Version 18.104.22.168 is considered a major upgrade as it contains a non-trivial new feature: it is now possible to directly extract data about specific arms within a given study. It also marks the first release where new features are made available on the usual (Silverlight-dependent) EPPI-Reviewer v.4 and the new web-based (does not require Silverlight) "Coding App". This release also includes also some minor enhancements and bugfixes.
New Feature: study Arms.
Clinical studies and many other experimental designs normally include at least two "Arms" (intervention and control arm). In EPPI-Reviewer, extracting the characteristics of each arm independently has always been possible, but neither intuitive nor easy to manage; this is because the main structure used to extract data (codesets) is common across the whole review and does not lend itself to easily describe 2 or more sub-parts of a given Study(/Item). To make data extraction against arms more manageable, we have introduced a new structure that allows to manually define arms for each included study. Users can now create, edit and delete Arms in the "Document Details" window (via the new "Study arms" tab).
When Arms are present for a given Study(/Item), a dropdown menu appears on the top of the "codes" area. Users can thus change the arm (a "Whole Study" option is always present and selected by default) and apply codes / "additional text" and/or selected text from the full-text document, to a specific arm.
To make this feature useful, all reports have been modified so that, whenever a code has been applied to an arm, the arm name will be mentioned accordingly.
A few things to note:
- Deleting an arm can be a very destructive action if and when many codes have been applied to the arm in question: upon deletion, the arm will be irreversibly destroyed, along with all the data associated with it. For this reason, the User Interface checks for codes applied to the to-be-deleted arm, warns the user if there are any, and forces the user to acknowledge that data will be destroyed.
- For convenience, reports include the Arm name, and not its unique ID number. Thus, the User Interface forbids the creation of arms with identical names within a given Study(/Item).
- It is not possible to assign codes from a Screening codeset to arms; when arms are present, screening codes can still be applied to the "Whole Study" level. Since we expect Arms to be identified after screening, we have added this restriction in order to avoid adding (potentially time-consuming) complexity to the "reconciliation" phase of screening exercises (for example, it's not clear what people should do if a item has been marked as excluded on the study level but an arm has been included).
Minor changes and bugfixes:
In the main user interface, it is now possible to create a new review immediately after logging on. Previously, in order to create a new review, it was necessary to open an exsisting review and then select the "my info" tab.
The procedure executing searches on free text has been tweaked slightly. Searching against free text can be computationally very expensive (as it allows for complex searches and might need to "look at" a lot of data), for these reason, such searches are known to be slow and prone to timing out. We made a small change to the searching routine, making free-text searches somewhat faster and more reliable.
EPPI-Reviewer 4, bugfix: there was a bug which allowed the same code to be added to the same item twice (if/when the user interface failed to prevent this). This bug had adverse consenquence only when running some reports and comparisons. It was introduced with version 22.214.171.124 (which made the new Arms structure visible to internal collaborators) and removed shortly after.
Coding App, bugfix: attempting to apply/edit the coding of an item which had its coding "locked" resulted in an error and made subsequent coding attempts fail - to keep working, users had to log off and reopen their review. This problem is now resolved.