Version 4.11.0.0 is a major release: it includes important new features regarding numeric outcomes. It is now possible to define "timepoints" for any given reference (understood here as a 'study') and associate a given outcome measure to a timepoint defined for the current item. Moreover, in EPPI-Reviewer version 4 (Silverlight) the data extraction form is now visible alongside the item details area, making it possible to fill-in the form while the Full Text document is visible in the PDF viewer. At the same time, EPPI-Reviewer Web has gained the ability to extract outcomes data, as well as editing timepoints. Other highlights include the appearance of a few more features in the Web version, performance enhancements (both versions) and a handful of bug-fixes.

New Features, both versions:

Outcomes data extraction and timepoints. This release introduces support for timepoints, used to record when a given Outcome was measured. Timepoints apply to a single reference (Item), they are defined by a numeric value and a unit of measure (minutes, hours, days, weeks, etc.). When extracting data for a given outcome measure, it's now possible to associate it with a single timepoint. This is useful to differentiate between outcomes measured within a single trial, when the same effect was sampled at different points in time, or to record the difference between separate trials.

The data extraction form, used to enter outcome data is now visible inline with the item details, allowing to view and read the PDF while filling in the form, moreover, when appropriate, it is also possible to associate an arm (defined at "Item" level) to a specific column in the outcome measure.

In EPPI-Reviewer Web, the same features have been implemented in parallel, so it is now possible to perform all data-extraction tasks, while using the new version exclusively.

New Features, EPPI-Reviewer Web:

In the Item Details page (full UI) simple edit operations on the coding tools are now available. Supported operations include: add a new code, edit an existing code, delete a code, reorder codes and coding tools.

In the Coding Only UI, it is now possible to view the full text of uploaded PDFs. It is also possible to assign codes to text therein.

Again in the Item Details page, the "find on Web" functionality has been added: users can now search for the current reference on Google, Google Scholar and Microsoft Academic.

In the searches tab (main page, full UI) it is now possible to bulk assign and bulk remove the search results to a given code.

Enhancements:

The routines used to fetch the data representing coding tools have been re-written from scratch. The result is a performance gain that we expect to be between 2 and 8 times faster. Importantly, the biggest performance gain should apply to bigger coding tools. This improvement applies to both version, but is particularly important for the Web version, where it is known that big reviews containing a total of 8000 or more codes (across all tools) could fail to load within the timeout period. We expect this problem to be permanently solved, now.

The Coding Progress / Review Stats figures can also take quite a long time to load. We've optimised how data about incomplete codings is fetched, making it at least 4 times faster.
Taken together, the two improvements above should ensure much shorter delays while "opening" a given review.

Bug Fixes:

When using Priority Screening, the "Review needs Indexing" option could get set to "true" unexpectedly and for no good reason. In turn this meant that EPPI-Reviewer would refuse to feed new items to screen, because that option signals that the current list was produced with outdated data. This was happening when a re-training was triggered from the Web interface and the training failed (this happens automatically when enough new screening decisions have been recorded). This bug is resolved: the "Review needs Indexing" option will not be activated when a failure occurs. Instead, the details of the failure will be logged in some detail. We expect to learn why such failures sometimes happen by inspecting these logs in the near future.

In the Web interface, when coding the the text of PDF documents, the system could break and prevent people from loading the coding details (in fact, the error was serious enough to send the users back to the login page - to preserve data integrity) if the PDF already contained some highlights (added by end users) when the file was uploaded. This problem is now solved: existing highlights are removed from the PDF when using the "in app" "View" functionality, but are shown and retained when the "Download" option is used.

In the Web version, Coding Only UI, Item Details page, when the current reference contained a very long URL, the page would become wide-enough to show the full URL in a single line. This meant that users had to scroll from left to write, in order to "see" all the details about the current reference. This problem is now solved.

When using Edge (Web version), clicking some "update"/"cancel" buttons in some particular pages would cause Edge to reload the whole application, which in turn, could result in "losing" the current data, forcing users to reload it explicitly (for example, when the list of references wasn't the default, people would then need to ask again for the specific list they needed). This problem is now solved (or at least, we've fixed it for all known occurrences).

Future plans:

We have ambitious plans for the next months, but details are still to be finalised, due to scope of the upcoming releases. The plan is to release new (and exciting) automation features, which are likely to appear on EPPI-Reviewer 4 first, to allow releasing them without additional delays. In terms of the Web version, we will keep adding existing features to the user interface. There are three main "areas" that are still missing from EPPI-Reviewer web: Deduplication, setting up Priority Screening and Meta Analysis. We expect to work on the former two, but might be delayed as we are also working on a new system to identify duplicates (which promises to be much better of the current one).