2020-11-30 Meeting notes

• George has now done some of the outstanding tickets but still has a couple more to do. He will produce a test build for Fran this week.
• Fiona has got hold of some details for accessing the server from the Alpha study and passed them to George. George needs to remind himself what is required and try the ftp link. It may not be the complete solution. Also ftp from NBT will be blocked so we may need to get a rule in place to allow the connection. George will confirm if this is required.
• Andrew hasn't progressed the tracing code.
• Andre has completed the Nephwork app and deployed it ready for the go live. All the users have been added to the platform ready for go live. Andre will start looking at some magic and transplantfirst issues next.
• George has confirmed the changes have been made by renalware to the laborders as required. There are still a couple of changes on the Trello board that need doing and George will follow this up with renalware and get them actioned.
• George hasn't documented the rrnos process and will do so.
• George hasn't deleted the PV Membership that was added in error and is causing duplicate NHS numbers on some radar records. He also hasn't documented the process for this though there is documentation for deleting a patient 

  https://confluence.ukrdc.org/pages/viewpage.action?pageId=39616561.

• 
  

  The BAPN covid matching in October was affected by  date issue meaning matching of patients may not have happened 100% correctly but didn't affect the pedw data. Retha will let Lucy know.
• RaDaR last login date. Andrew has talked with Garry and it seems they monitor user activity and if they haven't used the system in the last 6 months the account is removed. This means they are only interested in 6 months worth of data. It was decided to keep 7/8 months worth of data present. This means Andrew will restore the missing data to get to 6th months but will then only archive when there is more than 7/8 months of data in the logs. We will need to monitor and ensure logs don't grow too big in which case a cleverer approach may be required.
• George has done the addition of batch loading functionality to test the reload process of all files to allow the correction of missing data.
• The AKI extract for UoB is currently going through opt out having been submitted this am. Once that has been done the data should be ready for release.  

Patientview has had a couple of weeks of great instability. This was directly caused by the mass messaging of a group of patients. Due to the way the messages were retrieved when one of the recipients logged in the retrieving of their  messages would cause the process to run out of memory and thus crash the PV api process. This would then mean nothing on PV would work and it would require restarting. This has been resolved by deleting the errant messages but hasn't addressed the underlying issue of out of memory causing the process to crash and not recover. There is now a ticket in place [PV-4960] Amends to Conversation Recipient Logic - Solid State Group JIRA to prevent further IG issues of large group sends. Fiona and Shaun will look at the proposal as George thinks it is more than is required especially with PV being replaced. Only the minimal changes required to protect against further issues needs to be done. Shaun also asked that people ensure he is included in any emails as he was away and thus unaware of the issues so could have responded to the PV mailbox incorrectly if he hadn't realised what had happened. Fiona also asked that she be updated more frequently. 

Fiona also highlighted the fact that PV reconnected where she left off after each restart which she saw as an IG breach. On a single user machine this is not an issue but could be on a shared computer. George will verify that there is a ticket about this on SSG jira.

RDR-731 RDR-730 Barts & Kings data into RADAR Data appears to have been loaded but there are many duplicates. The same result must not be loaded twice into RADAR, RDR 730- Duplicated Meds

The issue here is how to remove the overlapping results. The duplication has been caused by the two feeds RDA and PV both covering the same period initially. In theory the RDA is more complete so the PV version of results and medication should be removed from RaDaR. This needs to be done at the UKRDC/RaDaR exporter end. The most logical solution is to delete the PV entries that overlap with the RDA feed. This should then allow the patient data to resync and the PV values should be removed. This will be an issue with all the new RDA feeds if a previous PV feed has existed and the data overlaps. Everyone needs to look at at the best way to achieve the deduplication and how it might be avoided with new migrations.   

RDR-709 & RDR 729 - Fatal error on trying to access Meds First reported 2/4/20 My solution would be load the name but not the units.  If  it does not match RADAR list. Leave it blank.

This issue remains and is caused by "invalid" units Fiona would like the invalid units blanked so at least meds are displayed.

RDR-741 Add min Max to export Needed for Dataset and PKB

The max min fields are not being exported when the data is exported for the admin menu list of observations whilst there are value sin the database. 

RDR-728 – Must have RADAR demographics UKRDC has a lot of missing data

RaDaR exports are using the ukrdc demographics in preference to the RaDaR ones when exporting when it should always use the RaDaR demographics as these are usually more complete.

RDR-695 Required for Biomarker data load

The barcode table has a number of duplicate barcodes (8) in them which makes the linking of the biochemistry results difficult as they could apply to two different patients. The duplications need resolving and corrected on RaDaR.

Retha sort some clarification over the PHE extract and why Shalini was finding discrepancies in numbers of audit patients. Georges criteria mistakenly referred to DOB instead of DOD meaning he included only patients alive at the start of 2019 also it is possible Shalini was using the wrong reference list.  George has asked for further clarification from Shalini to understand if the wrong comparisons were being done of if in fact there is an anomaly.