The CODEX Analytics app is available for more than 60 clinical trial outcomes databases. Each database is developed systematically based on relevant identification, screening, and assessment steps described in the Cochrane Handbook for Systematic Reviews of Interventions and reporting items in the PRISMA statement. Following the identification of references to be included or excluded, the clinical outcomes database is curated from medical literature and augmented to be used as an analysis-ready data set.

Note: Databases contain summary-level data for clinical trials as opposed to individual patient-level data. Detailed information regarding the development and augmentation of a database can be found in Database > Protocols.

CODEX consists of two components.

Users can choose to view data for one or several indications at the same time. Click “Change selected indications” to view a pop-up that lists all available indications.

Users can filter the database via the Filters on the left-hand side. This allows for the creation of personalized subsets of the database. The Summary section on the overview page shows an overview of the database and – in the case of the application of Filters.

To ensure the generation of informative plots and tables in CODEX, the Data Settings default option selects one analysis/representation per study. This is represented by unique study, strata, endpoint group, and analyzed population per endpoint-measure combination (see database specifications for details on column content). The user can alternatively select to show all data to see all reported measures for a selected endpoint of a study.

The following section describes the functionality and content of each page found along the horizontal navigation bar of CODEX.

CODEX opens onto the Overview tab, which provides a general overview of the database’s most reported efficacy and safety endpoints, graphical spread of clinical trial study sites, and evidence characteristics.

The Summary provides a quick overview of the full database in text and table format. In addition, the size of the original vs. the selected data set can be compared easily via the summary table.

The Study Sites Map provides a visual overview of the distribution of study sites across the globe; the more studies available within a given region, the darker the color. Hovering over a region shows the country name and the number of study sites for that region.

Two Endpoint Histograms can be found on the Overview page: the most common efficacy and the most common safety endpoints. Both histograms display the five most common endpoints of each category respectively, if available.

Three Gauge Charts give a quick overview of the percentage of randomized controlled studies (RCTs), blinded studies, and phase 3 studies in the database.

Note: You can now scroll down in CODEX – if the Evidence Characteristics plots are not visible on the Overview page, consider scrolling down.

The Database page provides a variety of summary tables about the outcomes and source database. Table columns can be used to filtering and sorting; each table can be downloaded as a CSV file.

The Database tab contains the following sub-tabs based on four categories:

The Studies page contains three types of study summaries: the Study details table provides a general overview of the study while the patient characteristics table summarises important patient covariate values within a given study; the Study summary table provides an overview of study details across all studies. The tables have the same functionality and features as the tables under the Database tab. The studies tab is completed by the Covariate overview plot (summarising covariates by study or patient number) and the covariate correlation plot.

Except for the Overview tables, the Exploration tab contains primarily plots.

All elements within the exploration tab are responsive to selections made via the sidebar filters and the user element items within the respective tab. All plots in CODEX are also interactive. Moving the mouse onto the plot area leads to the appearance of the so-called “mode bar” on the top right-hand corner of the plot. There, options such as zoom, pan, and plot download can be found. Further, additional information regarding a specific data point can be found when hovering over it.

The network plot shows the extent to which a given drug is represented in the database. The size of the drug node is proportional to the number of arms represented in the database for this treatment. The width of the edge represents the number of arms per treatment combination (two nodes). Checkboxes at the top allow for plotting by drug class and showing labels.

The time course by study plot depicts the treatment effect as observed over time. One study is shown per plot and each line represents a different study arm. Hovering on the points on the line shows information about the treatment, time, and the effect value of a given study arm. Control arms have grey plot lines, legend labels, and hover labels. Plots that do not contain control arms will use colors other than grey. The endpoint can be chosen by selecting from the dropdown menu or by typing the endpoint name. The type of plot can be changed to “comparative” to view summary statistics for comparisons to controls. Information in the corner title of the plot can be customized. The additional settings detail contains options to select the number of panels to display and to set the number of decimal places to show.

