This is a release almost entirely dedicated to bug-fixing (one bugfix includes also some small "enhancements"). It does not imply that we have stopped implementing new features, in fact, the opposite is true. The "new features" line of work has been proceeding and we have very substantial new things in the pipeline, however, these take time, so we decided publish this release, which includes the bug fixes that we thought should not be delayed. The headline bugfix resolves a problem that prevented users from starting new Cochrane reviews.
BugFix: Cochrane Reviews
Cochrane reviews done in EPPI-Reviewer include a few "unique" mechanisms: they are licensed via Cochrane Authentication (users authenticate with their Cochrane credentials and receive specific usage rights), while Cochrane systems control who has access to a given review (via the corresponding list of Authors). Until December 2020, Cochrane reviews also needed to be checked out from Archie in order to become editable in RevMan 5 and/or EPPI-Reviewer. This mechanisms was removed, but EPPI-Reviewer still relied on it, which meant that starting to work on a new Cochrane review in EPPI-Reviewer became impossible, without the help of our support team. To resolve this, we have changed how EPPI-Reviewer handles Cochrane reviews, so to make the "checked in/out" status of reviews irrelevant.
The user interface (both version) has changed accordingly, in a way that we believe is both simpler and self-explanatory.
BugFix: Uploaded documents in Item Details
We have recently added a new "open the first PDF document" icon in the "Item Details" page of EPPI-Reviewer Web. This exposed a bug that has been sitting unnoticed for quite a while. Especially in periods of "high workload", it was possible that the list of documents uploaded to a given reference might take one or a few seconds to reach the client, after changing the current item. This in turn made it possible for users to "View" a PDF that belonged to the previous item, right after moving to a different reference. To resolve this, upon changing item, the list of uploaded documents is now immediately cleared (while the new list is being retrieved). We have also improved the "please wait, I'm receiving data" animations that appear in the Item Details page. They should now better reflect what is being "received" (coding data or other data pertaining a reference) and be more reliable in their important function of preventing changes from being user-initiated at times when data is being updated.
BugFix: "search string" in new sources
In both versions (4 and Web), when importing new references, it is possible to specify the "Search String" that was used to obtain a given list of references. When importing via the PubMed tab, the search string is automatically saved as well. Most of the times the search string will be a complex and long boolean string. Unfortunately a bug interfered, and would truncate the search string at 1000 characters length. This happened only when importing new items, editing existing searches and saving the search string in this manner did not suffer the problem. However, especially in PubMed searches, where the search string is saved automatically, this resulted in saving a partial search string (for strings longer than 1000 characters). The problem is now resolved, although we regretfully do not have any way to rebuild the data that was never saved as a result of this bug.
BugFix: "get new duplicates" in large reviews
In large reviews, with hundreds of thousands of references, especially when they contained many duplicates groups, clicking on "get new duplicates" could fail systematically, which would in turn prevent to perform most/all deduplication activities. This was due to a subroutine that was timing out (taking more than the allocated time). The problem is now resolved: since "finding new possible duplicates" can take many minutes in such circumstances, we resolved this by "allowing" the routine to run for longer (from 30 to 90 seconds).
BugFix: JSON coding reports in EPPI-Reviewer Web
[Update: 03/02/2021] In EPPI-Reviewer Web, the "Outcome" elements of JSON coding reports were missing the OutcomeCodes section. Moreover, three elements (ItemTimepointId, ItemArmIdGrp1 and ItemArmIdGrp2) had a slightly different name, compared to the JSON reports produced by version 4 (the 'd' in "Id" was a capital "D"). Both issues are now resolved.
BugFix: deleting timepoints in EPPI-Reviewer Web
[Update: 15/02/2021] In the Item Details page, when a timepoint is deleted, the UI of EPPI-Reviewer Web could temporarily make the wrong timepoint disappear (i.e. not the timepoint that was deleted). The deletion actually happened correctly, but what was shown to the user was wrong. You could go to the next item and back, to find that the "correct" timepoint had been deleted. This problem is now resolved.
Enhancement: added Arms data to Outcomes tables in EPPI-Reviewer Web
[Update: 15/02/2021] Coding reports (quick coding report, per-item/tool/reviewer reports, per-item comparison reports) will include Outcomes tables, whenever this data is present. However, the table did not include information about Arms. Every numeric outcome can be assciated with two arms and this data is now included in the Outcomes tables, in its own new column, which remains empty if no arms are assigned to the current outcome.