2.1

Study characteristics, participants and design

The database with clinical records of one dental office was used in the present evaluation. The survival of resin composite restorations in posterior teeth was determined retrospectively for up to 18 years, and the influence of several variables in the outcome was investigated. The absence or presence of an intermediate layer of glass-ionomer cement underneath composite restorations was the main factor under analysis, considering survival, annual failure rate and types of failure as outcomes. The other evaluated factors were age and gender of participants, jaw, restored tooth, number of restored surfaces and type of composite.

The study was approved by the local Ethics Committee (N. 139.840) and the patients have signed a written informed consent. Original data were obtained from a private dental office in Caxias, RS, Brazil, and a single operator (PARR) placed all restorations. The first dataset refers to restorations placed between 1986 and 1990, whereas the second dataset refers to restorations placed between 1994 and 2002. During the above mentioned periods, all new Class I and II direct restorations were searched, which could include from 1 up to 5 restored surfaces (information present in the patient files), with or without the involvement of cusps (not described in patient files). For inclusion, patients should present full dentition or the restoration should be in occlusion and with at least one adjacent tooth. Patients should have stayed in continuous clinical follow-up, with at least 1 annual recall. In total, 128 patients were selected through the inspection of clinical and radiographic records and invited to visit the dental office. The recruitment was performed by letters and phone calls, and 97 (76%) adult patients agreed to participate in the clinical evaluations. For the present study, the same patient could be part of both datasets.

2.2

Clinical procedures

The terminology may be somewhat confusing when addressing liners and bases . For practical reasons, the term base will be used to describe the placement of intermediate layers covering most of the dentin part of the cavity. Also for practical reasons, the earlier dataset will be referred as S1, and the later as S2. In S1, restorations were placed using two composite resins, a minifilled hybrid composite with inorganic filler loading of 77 vol.% (P-50 APC; 3M ESPE, St. Paul, MN, USA), and a midifilled hybrid composite with inorganic filler loading of 57 vol.% (Herculite XR, Kerr, Orange, CA, USA). Bonding systems used were Scotchbond 2 (3M ESPE) for P-50 APC and XR Prime/XR Bond (Kerr) for Herculite XR. Restorations in S2 were performed using universal microhybrid composites (Z100, 3M ESPE; Tetric Ceram, Ivoclar-Vivadent, Amherst, NY, USA; Charisma, Heraeus Kulzer South America Ltda., São Paulo, SP, BR; or by the combination of these) with no substantial differences regarding filler loading (59, 60 and 56 vol.%, respectively). Bonding systems used were Scotchbond Multi-Purpose or Single Bond (3M ESPE). All restorations were placed under rubber dam isolation. Cavities were prepared using diamond burs, and low-speed steel burs were used to remove carious tissue. No bevels were made, and preparations were restricted to the removal of carious tissue and/or failed restorations. In deep cavities, including both S1 and S2 datasets, a thin layer of calcium hydroxide (Dycal, Dentsply Indústria e Comércio Ltda, Petrópolis, RJ, BR) and conventional glass-ionomer cement (Ketac-Fil, 3M ESPE) were used to cover the deeper parts of the pulpal wall. In S1, the conventional glass-ionomer cement (GI; Ketac-Fil, 3M ESPE) was used as base in a closed sandwich technique, where the dentin was covered with GI, and the outline of the restoration was completely in composite resin. All other procedures were performed in the same way. The cavities were etched with 35% phosphoric acid, and bonding systems were applied according to the manufacturers’ instructions. The composites were placed incrementally, with layers of 2-mm under a horizontal technique, and cured for 40 s using a quartz–tungsten–halogen curing unit (Visilux, 3M ESPE). In class II cavities, contoured metal matrices and wooden wedges were used to restore the proximal boxes, each at a time in case both proximal surfaces were involved. Finishing procedures were performed with fine-grit diamonds and flexible rubber points/cups (FlexiPoints/FlexiCups, Cosmedent, Chicago, IL, USA) with aluminum oxide paste (Enamelize, Cosmedent). The proximal surfaces were finished by means of scalpel blades, abrasive finishing strips (3M ESPE), and finishing/polishing discs (Sof-Lex, 3M ESPE) for marginal ridges.

2.3

Evaluation of restorations

First, the history of the restorations was collected from the patient files. The date of placement and subsequent interventions were recorded. The restoration was considered as failed in case of replacement or repair, and the reason for failure was registered (as judged by the treating clinician). Also, a clinical evaluation was performed for all restorations still in function, by calibrated examiners according to FDI criteria . The patients were examined at the dental office, and photographs were taken for future reference and documentation. The surfaces were air-dried and inspected with dental mirror and explorer, and the examiners were blinded to the materials. In case of disagreement, the examiners evaluated the restorations jointly, until a consensus was reached. For the analysis, different levels of each criterion were simplified according to re-treatment need: no intervention needed (success) and requiring intervention (failed).

2.4

Statistical analysis

Statistical analysis was performed using the software package Stata 11.0 (StataCorp LP, College Station, TX, USA). Fisher’s exact test was used to analyze the differences between materials and failure distribution and between techniques and type of failure distribution ( α = 0.05). Kaplan–Meier method was used to generate survival curves up to 18 years, because follow-up periods were different in S1 and S2 datasets. A multivariate Cox regression analysis with shared frailty was used to investigate the influence of the variables of interest (technique, material, age, gender, jaw, tooth and number of restored surfaces) in restoration survival. The hazard ratios with respective 95% confidence intervals were determined. Only variables presenting p < 0.100 were selected for multivariate analysis.

