Design and registration
The design and implementation of this protocol were conducted in strict accordance with Preferred Reporting Items for Systematic Reviews and Meta-analyses Protocols (PRISMA-P) statement (checklist in Additional file 1) . The protocol of this study was registered in the international prospective register of systematic reviews (PROSPERO). The unique registration ID is CRD42020204053.
Type of participants
All patients undergoing spinal surgery will be included in this study. There is no limitation on age, gender, nature of disease, focus of disease, etc.
Type of intervention
All clinical studies that used 3DP for preoperative planning and spinal surgery procedures will be included in this study. The 3DP equipment, software and materials are not restricted. Control interventions in the control group was traditional spinal surgery without 3DP including CT-based computer-assisted guide system, X‑ray fluoroscopy, and other methods. For the studies that comparison of two experimental interventions with no control intervention group will be excluded.
Type of Outcomes
The primary outcomes include the accuracy and precision, perioperative blood-loss volume, and X-ray fluoroscopy times. Other outcomes such as complications, the operation time, Visual analogue scale (VAS), Japanese Orthopedic Association (JOA), the Neck Disability Index (NDI), hospitalization length, and patient expenditures were defined as secondary outcomes.
Type of studies
Randomized controlled trials (RCTs), case-control studies, and cohort studies concerning the use of 3D printing in spinal surgery will be included in this review. The reviews, abstracts, case reports, case series, editorials and opinion pieces, letters to the editor will be excluded in this review. This review will exclude non-clinical research such as laboratory research, animal research, and cadaver research, etc. The studies that 3D printing technology was not used for spine surgery planning or surgery procedures such as education, surgical, and simulation purposes, etc. will be excluded. The studies that 3D printing technology was used to develop patient-specific instrumentation (PSI) (e.g. cutting jigs etc.) also will be excluded. Only the studies that published using English will be included.
Three English online databases will be searched from inception until August 31, 2020. They are EMBASE (via embase.com), Medline (via PubMed), and Cochrane Central Register of Controlled Trials (CENTRAL) in the Cochrane Library. The English terms were used individually or combined ‘‘three-dimensional printing (3DP),’’ ‘‘additive manufacturing technique,’’ ‘‘rapid prototyping (RP),’’ ‘‘spine,’’ ‘‘spinal fusion,’’ ‘‘spinal surgery’’.
Other literature materials will be obtained by searching the reference list of qualified studies.
Data management and processing
Document collection and management
Document records retrieved according to the pre-defined search strategy will be imported into the document manager, EndNote X7. Duplicate document records will be identified and deleted based on related information such as title, publication date, author, and periodical, etc. The titles and abstracts of the remaining literature will be reviewed and studies that might meet the inclusion criteria will be identified by two independent authors. After two authors read the full text of the literature independently, qualified literature will be included according to the inclusion criteria established in this study. The final decision will be made via through group discussion with a third reviewer if there is uncertainty at this stage. The process of study inclusion will be carried out and reported in accordance with the PRISMA flow diagram (Fig. 1).
Data extraction (selection and coding)
The data extraction process will be completed by two authors. A reviewer will use a tabular summary to guide extraction and record the data and the other author will check it. The following information for each report will be extracted: 1) General information such as country, region, author, title, journal, publication date, etc., 2) Information about the research object such as the source of patients, the nature of the disease, the severity of the disease, and the location of the disease, etc., 3) Information related to the operation, such as the location of the operation, the method of the operation, the level of the surgeon, the biological materials that may be used in the operation, etc., 4) Information related to outcome evaluation, such as outcome evaluation indicators, outcome evaluation tools, outcome evaluation time, follow-up information, etc., 5) Information related to study design, such as randomized controlled studies, case-control studies, cohort studies, etc.
Conversion to the desired format of the data will be performed before data analysis, for example, standard error will be converted into standard deviation. For the studies that are not pair-wise comparisons of interventions, the data of two groups that best meet the inclusion criteria will be extracted. The Cochrane review writing soft (RevMan5.3) will be used to manage and synthesize data. The final decision will be made via through group discussion with a third reviewer if there is uncertainty at this stage.
Risk of bias (quality) assessment
The MINORST (methodological index for non-randomized studies) item recommended for non-randomized controlled interventional studies in surgery will be used to assess the quality of non-randomized controlled studies. The “Risk of bias” (ROB) table will be used to assess the quality of randomized controlled studies . This stage will be completed independently by two authors. The final decision will be made via through group discussion with a third reviewer if there is uncertainty at this stage.
Handling of missing data
When encountering missing data at any stage, we will do our best to obtain the required data, including but not limited to contacting the original author via email. If data acquisition fails, only the available data will be analyzed. The potential impact of missing data on the final composite result will be discussed in the discussion section.
Identifying and addressing heterogeneity
The chi-squared test will be used to identify whether observed differences in results are compatible with chance alone. The inconsistency across studies will be quantified using I2 statistic to assess its impact on the meta-analysis. When I2 value is less than 75%, which means that the included studies have good homogeneity, the overall effect will be synthesized in a meta-analysis. Otherwise, a series of measures will be performed to deal with heterogeneity such as check the data, perform subgroup analysis, carry out meta-regression, perform a random-effects meta-analysis, change the effect measure and exclude the studies, etc.
Detecting reporting biases
Funnel plots asymmetry will be used to distinguish the reporting biases only when there are at least 10 studies included in the meta-analysis for each outcome respectively, because when the number of studies is too small, the test power is too low to identify true asymmetry32. Potential source of funnel plot asymmetry will be mainly considered from the following aspects: differences in methodological, intervention effects, language biases, location biases, delayed publication, etc. The different possible reason for funnel plots asymmetry will be interpreted and discussed in part of discussion for this study.
Planning the analysis
An overall statistic will be performed using Cochrane Review Manager (Revman5.3) if quantitative synthesis is appropriate. One preferred method for each outcome, the fixed-effect method or the random-effect, will be performed based on whether there is heterogeneity between studies. Otherwise, only the qualitative analysis will be performed.
Subgroup analysis will be undertaken base on the results of heterogeneity test and the number of included studies for each outcome. Subgroup analyses will be done for subsets of interventions (such as different surgical approaches), subsets of patients (such as children, adolescent, adult, and elderly), subsets of surgical sites (cervical spine, lumbar spine, thoracic spine).
Sensitivity analysis will be done to identify whether the overall results and conclusion are affected by the different decisions that could be made during the review process. It will be undertaken completed in two steps: first, including all included studies in the meta-analysis, and second, deleting the poor quality studies (such as small size, low methodological design, ultra-wide confidence interval, etc.) to identify the influence of special studies on the overall results and the conclusions.
Grading the quality of evidence
The Grading of Recommendations Assessment, Development and Evaluation (GRADE) system will be used to evaluate the quality of evidence for outcomes which rate the quality as very low, low, moderate, or high levels.