant treatment. 30 Smoking was, however, not a significant factor to directly affect the failure of prostheses, according to the Cox regression model.
Prostheses opposed to either natural dentition or fixed prosthesis presented a higher failure rate than prostheses opposed to removable complete dentures or overdentures, or to partially dentate jaws with or without removable partial denture. It is believed that intraoral force T A B L E 4 Prevalence of complications, according to prosthesis material, jaw, sex, and bruxism. The statistical unit is the implant, not the patient neither the implant-supported fixed partial denture magnitudes transmitted to prostheses may be higher when opposed to fixed dentition/prosthesis when compared to removable dentures.
ISFPDs with a cantilever prosthetic arm presented a higher risk of prosthesis failure, as well as a higher risk for screw loosening or screw fracture. This is in agreement with some previous studies 31,32 that showed a relatively high number of technical or mechanical complications in ISFPDs with cantilever. The incidence of abutment screw problems may be largely attributed to the increased bending moments and rotational forces experienced with the cantilever design. 33 Moreover, ISFPDs with two pontics had a higher risk of chipping of the ceramic veneer than prostheses with one or no pontic. Despite the fact that the pontics were not always adjacent to one another and that the dimensions of the pontics were not measured by the authors of the present study, this may have be related to the dimensions of the pontics. The occluso-gingival thickness of the pontic has an effect on deflection of framework. Moreover, long anterio-posterior metal substructures flex under heavy or complex loads leading to porcelain fracture. A fixed partial denture with two-tooth pontic span will bend eight times as much as a single-tooth pontic fixed partial denture will, if everything else remains unchanged. 34 Ceramic veneer chipping was also more prevalent in prostheses supported by implants with internal abutment connection in comparison to prostheses supported by implants with external abutment connection. Unfortunately, no reasonable explanation for this result could be found.
Another finding from the regression models was that ISFPDs supported by turned implants presented higher risk of screw loosening in comparison to prostheses supported by moderately rough implants. This could be related to the fact that the older turned/ machined implants had external abutment connections and the more recent moderately rough implants had internal abutment connections, 35 but this was not confirmed by the statistical model for the present material.
The results of the present study suggest that prosthetic material seems to have no significant impact on survival rates of ISFPDs, in agreement with some reviews on the subject. 4,10,36 However, one of the reviews highlighted the considerable heterogeneity of studies with large variation in number of restorations per material group available for analysis, making comparisons difficult. 10 The limitations of the present study include the fact that this is a retrospective study, which inherently results in flaws, manifested by gaps in information and incomplete records. The lack of information on several factors, including biological factors such as oral hygiene status, bleeding on probing, probing pocket depth, and the periodontal and peri-implant status, which may have some influence on the results, 37,38 is also related to the retrospective nature of the present study. Moreover, since this was not a prospective study, treatment was not standardized. As a result, the distribution of different materials for the manufacture of the prostheses was not balanced, as it followed the trend of each time period. It can also be observed that several professionals were involved in the treatment of these patients for the long time of observation of the study, and there must have been some variability of surgical and prosthetic approaches and dental laboratory techniques applied by these different professionals. An example would be the dimensioning of the framework, which was not standardized. Another example would be the influence of different surgeons on the implant failure rate. 39,40 Furthermore, the diagnosis of bruxism was based on self-report of clenching/grinding during sleep or during wakefulness, plus the inspection part of several clinical examinations during the treatment. The patients were not examined by electromyography and/or polysomnography.

| CONCLUSIONS
ISFPDs presented good long-term prognosis. Failure of one or more supporting implants was the main reason for the failure of ISFPDs.
The results of the present study add evidence to suggest that bruxism and the presence of prosthetic cantilever arms may contribute to the increased rate of mechanical complications in ISFPDs, as well as prosthesis failure. Note: Only the factors that were considered statistically significant (P < .1) in the univariate model and did not present multicollinearity were included in the multivariate model. Abbreviations: 95% CI, 95% confidence interval; FPD, fixed partial denture; RPD, removable partial denture; w, with; wt, without. a Prostheses with three pontics were considered to have not so many cases to be included here (failures/total number of cases, % of failure): 0/3, 0%.