Screen readers such as JAWS have a difficult time understanding and translating data that is visually laid out in a table format but is not really a table. When a screen reader is reading something like the items in a bootstrap grid (such as the list of search results produced by a patron search, a list of items out on a patron's account, a list of bills, patron summary info, etc.), it reads every element continuously but does not explain to the user how the grid is laid out
Example of how NVDA (free screen reader software) currently reads the column headers and the first line in a patron search results grid (I hid some of the columns to simplify the example - screenshot attached):
Number
Checkbox Not Checked Row Select a Column
ID
Card
Profile
Visited Link Last Name
Visited Link First Name
Clickable Visited Link 1
Checkbox Not Checked Select Row
Clickable 2799
Clickable 373565
Clickable Patron
Clickable Smith
Clickable Alexandra
Same example if it were in an HTML table:
Table with two rows and seven columns
Row 1
Column 1 Number
Column 2 Checkbox Not Checked Row Selector Column
Column 3 ID
Column 4 Card
Column 5 Profile
Column 6 Visited Link Last Name
Column 7 Visited Link First Name
Row 2
Number Column 1 Visited Link 1
Row Selector Column Column 2 Checkbox Not Checked Select Row
ID Column 3 2799
Card Column 4 373565
Profile Column 5 Patron
Last Name Column 6 Smith
First Name Column 7 Alexandra
Bootstrap grid navigation might be improved by using aria to make screen readers understand them as tables, but it will require experimentation and testing to be sure.
Related bug in the OPAC: https:/ /bugs.launchpad .net/evergreen/ +bug/1467645