The time course by treatment plot depicts the treatment effect as observed by treatment. One treatment is shown per plot and each line represents a different study arm. Hovering on the points on the line shows information about the study strata, treatment, time, and the effect value of a given study arm. The endpoint can be chosen by selecting from the dropdown menu or by typing the endpoint name. The type of plot can be changed to “comparative” to view summary statistics for comparisons to controls. The additional settings detail contains options to select the number of panels to display and to set the number of decimal places to show.

The forest plot provides a summarising overview of the treatment effects of all studies together with a 95% confidence interval. Each row represents a different treatment group. Two types of forest plots are available: descriptive and comparative. The measure statistics as seen in the plot are summarised in Table 1. Columns can be added to both the left and right sides of the plot. The additional settings detail has options for 1) Selecting decimal places 2) Selecting the number of rows to display 3) Log transforming the x-axis for ratios 4) Reversing the data order and 5) Plotting points symmetrically around the treatment effect line.

The funnel plot depicts potential publication bias within the database. A variety of measure types for the observed treatment effect can be selected based on the type of endpoint selected in Table 1. The default index of precision on the y-axis is the inverse standard error. Information in the hover can be customized.

The endpoint correlation plot allows a visual comparison between two endpoints. Two types of scatterplots are available: descriptive and comparative. Points can be colored by treatment, treatment class, indication, or study design. A slider for smoothness can be used to adjust the correlation. Additional information is provided when hovering over each point.

As in many of the plots above, the covariate vs. endpoint plot allows for the comparison of two elements directly or respectively of the control treatment. Data points can be grouped by color: by randomized.drug.class, randomized.drug, publication.type, and publication.year.

Time bar plots complement time course plots, making it possible to view responses to treatments at a given point in time, grouped by study, treatment, and endpoint.

The overview tables show a summary of the database based on the value selected from the “Summary type” dropdown menu on the left-hand side above the table. The level of detail for the summary can be adjusted via the “Summary columns” dropdown. Several characteristics can be selected at once and easily rearranged to display them in the table in the specified order.

Settings for the pairwise meta-analysis include a selection for endpoint, plot type (descriptive or comparative), model type (fixed or random), subgroup to generate the analysis by, selections for the left and right text columns that surround the plot, and a page slider. The user can further select the number of rows and decimal places to view, set x-axis log transformation, reverse data order, and show a symmetrical forest plot. There is also an option to include/exclude imputed variability measures.

The pairwise analysis provides a summarising overview of the treatment effects of all studies together with a 95% confidence interval. Each row represents a different treatment group. The diamond at the bottom of the plot shows the result when all individual studies are combined and averaged. The horizontal points of the diamond are 95% confidence interval limits. Additional summary-level information includes the prediction interval and heterogeneity information.

The summary is an R output of the meta-object (an object created from the R package metafor) generated by the selections in the sidebar settings.

Analysis set to display in the meta-analysis plot. The checkbox allows the inclusion or exclusion of the time group.

Settings for the network meta-analysis include a selection for endpoint, intervention, plot type (descriptive or comparative), reference treatment (if comparative plot type is selected), covariate to generate the analysis by, and covariate effect (global or by intervention). As with the pairwise MA, the user can select the number of rows and decimal places to show in the plots, log transform the x-axis and show symmetrical plots as well as include/exclude imputed variability measures.

The treatment plot provides a summarising overview of the treatment effects of all interventions together with a 95% confidence interval. For the comparative plot type, each row represents a different intervention and its summary statistic against the reference treatment chosen.

The treatment table displays the intervention, covariate, and their respective summary statistics.

The study plot displays a forest plot with added prediction lines.

The network plot displays direct and indirect comparisons, which allows us to compare the effects of several interventions simultaneously.

The funnel plot residual displays residuals on the x-axis and standard error on the y-axis.

The summary is an R output of the meta-object (an object created from the R package metafor) generated by the selections in the sidebar settings.

Analysis set to display in the meta-analysis plot. The checkbox allows the inclusion or exclusion of the endpoint group.

The Settings tab contains

For any suggestions, comments, feedback, inquiries, or bugs that arise during the use of CODEX, you can reach us via Contact us in the top right corner or email us directly at codex-feedback@certara.com. Thank you!