Following the Cochrane methodology, we searched MED-LINE and Embase for reports of prospective randomized controlled studies comparing at least one composite and one ceramic material for inlay or onlay manufacturing, with a minimum follow-up of 6 months, and without any date or language restriction up to 11 October 2012.

In vitro: 91 studies

Only one ceramic or one composite (no control or luting agent/base material randomized): 20 studies (27 reports)

Ceramic versus ceramic: three studies

Composite versus composite: two studies

Ceramic versus composite (non-randomized or retrospective study): five studies (eight reports)

Ceramic versus composite (prospective randomized study): two studies (four reports)

Only two randomized studies were identified, which compared ceramic and composite materials for inlay or onlay manufacturing. In 2005, a study compared 80 VITA Mark II (ceramic; Vita Zahnfabrik, Bad Säckingen, Germany) and Paradigm (composite; 3M Espe, St Paul, MN, USA) CAD/CAM inlays in 43 adults after 3 years with use of the US Public Health Service (USPHS) modified criteria 29 . The composite inlays performed better for only two items: color match and restoration fracture 25 . In 2006, a study compared 58 CEREC, Vita Dur N (two ceramics; Vita Zahnfabrik), Brilliant DI (Coltene/Whaledent AG, Altstätten, Switzerland) and Estilux (two composites; Heraeus Kulzer GmbH, Hanau, Germany) inlays in 37 patients after 10 years with use of the California Dental Association criteria 30 : survival was similar for all inlays when repairs were not considered failures (75 to 80%) and was better for CEREC ceramic inlays than other inlays when repairs were considered failures (80% versus 51 to 67%) 26 . Data on the material to use for manufacturing inlays or onlays are thus controversial and a RCT is needed.

The main objective of the CEramic and COmposite Inlays Assessment (CECOIA) randomized controlled trial (RCT) is to compare the clinical efficacy of composite and ceramic inlays or onlays for treating moderate substance loss of posterior teeth in adults according to recent consensus outcomes. Secondary objectives include the overall quality, wear and survival of inlays and onlays made of both materials, and prognostic factors of restoration failure, including patient-related items.

Patients are eligible to participate in the trial if they are adults aged 18 to 70 years, can tolerate restorative procedures and have a posterior moderate-sized dental caries or aged restoration necessitating an inlay or onlay. Exclusion criteria are allergy to one of the materials used, bruxism, severe or acute periodontal or carious disease (greater than or equal to four primary or secondary restorations due to caries in the preceding year) and poor oral hygiene; the tooth to be treated should not need endodontic treatment or retreatment, show mobility >1 mm or a periodontal socket >3 mm or support a removable partial denture.

Patients with a tooth showing a subgingival margin after cavity preparation, that cannot be isolated with use of a rubber dam, or that has cusps that all need to be covered by the restoration are not eligible.

Only one tooth per patient is eligible. If a patient needs more than one inlay/onlay restoration, the tooth with expected cervical limits that are the most coronal, is the eligible tooth. Other required inlays or onlays will be manufactured by the dentist with the usual material (leucite-reinforced glass-ceramic). In case of pulpal exposure, the operator will decide whether a direct pulp capping (with calcium hydroxide) or an endodontic treatment is necessary, randomize the patient after this treatment has been conducted and fill the corresponding fields in the adverse events section of the case report form (CRF).

Operators will be eligible for inclusion if they have at least 3 years of clinical experience and at least 1 year of experience with chairside CAD/CAM, agree with the intervention protocol, and have no preference for either composite or ceramic to manufacture inlays and onlays.

Evaluators of restorations during follow-up will be two dentists different from the operators.

Patients will be included and treated in seven centers in France: the dental care departments of two hospitals (Hôpital Charles Foix, Ivry-sur-Seine and Hotel-Dieu Saint-Jacques, Toulouse) and five private practices (four in Paris and one in Lyon). Follow-up data will be collected in these seven centers. Any patient with the eligible criteria visiting one of the included centers will be asked to participate in the study. The consent form can be consulted at http://cecoia.fr: extra section.

After powder spraying (CEREC Optispray, Sirona Dental Systems), the operator will scan the preparation with use of a digital camera and design the restoration by use of CEREC software (Sirona Dental Systems). If eligibility criteria are still satisfied, the operator will then randomize the tooth to a treatment (randomization procedure described below), insert the corresponding block inside the milling machine and press the button for the restoration to be milled. The operator will then check the approximal contacts of the resulting restoration, correct them if need be, remove the machining lug and weigh the restoration.

The operator can glaze (IPS Object Fix Putty, glazing paste and stains, Ivoclar Vivadent) or polish the inlay or onlay using the polishers provided in the sequence and diamond paste (OptraFine, Ivoclar Vivadent). The intaglio surface will then be treated with hydrofluoric acid (Porcelain Etchant gel, Bisico, Schaumburg, IL, USA) for 60 seconds, rinsed, dried, silanated (Monobond Plus, Ivoclar Vivadent) and left to dry for at least 3 minutes before sealing.

