Use Case - Prevalent CKD Calculator
- Philip Main
- James Medcalf
Description
To calculate the care planning decisions for prevalent ckd patients.
Target User(s)
Stats team in Jupyter notebook. In due course, and when correct aggregate statistics agrees with AKC KQIP project to use in UKRDC app dashboard statistics.
Specification
A single-row-per-patient data extract for a prevalent cohort of patients in a user-specified centre and prevalence point (date on which the prevalence is calculated) who are under CKD follow-up. The prevalent CKD cohort will be initially defined as patients with treatment records containing an admisionreasoncode in the group (900,901,902,903, 92,93,94) that span the prevalence point (i.e., records that start before and end after this date or remain open).
The first deliverable is a function that enables data extraction, to a flat format, using high-level syntax, within a Jupyter Notebook. This should prioritize simplicity, though configuring output (for example choosing columns) may be considered if it does not introduce complexity. The extracted data should include the following:
Item | Definition |
|---|---|
Nhs Number (preferentially or other NI if not available) | This should be used as the patient primary identifier. Particularly when linking between the xml archieve and the ukrdc proper. |
Admission reason code |
|
From time |
|
To time |
|
Date of Death |
|
Postcode |
|
Age | decimal age at prevalence point? |
Sex |
|
Ethnicity |
|
Most recent lab calculated eGFR | Most recent egfr or creatinine sent prior to the prevalence point in the case of the later the egfr should be calculated from the test result. |
Most recent creatinine | Needed where centre does not send lab calculated eGFR. Would need to be calculated using CKD-EPI formula (2009) without ethnicity correction (ie creatinine, age and sex). Maybe one for a standard library call Phil as will be used repeatedly? |
KRT Care planning decision | One of: None (blank), Open (record still open), Assessment outcome, Closed (record closed no outcome). |
Transplant Care planning decision | As above but for transplant. |
Optionally: other biochemistry results/observations. Admit reason description, registry code type...fill in…
Second deliverable: automatic generation of aggregate stats based on the CKD care-planning cohort. These would need to be defined by the AKC KQIP project and defined for stats to test.
Questions and Answers
Do we expect multiple overlapping ckd records? If so how do we agrigate the to a one row per patient basis?
Components/Dependencies
Libraries
(delete where appropriate)
dashboard stats
Jupyter notebook
UKRDC stats dependencies
TBC
Steps
Reviewers