Description

The existing cohorts will change three rigorously defined cohorts. One new, one incident, one prevalent. To achieve this the core cohorts of patients being used to do the calculations will need to be tweaked. Then the algorithm being used to do the calculation will also need to be tweaked. This will mean incident and prevalent cannot be defined more recently than 90 days ago.

Target User(s)

Components/Dependencies

Libraries

(delete where appropriate)

• dashboard stats

UKRDC stats dependencies

• abc base class

• Dialysis stats calculator

Steps

• Prototype new algorithm

• Rewrite calculator

• Future proof design to ensure it’s fast and compatible with caching

Algorithm

for a reference time T, a start window time of t0 = T-90 and end window of t1 = T+90

1. Get every treatment with a modality of TX, HD, and PD in a window t0 < t <t1 for any patient at the unit who has at least one treatment coincident with the window t0<t<T.

2. For each patient calculate the beginning of the a window defined as the period they are under treatment with no gap of greater than 3 months or transfer in.

3. For each patient calculate the end of the window defined as above, their CKD diagnosis status and the length of the time window.

4. Calculate the most recent treatment modality at time T.

5. Patients where the beginning of the time window is in the range t0 <t<T and whose treatment window overlaps with T are considered new. Patients whose treatment window coincides with T are considered prevalent. New patients with CKD diagnosis or a treatment window of greater than 90 days are considered incident.

6. Aggregate stats based on satellite unit (needs futureproofing against dataset changes), cohort, modality type, and in the future longitudinally (in time).

Reviewers

@James Medcalf

@Shalini Santhakumaran