The operator will polish the inlay or onlay using the polishers provided in the sequence, and may modify the color (Kolor Plus, Kerr) of pits and fissures. The intaglio surface will be sandblasted with 50 μm alumina, rinsed, dried, silanated (Monobond Plus) and left to dry for at least 3 minutes before sealing.

A dental dam (DermaDam medium, Bisico) will be used. The tooth surface will be cleaned by air abrasion (RONDOflex, KaVo, Biberach, Germany). Enamel will be etched with orthophosphoric acid (37.5%) for 15 seconds, rinsed thoroughly and dried gently 37 . Adhesive (Optibond XTR) will be applied by gently brushing the tooth surface for 15 seconds, followed by a 3-second air spray and light polymerization of the adhesive for 20 seconds. The inlay will be handled with use of a stick (Stik-N-Place, Directa); adhesive cement (NX3 yellow, Kerr) will be applied generously on the intaglio surface of the restoration. The inlay or onlay will be positioned and maintained. It may be light polymerized for 1 or 2 seconds. Excess cement will be carefully removed by use of dental floss and a curette. Glycerine gel will be applied on the limits of the restoration, followed by light polymerization of the cement for 40 seconds per face. The occlusion will then be adjusted, and the corrected surfaces and cement interface will be polished.

The primary outcome, clinical efficacy of materials, will be measured by use of the Fédération Dentaire Internationale (FDI) World Dental Federation instrument for assessing dental restorations, described in 2007 28 and updated in 2010 38 . This instrument contains three dimensions (18 items): biological (six items), functional (seven items) and esthetic (five items). Each item is assessed by clinical examination on a 5-point Likert scale (1 corresponding to a perfect restoration and 5 corresponding to a restoration that needs to be replaced), and collected in the CRF. All items but one are assessed by the dentist; the remaining item is patient-reported satisfaction. The primary outcome is the worst score for all items (ranging from 1 to 5) at 2-year follow-up (the best material will be the one with the lowest score).

Operators and evaluators, who will assign scores, will be trained in the FDI criteria by means of the e-calib web-based  software (http://zep01793.dent.med.uni-muenchen.de/moodle/) and group training sessions. They will use the evaluation kit specifically designed for evaluating the FDI criteria (EX-KIT 150/250, Deppeler, Rolle, Switzerland).

Secondary outcomes will include each item of the FDI instrument, patient-relevant outcomes, quantified wear analysis (through silicone impressions) and overall quality of the restoration (as assessed by dentists). Survival may be evaluated if the follow-up is extended.

The restorations will be evaluated after 1 week by the operator, and after 1 and 2 years by two independent evaluators (Table  1). Follow-up is planned and funded for 2 years; it may be extended to 5 years (as recommended for indirect dental restorations by the FDI) if the grant can be extended.

Investigators will use a CRF (available at http://cecoia.fr) to record all items required for outcomes analysis. The CRF comprises two adverse events forms (one concerning general health and one concerning inlay/onlay-related events). Patient data will be anonymous because patients will be identified by their inclusion number (the first letter of their first and last name and date of birth only will be registered in the CRF). A clinical research assistant (RB) will visit each center every 20 inclusions to monitor the collection of data (by checking that no CRF field is incomplete) and assess the quality (by comparing the data in the medical record, entered through the online inclusion and randomization software RandoWeb (Assistance Publique – Hôpitaux de Paris (AP-HP), Paris; http://randoweb.aphp.fr), written in the CRF). The data will be entered twice in the database by operators and checked by a data manager (more information about data management procedures is available at http://cecoia.fr: extra and protocole initial sections). Some elements in the CRF allow for checking for operators’ adherence to the protocol.

The main analysis will compare the final values of the FDI score (worst score over the three dimensions) between the ceramic and composite groups. The main analysis will be adjusted on the following pre-specified variables: inlay/onlay, premolar/molar, vital/non-vital tooth and operator 53 55 . An ordinal logistic regression model will be used. The operator variable will be modeled as a random effect. The main analysis will take into account missing outcome data by multiple imputation, with the assumption that data are missing at random. We will report the unadjusted analysis as well; that is, the contingency table showing the distribution of FDI scores in the ceramic and composite groups. The distribution of FDI scores will be compared by Fisher’s exact test. All P values will be two-tailed, with significance level 0.05.

The same analyses will be used to compare both treatments by each of the three dimensions (with an α risk of 1.7% for each dimension). Secondary analyses will also involve FDI items, quantified analysis of wear (by silicone impressions) and analysis of the overall quality of the restoration (assessed by dentists).

We will perform subgroup analyses 56 of the following variables: inlay/onlay, premolar/molar, vital/non-vital tooth, inlay/onlay volume, canine or group lateral guidance and occlusal tapping before luting of the inlay/onlay. If interaction tests are performed for six subgroups independent of each other and each at a significance level of 5% (two-sided), the risk of finding at least one false-positive statistically significant interaction (that is, due to sampling fluctuations) is 26% (= 1 − (1 – 0.05)⁶).