Aplicación de la base de conocimiento

1. Identifying patients likely to have been receiving renal replacement therapy

2. Confirming that renal replacement therapy was for the treatment of ESRD, rather than recoverable AKI

3. Restricting the cohort to an adult and incident cohort of treated ESRD patients all with upto three years follow-up (i.e. removing prevalent ESRD patients; those who had started RRT before the HES dataset began)

A summary flowchart of the procedures is found in Figure 2-4.

2.7.1 1^st HES Step: Identifying codes and potential patients treated with ESRD and/or renal replacement therapy

For the first step, diagnostic and procedural codes relevant to RRT and ESRD were manually cross-mapped through the versions of coding manuals. The diagnostic RRT codes were identified from the ICD versions 7-10 and categorised into terms relating to transplantation (incident or prevalent) or dialysis (including terms for haemodialysis, peritoneal dialysis and dialysis ‘unspecified’). Procedural codes relevant to RRT were identified from the OPCS versions 2-4 and categorised into similar categories: dialysis and transplantation. A summary list of the descriptions of these codes is shown in Table 2-1.

In the HES derivation, the codes for ESRD and renal replacement therapy were searched for in all patients at the episode level of each hospital admission. Patients with no mention of any of these ‘RRT-related codes’ were excluded, reducing potential patients to from 56 million to 140,616. Patients identified as having residency (identified through postal districts)

outside England (n=1,598) were then excluded as there was potential of discontinuity of follow-up.

Table 2-1: Diagnostic and procedural codes used to identify patients with renal disease in versions of A) International Statistical Classification of Diseases and Related Health Problems and B) Office of Population Censuses and Surveys Classification of Surgical Operations and Procedures

2.7.2 2^nd HES Step: Distinguishing maintenance from temporary RRT

Thereafter, the records of the remaining 139,018 patients were interrogated using steps designed at confirming that the pre-defined criteria of “maintenance RRT” were satisfied.

This was necessary as clinical coding manuals have not reliably distinguished between dialysis that was delivered in the setting of severe acute kidney injury and regular dialysis provided as part of maintenance treatment for ESRD. It would have been wrong to assume that the co-existence of clinical discharge codes for CKD and dialysis represented maintenance dialysis as CKD is itself a strong risk factor for AKI.¹⁰¹ A series of hierarchical rules designed to confirm whether maintenance RRT had occurred were therefore employed.

Rules for defining ‘maintenance RRT’ in all-England HES

Maintenance RRT Rule 1, Kidney transplantation:

The occurrence of any code included in RRT type = ‘Prevalent kidney transplantation’ or ‘Incident kidney transplantation’ (Table 2-1).

Explanatory note: Transplantation is only ever performed in patients with ESRD.

This was the first rule satisfied in 21,485 patients.

Maintenance RRT Rule 2, Maintenance peritoneal dialysis:

The occurrence of a RRT code = Peritoneal Dialysis, or survival of at least 90 days from the insertion of a peritoneal dialysis catheter (Table 2-1).

Exception to maintenance RRT Rule 2: Those who fulfilled rule 2 (peritoneal dialysis) as their first RRT event, but did not subsequently fulfil any of the other rules and had a diagnosis of AKI associated with all their PD spells were not considered a maintenance RRT patient.

Explanatory note: Peritoneal dialysis is rarely used to treat AKI and the insertion of a peritoneal dialysis catheter, in the absence of term for AKI, was considered maintenance RRT.

This was the first rule satisfied in 21,384 patients.

Maintenance RRT Rule 3, Definite maintenance dialysis:

The occurrence of a hospital inpatient episode with any code which included in RRT type ‘Dialysis’ in a participant who had:

3.1) A diagnosis of ESRD (Table 2-1) any time prior to, or within 365 days after the start of the episode

Or

3.2) The insertion of an arterio-venous (AV) fistula or graft (Table 2-1) any time prior to, or within 365 days after the start of the episode

Explanatory note: Patients who were identified as having a code for dialysis in the context of a prior mention of end-stage renal disease or evidence of permanent haemodialysis access creation were considered to have commenced maintenance RRT. Tunnelled central venous catheters were deliberately not included in rule 3.2 as these can be inserted for a variety of other reasons whereas an AV fistula or graft is almost only used for long-term dialysis. Similarly, including patients with only a record of permanent haemodialysis access (i.e. a arteriovenous fistula or graft) but without a record of RRT was not considered appropriate as these conduits are inserted upto 12 months before patients are anticipated to start maintenance and may never be subsequently used.

This was the first rule satisfied in 46,895 patients.

Maintenance RRT Rule 4, Probable maintenance dialysis:

The occurrence of at least two episodes with any code included in RRT type =

‘Dialysis’, with at least 90 days between the start of the first ‘Dialysis’ episode and the start of any subsequent spell containing a ‘Dialysis’ code that did not have a record of a acute renal disease diagnosis in that hospital spell (Table 2-1).

Explanatory note: Those who fulfilled rule 4 (probable dialysis) as their first RRT event and did not subsequently go on to fulfil any of the other rules should be considered ‘Possible dialysis’.

This was the first rule satisfied in 5,792 patients.

This series of rules was applied to the 139,018 patients of which 43,462 did not fulfil any rule and were therefore excluded, leaving a total of 95,556 patients who were confirmed to be receiving maintenance RRT.

2.7.3 3^rd HES Step: Distinguishing incident from prevalent RRT patients The cohort was then restricted to incident RRT patients by excluding those who were identified as having as prevalent transplant (n=14,420). The cohort was then also restricted to patients starting maintenance RRT between 2000 and 2008, removing a further 31,049 patients, leaving 42,730 patients. A flowchart of the derivation steps for the HES ESRD cohort is provided in Figure 2-4.

Figure 2-4: Treated end-stage renal disease cohort derivation (all-England HES 2000-2008)

2.7.4 Exclusion criteria applied to All-England HES derivation, 2000-2008 The cohort was then restricted to adults (≥ 18 years old) only.

In less than 5% of patients, missing demographic data at the defined start date of the cohort prevented their inclusion.

Incident RRT patients who died within 90 days of starting maintenance RRT were excluded (see Figure 2-4).

2.7.5 Defining entry date to the cohort

For each patient that met the criteria that defined maintenance RRT, the date of first maintenance RRT was defined as the earliest date of:

 The date of first incident transplantation code

 The start date of the episode for first record of peritoneal dialysis, or the date of insertion of a peritoneal catheter, when not in the context of acute kidney injury

 The start date of the first episode of dialysis that was used to define maintenance dialysis