Getting medical histories for participants, September 2022¶
There is a more up-to-date version of this tutorial from August 2023.
Description¶
The Genomics England dataset includes a rich array of clinical data for all participants, rare disease probands and relatives, and cancer participants. Beyond the phenotypes recorded when participants were recruited into Genomics England, medical history was retroactively retrieved from NHS England for all participants and continues to be updated, allowing you to analyse secondary phenotypes, common disease and risk factors.
This training session will introduce you to the type of data we have available, including hospital episode statistics and mental health data, and the time periods when different data types were collected. We will show you how to access these data in table and graphical format using Participant Explorer, and compare medical history between participants. The raw data are stored in LabKey, so we will cover the tables that include these data and their structure, plus how to access these programmatically.
Timetable¶
14.00 Welcome and introduction
14.05 NHS England data in the RE
14.15 Mental health data in the RE
14.20 Accessing NHS England data with Participant Explorer
14.30 Comparing participants’ medical history with Participant Explorer
14.40 LabKey tables: Hospital Episode Statistics
14.50 LabKey tables: Mental Health
15.00 Accessing medical history programmatically
15.15 Questions
Learning objectives¶
After this training you will be able to:
- Understand what medical history data is available for participants in the GEL RE
- Visualise and compare medical histories using Participant Explorer
- Access the LabKey tables of medical history data
Target audience¶
Trainees who are familiar with the RE who wish to learn more about finding participants’ medical histories. This training session will include a short section on accessing data programmatically, so some familiarity with Python or R is helpful but not required.
You are only allowed to attend this session if you are eligible for data access. This means that you are a Research Network member, have met the necessary verification checks and passed our Information Governance training course. If you do not meet these criteria by 19th September 2022, you will be unregistered for this session.
Time¶
September 20, 2022 02:00 PM in London
Materials¶
You can access the redacted slides and video below. All sensitive data has been censored.
The notebooks used in the training session can be found in the RE under:
/gel_data_resources/example_scripts/workshop_scripts/medical_history_20220920
Give us feedback on this tutorial
Q&A
Is the information as to when these particular data are collected from and to listed anywhere? The data dictionary does not mention that A&E data was discontinued, and that ECDS data is only connected from 2018ish.
The graph of longitudinal period coverings emily showed earlier is present in the release notes. There is also a table below it with the exact date range for each table
Are only the ICD10 codes used for the diagnosis? I have read in the guide that before 1995, ICD9 codes were used.
the ICD codes will contain both ICD9 if the entry from before 1995, for entries form 1995 the entry will be ICD10
In the cancer registery, is there discharge information?
The table cancer_register_nhsd does not contain any discharge information. It has details on registration year, diagnosis date, cancer site, cancer type and cancer behaviour
Are comorbidites for participants recorded anywhere before the ecds table was opened?
Unfortunately comorbidities are only currently present in the ecds table
So there’s no information from prior to 2018?
Sorry, missed this question. There is some comordities data in the NCRAS tables which may be prior to 2018 but NHSE comorbidities are only currently from 2018 in ecds.
I believe their conditions may be listed in a differently named column (don't know it off the top of my head, sorry!)
Are these data available for only the probands or are they also available for family members who have genetic data available?
live answered
Brilliant. Thank you!
How many patients does GEL have digital records for? Thanks! Amy
There are around 90K participants in the 100,000 genomes project data release.
Is there any specific dataset covering intellectual disability participants?
these types of groupings are listed in the rare_disesases_participant_disease table which can be used in conjunction with other tables to help build your cohort
Just for confirmation, the date of birth is not available, just the year of birth, right?
Correct
If a genomic analysis was done previously, on an earlier release of the aggregate dataset, can you access the correct demographics from LabKey to match with the dataset which you used for the genomic analysis (given the time point when you did that analysis)?
The aggregate is compiled from the delivered genomeVCFs for each participant. It has filters applied to ensure that non-consented participants cannot be selected. There is no dynamic update on the aggV2?
Since there are many diagnosis columns, the epistart date to which diagnosis corresponds? (using the hes_Apc dataset) Since there are many diagnosis columns, the epistart date to which diagnosis corresponds? (using the hes_Apc dataset)
The multiuple columns allow clinicians to assign miltuiple codes for each participant at each visit. You will find multiuple rows for each participant, that will be one row for each visit or consultation event. Sorting these rows by date will allow you know the timeline for the diagnoses.
Does the NCRAS table contains data from the national cancer registry and analysis service? How far back do these records go? Are they less complete than the NHS records?
There are 8 NCRAS tables, av_treatment and av_tumour go back the furthest to 1985. The NCRAS tables contain a lot more data than the cancer_register_nhsd table
*11 tables! And to clarify on my previous point, the data does not go back as far but includes data not present in the cancer_register_nhsd table such as details of care plans and specific treatments. You can look in the data dictionary for more details on these tables
Where are these notebooks? I can find the jupyter notebook but not the R markdown
it may not have the access permissions set just yet
How do we get access to Tier1 variants from rare disease in VCF format and query them using bcftools?
Once you have the list of participants that you are interested in you will be able to get the vcf file paths from the genome_file_paths_and_types table that you will be able to send to BCFtools.
Do we need to do this via HPC?
So could it become impossible to access the correct demographics for an earlier aggV2 version?
The aggV2 is purely an aggregated genomeVCF for the 100,000 Genomes Project, the additional data, of the type that Emily has been presenting, will be up to date.
What is the difference between the main program data and NHS GMS data released recently? I am not sure I have understood it correctly.
The main programme data is the data on participants recruited as part of the 100K genomes project. The NHS GMS data is data on patients recruited through the NHS GMS. Similar data should be available on both sets of participants, although we do not currently release any NHSE clinical data for the NHS GMS participants
Could we get the list of proposed training sessions emailed? I can shared with our team who are all GeCIP trained and we might be able to provide input
live answered
Perfect! Thanks
Are cancer registered users able to access rare-disease BAM and VCF files?
The only restriction will be at the participant level, once a participant withdraws their consent we remove their data from future research
Is there any limit regarding the queues in the hpc cluster?
the scheduler that we use has a “fair share” routine that looks to ensure that all researchers are able to perform their work
Where can I find the pedigree information for the probands.
this information is only available for Rare Disease participants in the rare_diseases_pedegree_member table
just to add to this, the Participant Explorer also displays a list of relatives on the participant details page (including a link that opens the pedigree_member table in labkey)
Can you remind me how to find the data dictionaries? Sorry, I zoned out when you showed us
in the RE user guide under “Data in the RE”, there is a page for each data release, with links to the data dictionary and release note at the bottom.
Thanks for your answers so far - I am still not certain whether it is possible to access the summary demographics for an earlier version of aggv2 There is only 1 version of the aggV2, it contains only variant information.
For demographic information you would need to use the data sources that Emily has presented here
Which tables in the medical history are considered the most reliable ones to use?
live answered
Please, are you including how to run code scripts (typical examples) with GEL data in the documents you are sending to us?
live answered
How can I get detailed information of the variants like for allele 1 and allele 2 for the proband or for the relative for rare disease
https://vimeo.com/manage/videos/742691461/d9fdde40c6