Summary of the findings
To our knowledge, this is the first study to identify and evaluate systematically the completeness of retinal disease-related RWD sources. Our study identified many sources of RWD that provide insightful and comprehensive evidence on retinal diseases and their treatments. Of the 128 data sources identified, 64 were analyzed in detail, 10 of which were considered to be useful sources for carrying out real-world studies for most or all of the outcomes of interest, including patient information, clinical and safety outcomes, treatment information, and patient-reported and economic burden.
Value of real-world evidence in ophthalmology
RWD provides information about the effectiveness of a treatment in a real-life setting in a wider patient population than in a RCT. RWD adds to the evidence base for a disease and its treatment by providing information on topics such as long-term safety, treatment patterns, disease progression and burden of illness, which are not designed to be assessed in clinical trials. RWD are increasingly being considered alongside data from RCTs in healthcare decision-making. For instance, the US Food and Drug Administration uses RWD in specific contexts (for example, with devices) based on existing evidentiary standards. Similarly, the National Institute for Health and Care Excellence (NICE) in the UK considers RWD in addition to more traditional clinical evidence in technology appraisals. Therefore, robust data sources that permit high-quality, observational research on outcomes of interest are important to meet the increasing demands for real-world evidence from healthcare decision-makers. Among the data sources analyzed in this study, we confirmed that the Southwestern Ontario Database (Canada), the Swedish Macula Register and Medisoft (UK) record data on all outcomes of interest as recommended by the ICHOM guidelines, each scoring the full 21 points. As such, these 3 RWD sources will be useful for developing comprehensive real-world studies.
Characteristics of data sources identified
RWD have provided important insights in the ophthalmology field, particularly in terms of treatment outcomes for anti-VEGF regimens, by permitting the assessment of outcomes for patients who may not typically be included in RCTs. Our study identified that there are many sources of RWD for retinal diseases and treatments worldwide. A high proportion of the sources recorded data on patient treatment and management (90% [58/64]), with 60% of these sources capturing information on whether or not patients had received anti-VEGF therapy. In addition to treatment information, many of the data sources provided information on the patient and their disease characteristics, as well as clinical outcomes. In contrast, relatively few data sources collected safety outcomes, patient-reported burden and economic data. In total, 10 data sources (6 from Europe, 1 from Canada, 2 multinational and 1 from the USA) provide good or moderate data for multiple outcomes of interest. It is important for retinal data sources to measure multiple outcomes to be able to provide a true sense of treatment effectiveness in clinical practice; measuring visual acuity, for example, is meaningless without capturing information on treatment patterns. By providing comprehensive information on clinical, treatment, safety, human and economic outcomes, these data sources can be considered to be robust for performing real-world studies.
To assess the completeness of the identified data sources, data items for outcomes of interest were selected and sources were scored on the presence or completeness of the data captured. For some outcomes of interest, our study found that the type and depth of information available differed across the data sources. This might reflect differences in how the data are captured and/or reported in publications. To permit comparison across real-world sources, it is important to standardize the types of data items collected and the outcome definitions used. In addition, the most suitable and relevant data items should be captured. Data items collected for appraisal of the data sources in this study were based on recommendations by the ICHOM guidelines, as outlined by Rodrigues et al. (2016) . The data items collected in the completeness assessment in our study is based on recommendations for outcomes that matter most to patients and which therefore can be considered as robust data items for assessment . Although many RWD sources capture the data items recommended by the ICHOM guidelines, this study has highlighted that several sources omit key outcomes measures. Therefore, there is a need for new RWD sources that standardize and align the data collected.
Strengths and limitations
This is the first analysis to identify and assess all worldwide RWD sources for patients with retinal disease treated with anti-VEGF therapy. This analysis used a robust and systematic approach to identify all the data sources available, and collected data items that were recommended in the ICHOM guidelines. The completeness assessment profile for each data source was based on previously published methodology for a similar study assessing the location and quality of renal registries .
A systematic review of electronic literature databases was performed to identify the relevant data sources. It is possible that our analysis may have missed some data sources, although we did perform online searches of congresses and other websites, as well as conducting searches for proprietary data sources to identify any further sources that may not have been included in peer-review articles. Our searches were performed up until July 2016, so any data sources published after this date are not included in this analysis. At the time of the searches, the American Academy of Ophthalmology Intelligent Research in Sight (IRIS) registry had limited publications; as a result, this data source has not been highlighted in this study.
The bespoke grading system for data sources illustrates the breadth and type of data collected within each outcome of interest, but our study did not assess the data sources in terms of how the data are collected, the quality-check mechanisms in place, or how closely the data items collected compare with those recommended by the ICHOM guidelines. The assessment of the data sources may have been influenced by the amount of data available for a particular data source; for example, a source may have been graded 'poor' or 'unknown' for a particular outcome if retinal disease-related data from the source remains unpublished. For several of the data sources, we contacted the administrators to obtain further information that may not have been in the public domain to enable us to provide the most comprehensive assessment possible of these data sources. However, to limit the scope of the project, contacts were only pursued for a subset of the data sources. In addition, the completeness assessment was only undertaken for the multinational/international data sources, or sources from a shortlist of countries of interest. Therefore, the findings of this study should be interpreted with caution, as they are mostly relevant to international data sources or those from countries with published data. As a follow-up to this analysis, all the data sources identified in this SLR, including the proprietary data sources, could be assessed for their completeness.