2.1

Search strategy

A comprehensive literature search was undertaken through the MEDLINE (National Library of Medicine) via PubMed, Web of Science (Core Collection) and Scopus databases to identify relevant articles through November 2015 with no limit on publication year. Specific medical subject headings (MeSH) and key words (“free-text words”) were used as search terms. The electronic database searches were undertaken independently by two of the authors (I.L.A. and A.M.E.M.) using the following search strategy words and terms: ((((((((zirconium [MeSH Terms]) OR zirconi*) OR yttria*) OR y-tzp) OR y tpz)) AND (((((((((((Air Abrasion, Dental [MeSH Terms]) OR Air Abrasion*, Dental) OR Abrasion*, Dental Air) OR Air Abrasion*) OR airborne particle) OR airborne-particle) OR silica coating) OR silica-coating) OR air particle) OR air-particle) OR sandblasting)) AND (((((((((((strength) OR resistance) OR flexural*) OR fatigue) OR aging [MeSH Terms]) OR aging) OR ageing) OR structural stability) OR phase transformation) OR long term) OR long-term)).

2.2

Eligibility criteria

The criteria for inclusion of studies were the reporting of in vitro evaluations of the effects of airborne-particle abrasion protocols on: (a) the mechanical strength, regardless of the mechanical testing configuration adopted (uni- or biaxial; static or cyclic loading) and/or (b) phase transformation on specimens with any geometry made from Y-TZP, either immediately or after any type of aging.

The final decision on the inclusion of a given study was made based on the full-text papers of the potentially relevant studies, in accordance with the exclusion criteria: only studies that had proper control groups without surface conditioning were included. Studies with experimental groups in which technical procedures were performed on the specimens after sandblasting, such as thermal treatments or sintering, were excluded, as they did not enable the assessment of the isolated effect of sandblasting. For studies assessing the mechanical strength, the data were required to be reported as mean in megapascals (MPa) and standard deviation (SD), sandblasting was required to have been performed only on the surface subjected to tensile stress during testing, and bilayer samples were not accepted. Studies evaluating the phase transformation of Y-TZP were required to report at least one outcome supported by quantitative data such as the relative amount of transformed monoclinic zirconia ( F _M ) or the transformed zone depth (TZD). Studies that did not enable the evaluation of the effect of sandblasting in isolation on the phase transformation were also excluded. Studies that did not provide all required data, even after at least two attempts to contact the authors via e-mail, were also excluded.

2.3

Screening and selection

First, the titles and abstracts were reviewed independently by two of the authors (I.L.A. and A.M.E.M); the studies were selected for full-text reading if the titles and abstracts met the inclusion criteria. In a second stage, abstracts reviewed in the first step were selected with the consensus of both authors. If a consensus was not reached, the abstract was set aside for further evaluation. In the third step, the full-text articles corresponding to the abstracts selected in the second step were retrieved and reviewed by the two authors. Inclusion in this review was based on consensus between the two investigators. Disagreements were discussed with a third author (A.F.M.). Finally, the reference lists of all articles selected in the third step were manually reviewed, and the full texts of studies that could potentially fulfill the inclusion criteria were examined.

2.4

Data extraction

A protocol for data extraction was defined and evaluated by two of the authors (I.L.A. and A.M.E.M.). Data were extracted independently from the full-text articles selected for inclusion using a standardized form in a program (Office Excel 2013 Software, Microsoft Corporation, Redmond, WA, USA). The authors categorized similar information into groups according to the two main outcomes of interest of mechanical strength and phase transformation. The authors of the selected studies were contacted via e-mail if data were missing or more information was needed when studies did not report the precise values of flexural strength (mean and SD) or t→m phase transformation (by F _M or TZD), or when the articles reporting the studies presented the results in graphs or figures.

2.5

Assessment of risk of bias

The evaluation of the risk of bias was based on and adapted from a previous study . The quality of studies evaluating mechanical strength and that of studies evaluating phase transformation were assessed separately. The quality assessment for studies evaluating mechanical strength was performed according to the following parameters: (a) randomization of specimens, (b) standardization of obtaining of samples, (c) standardization of airborne-particle abrasion procedures (time, pressure, and distance), (d) single-operator protocol execution, (e) description of sample size calculation, (f) blinding of the operator of the testing machine, and (g) performance of specimen dimensioning, test designs, and flexural strength calculations according to standard specifications such as ISO or ASTM. For the quality assessment of studies on phase transformation, the parameters (a)–(d) used to assess the studies evaluating mechanical strength were retained. Further, the blinding of the operator during phase transformation analysis by methods such as X-ray diffraction or Raman analysis was also considered. If the authors reported the parameter clearly, the paper received a score of 0 for that specific parameter; if a particular parameter was reported but was inadequate or unclear, the paper received a score of 1; and if it was not possible to find the information, the paper received a score of 2. For studies evaluating mechanical strength, papers that received scores from 0 to 4 were classified as having a low risk of bias, 5 to 9 as medium-risk, and 10 to 14 as high-risk. For those on phase transformation, papers scored from 0 to 3 were classified as low-risk, 4 to 7 as medium-risk, and 8 to 10 as high-risk.

2.6

Data analysis

For the meta-analysis, the flexural strength data of air-particle abrasion compared to those of the control groups were analyzed both globally and by subgroup. Subgroup analyses assessed the effects of different blasting parameters such as particle size, pressure, or time duration and the presence or absence of aging conditions. For subgroup analyses involving the air-particle abrasion parameters, two strata were created for each parameter.

Every possible comparison was performed; when appropriate, specific formulas were applied to combine multiple values into single sample sizes, mean values, and standard deviations for the experimental, control, or both groups within a given study . In the subgroup analyses, each distinct blasting parameter and the presence or absence of aging conditions resulted in one possible comparison, even for studies subjected to the data-combining strategy. For the studies assessing aging procedures or other interventions, such as grinding or polishing, in addition to particle abrasion, the comparisons between the abraded and control groups necessarily involved the presence of aging or the other intervention in both groups, resulting in a single possible comparison correlating to each condition.

The pooled effect estimates were obtained by comparing the means of each of the flexural strength values; the estimates were expressed as the raw mean difference among the groups. A p value ≤ 0.05 was considered to be statistically significant ( Z test). The heterogeneity of the treatment effects among the studies was assessed using both Cochran’s Q test and the inconsistency I ² test, in which values greater than 50% were considered to indicate substantial heterogeneity . All analyses were performed using a random-effect model, using Review Manager Software version 5.1 (Cochrane Collaboration, Copenhagen, Denmark).

Meta-analysis was performed only for the flexural strength data. The phase transformation data were not included in the meta-analysis, as the studies provided insufficient data and the information required for meta-analysis was missing. Thus, the phase transformation data were included only in the systematic review and analyzed descriptively.