2.1

Study characteristics, participants and design

The database with clinical records of one dental office was used in the present evaluation. The survival of resin composite restorations in posterior teeth was determined retrospectively for up to 18 years, and the influence of several variables in the outcome was investigated. The absence or presence of an intermediate layer of glass-ionomer cement underneath composite restorations was the main factor under analysis, considering survival, annual failure rate and types of failure as outcomes. The other evaluated factors were age and gender of participants, jaw, restored tooth, number of restored surfaces and type of composite.

The study was approved by the local Ethics Committee (N. 139.840) and the patients have signed a written informed consent. Original data were obtained from a private dental office in Caxias, RS, Brazil, and a single operator (PARR) placed all restorations. The first dataset refers to restorations placed between 1986 and 1990, whereas the second dataset refers to restorations placed between 1994 and 2002. During the above mentioned periods, all new Class I and II direct restorations were searched, which could include from 1 up to 5 restored surfaces (information present in the patient files), with or without the involvement of cusps (not described in patient files). For inclusion, patients should present full dentition or the restoration should be in occlusion and with at least one adjacent tooth. Patients should have stayed in continuous clinical follow-up, with at least 1 annual recall. In total, 128 patients were selected through the inspection of clinical and radiographic records and invited to visit the dental office. The recruitment was performed by letters and phone calls, and 97 (76%) adult patients agreed to participate in the clinical evaluations. For the present study, the same patient could be part of both datasets.

2.2

Clinical procedures

The terminology may be somewhat confusing when addressing liners and bases . For practical reasons, the term base will be used to describe the placement of intermediate layers covering most of the dentin part of the cavity. Also for practical reasons, the earlier dataset will be referred as S1, and the later as S2. In S1, restorations were placed using two composite resins, a minifilled hybrid composite with inorganic filler loading of 77 vol.% (P-50 APC; 3M ESPE, St. Paul, MN, USA), and a midifilled hybrid composite with inorganic filler loading of 57 vol.% (Herculite XR, Kerr, Orange, CA, USA). Bonding systems used were Scotchbond 2 (3M ESPE) for P-50 APC and XR Prime/XR Bond (Kerr) for Herculite XR. Restorations in S2 were performed using universal microhybrid composites (Z100, 3M ESPE; Tetric Ceram, Ivoclar-Vivadent, Amherst, NY, USA; Charisma, Heraeus Kulzer South America Ltda., São Paulo, SP, BR; or by the combination of these) with no substantial differences regarding filler loading (59, 60 and 56 vol.%, respectively). Bonding systems used were Scotchbond Multi-Purpose or Single Bond (3M ESPE). All restorations were placed under rubber dam isolation. Cavities were prepared using diamond burs, and low-speed steel burs were used to remove carious tissue. No bevels were made, and preparations were restricted to the removal of carious tissue and/or failed restorations. In deep cavities, including both S1 and S2 datasets, a thin layer of calcium hydroxide (Dycal, Dentsply Indústria e Comércio Ltda, Petrópolis, RJ, BR) and conventional glass-ionomer cement (Ketac-Fil, 3M ESPE) were used to cover the deeper parts of the pulpal wall. In S1, the conventional glass-ionomer cement (GI; Ketac-Fil, 3M ESPE) was used as base in a closed sandwich technique, where the dentin was covered with GI, and the outline of the restoration was completely in composite resin. All other procedures were performed in the same way. The cavities were etched with 35% phosphoric acid, and bonding systems were applied according to the manufacturers’ instructions. The composites were placed incrementally, with layers of 2-mm under a horizontal technique, and cured for 40 s using a quartz–tungsten–halogen curing unit (Visilux, 3M ESPE). In class II cavities, contoured metal matrices and wooden wedges were used to restore the proximal boxes, each at a time in case both proximal surfaces were involved. Finishing procedures were performed with fine-grit diamonds and flexible rubber points/cups (FlexiPoints/FlexiCups, Cosmedent, Chicago, IL, USA) with aluminum oxide paste (Enamelize, Cosmedent). The proximal surfaces were finished by means of scalpel blades, abrasive finishing strips (3M ESPE), and finishing/polishing discs (Sof-Lex, 3M ESPE) for marginal ridges.

2.3

Evaluation of restorations

First, the history of the restorations was collected from the patient files. The date of placement and subsequent interventions were recorded. The restoration was considered as failed in case of replacement or repair, and the reason for failure was registered (as judged by the treating clinician). Also, a clinical evaluation was performed for all restorations still in function, by calibrated examiners according to FDI criteria . The patients were examined at the dental office, and photographs were taken for future reference and documentation. The surfaces were air-dried and inspected with dental mirror and explorer, and the examiners were blinded to the materials. In case of disagreement, the examiners evaluated the restorations jointly, until a consensus was reached. For the analysis, different levels of each criterion were simplified according to re-treatment need: no intervention needed (success) and requiring intervention (failed).

2.4

Statistical analysis

Statistical analysis was performed using the software package Stata 11.0 (StataCorp LP, College Station, TX, USA). Fisher’s exact test was used to analyze the differences between materials and failure distribution and between techniques and type of failure distribution ( α = 0.05). Kaplan–Meier method was used to generate survival curves up to 18 years, because follow-up periods were different in S1 and S2 datasets. A multivariate Cox regression analysis with shared frailty was used to investigate the influence of the variables of interest (technique, material, age, gender, jaw, tooth and number of restored surfaces) in restoration survival. The hazard ratios with respective 95% confidence intervals were determined. Only variables presenting p < 0.100 were selected for multivariate analysis.