Study design
This systematic review was registered on PROSPERO (CRD42020170531). It has been reported according to the guidelines of Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) Protocol and checklist (additional file 1) [32]. In brief, this study will be performed in four steps: (a) multiple literature sources will be searched to examine relevant records; (b) titles, abstracts, and full-text identifying will be carried out in accordance with predefined eligibility criteria; (c) all essential data will be extracted; and (d) a recommended study quality assessment tool will be utilized to appraise study quality before a meta-analysis will be pursued.
Eligibility criteria
This study consists of following inclusion criteria: (a) only randomized controlled trials (RCTs) will be eligible if they assess the efficacy and safety of DTFC alone in patients with POAG which meet the criteria; (b) we will include all RCTs involving participants with a confirmed diagnosis of POAG in spite of country, race, gender, age, and severity of POAG; (c) RCTs will be included if they randomize participants to utilize DTFC or to other comparators, such as placebo; (d) studies published up to the present in any language and publication status will be included.
The exclusion criteria are presented as follows: (a) all animal study, review, case report, uncontrolled trial, non-RCTs and quasi-RCTs will be excluded; (b) other types of glaucoma will be excluded if they are elaborated clearly, however, any undefined type of glaucoma will be considered for further identifying; (c) glaucoma secondary to other diseases will be excluded; (d) any other treatment combined with DTFC in the experimental group, and any forms of DTFC as comparator in the control group will be excluded in this study; and (e) trials report one of the outcomes of interest.
Outcome measurements
The outcomes of interest for analysis will include primary outcomes of mean intraocular pressure and best corrected visual acuity; and secondary outcomes of contrast sensitivity, bioelectric activity of the retina, rate of progression of glaucoma, quality of life (as assessed by 36-Item Short Form Survey), and adverse events.
Information sources and search method
A comprehensive search will be performed to identify relevant records from origin to the present in the electronic databases of Cochrane Library, MEDLINE, EMBASE, CINAHI, ACMD, China National Knowledge Infrastructure, and WANFANG databases. We will not apply limitations to the language and publication status. We will only consider RCTs that appraised the efficacy and safety of DTFC for POAG. We have summarized search strategy sample for Cochrane Library (table 1), and will create similar search strategies for other electronic databases. The search terms include “glaucoma”, “intraocular pressure”, “ocular hypertension”, “intraocular hypertension”, “open-angle”, “primary”, “optic neuropathy”, “timolol”, “Timoptic”, “Istalol”, “Timoptic-xe”, “dorzolamide”, “Trusopt”, “fixed”, “combination”, “randomized controlled trials”, “random”, “randomised”, “randomly”, “allocation”, “placebo”, “blind”, “clinical trials”, and “controlled trials”. In addition, we will scrutinize other sources, such as Google Scholar, conference proceedings, and reference lists of included trials.
Table 1
Search strategy used in Cochrane Library database
Number | Search terms |
1 | Mesh descriptor: (glaucoma, open-angle) explode all trees |
2 | ((glaucoma*) or (intraocular pressure*) or (ocular hypertension*) or (intraocular hypertension*) or (open-angle*) or (primary*) or (optic neuropathy*)):ti, ab, kw |
3 | Or 1–2 |
4 | Mesh descriptor: (timolol) explode all trees |
5 | (dorzolamide) explode all trees |
6 | ((timolol*) or (Timoptic*) or (Istalol*) or (Timoptic-xe*) or (dorzolamide*) or (Trusopt*) or (fixed*) or (combination*)):ti, ab, kw |
7 | Or 4–6 |
8 | MeSH descriptor: (randomized controlled trials) explode all trees |
9 | ((random*) or (randomised*) or (randomly*) or (allocation*) or (placebo*) or (blind*) or (clinical trials*) or (controlled trials*)):ti, ab, kw |
10 | Or 8–9 |
11 | 3 and 7 and 10 |
Study selection
All citations will be imported into Endnote X9 and duplicates will be removed. Two researchers (YXQ and HWL) will independently identify titles/abstracts of sought records to eliminate unrelated studies. Then, full papers of potential trials will be obtained and further inspected against all eligibility criteria. If any differences are identified, we will invite a third researcher (QS) to solve them by discussion and a final decision will be reached. Any reasons for excluded studies will be listed. The process of study selection will be reported following PRISMA statement, and will be exerted in a PRISMA flowchart.
Data extraction and management
Two researchers (XJS and LH) will independently extract data from all included RCTs using a predefined, structured and standard data-extraction form. Any divergences will be resolved through discussion with a third researcher (QS). The extracted information consists of study information (e.g. title, first author, year of publication, and country), patient characteristics (e.g. diagnosis criteria, and eligibility criteria), study methods (e.g. sample size, randomization, blind, and concealment); details of intervention and controls (e.g. treatment types, dosage, and duration), outcome measurements, results, findings, follow-up information, adverse events, and conflict of interest. We will express continuous data using means, standard deviations, standard errors with 95% confidence intervals (CIs), and dichotomous data using frequencies or percentages (%) with 95% CIs.
Dealing with missing data
We will contact primary authors to request any unclear or missing data. If it can not be obtained, we will analyze available data using intention-to-treat analysis, and will discuss its possible affects to the study findings.
Risk of bias assessment
Two researchers (YXQ and HWL) will independently appraise methodological quality for all eligible RCTs using Cochrane Handbook for Systematic Reviews of Interventions Tool [33]. It covers seven aspects, and each item is further rated as “high risk of bias”, “unclear risk of bias” or “low risk of bias”. Any disagreements will be resolved by a third researcher (QS) through consultation and a consensus will be reached.
Quality of evidence rating
Two researchers (XJS and LH) will independently assess the overall strength of the evidence using Grading of Recommendations Assessment, Development and Evaluation tool [34]. It includes study limitations, inconsistency of results, indirectness of evidence, imprecision, and reporting bias [34]. It categorizes the quality of evidence in one of four levels: high quality, moderate quality, low quality, and very low quality [34]. Its results will be demonstrated in the table of Summary of Findings. A third researcher (QS) will help to solve any disagreements.
Statistical analysis
RevMan 5.3 software (Cochrane, London, UK) will be utilized to perform statistical analysis. In terms of treatment effect measures, mean difference (MD) or standardized MD and 95% CIs will be used for continuous outcomes, and risk ratio and 95% CIs will be utilized for dichotomous outcomes. We will appraise the inconsistencies cross studies using I² test [35]. We define I² ≤ 50% as having minor heterogeneity, and will use a fixed-effects model to pool the data [36]; while I² >50% as having remarkable heterogeneity, and will apply a random-effects model to synthesize the data [37]. If it is possible, we will conduct meta-analysis when minor heterogeneity across sufficient data on outcomes is extracted. If the outcome data can not be pooled, we will present a narrative analysis of individual trials. If sufficient data is available, subgroup analyses will be performed to identify the potential sources of heterogeneity according to the variations in study and patient characteristics, different types of interventions and controls, and different study quality. A sensitivity analysis will be conducted to test the stability of conclusions by eliminating low quality trials. If necessary, we will perform a funnel plot and Egger’s regression test to check reporting bias when over 10 RCTs are included.
Dissemination
This study will be published in print, conferences or by peer-reviewed journals.
Amendments
Any changes to this protocol will be noted with reference to saved searches and analysis.