HRG design concepts

Casemix classification design is governed by the Casemix Design Framework, which provides comprehensive guidance for stakeholders involved in the design process regarding scope, format, data and Healthcare Resource Group (HRG) performance requirements.

The Casemix Design Framework outlines the key principles that are essential to any casemix classification and has evolved with each iteration of our classification. The Framework includes details of the Casemix design fundamentals and editorial rules used when developing our HRG designs, as well as how we measure the performance of our HRGs.

Our stakeholders are comprised of representatives from royal colleges, clinical professions and associations, Policy colleagues from NHS England and Improvement, and professional bodies within the independent sector.

In brief, the design rules stipulate that:

• HRGs must be clinically meaningful and contain activity with similar expected resource intensity. This not only ensures that HRGs provide a valuable dialogue mechanism between clinical and finance professionals, but that average costs or national tariffs at the HRG level do not systematically under- or over-represent the resource use of the care provided when treating particular groups of patients
• Data used to define HRGs should be routinely available to minimise the burden of data collection on the NHS
• There should be a manageable number of HRGs to cover all patients, ensuring that the administrative burden of processing and evaluating HRG-level data in terms of costing and reimbursement is kept to a minimum

HRGs are identified by a five-character code structure:

The first four characters together are classed as the HRG root.

Every OPCS-4 procedure code and ICD-10 diagnosis code is mapped to an HRG root within the HRG4+ design, with the exception of procedure codes that are ignored for grouping. The base HRG root to which a code is mapped, when all other logic conditions for the code have not been met, can be found in the HRG1 field in the Code to Group tab of the Code to Group Excel workbook that accompanies each Grouper release.

General principles for the HRG design are that:

• HRGs are divided into clinically meaningful sections (chapters and subchapters)
• the lower the HRG number, the higher the expected resource use of that HRG in relation to other HRGs within the subchapter (though this may not be the case where more-resource intensive HRGs have had to be 'slotted in' to an existing subchapter structure)
• the final character split within the HRG code structure is a single character code that further describes activity, such as patient age, length of stay or the presence of complications/comorbidities in the patient record - split characters do not however have to be standardised across the HRG design
• a value of 'Z' as the last character indicates that no split is present

The Code to Group Excel workbook provides details of constituent elements that contribute to HRG grouping. It contains reference data, such as the ICD-10 and OPCS-4 codes used in the design, procedure and diagnosis hierarchies relevant to a specific design, and the Complication and Comorbidities lists for HRG subchapters. It also includes a list of chapters and subchapters relevant to an individual Grouper product, as well as a comprehensive list of HRG codes and labels.

As the HRG design necessarily changes over time, users must ensure they are using the Code to Group Excel workbook specific to the Grouper software being used. These must match in both purpose (Costing/Payment) and financial year.

Setting independence means that if a procedure can be performed across different care settings, the same HRG can be derived regardless of setting. For example, an endoscopy would generate the same HRG regardless of whether it was performed as an outpatient, day case or inpatient procedure. It is important to understand that setting independence applies to procedure-driven HRGs only. It does not apply to diagnosis-driven HRGs, nor to HRGs that are derived from data items other than the procedure (OPCS-4) or diagnosis (ICD-10) primary classifications.

Non-admitted consultation (Outpatient) HRGs are derived where no significant procedure code is recorded in the patient record, or where the only code recorded is from OPCS-4 code category X62.- Assessment. For outpatient data, HRG derivation is not dependent on diagnosis as these data are not yet mandated in the Outpatient Commissioning Data Set.

In certain settings, for example outpatient clinics, it is possible that a procedure may not be carried out, or one may not always be recorded, meaning that a procedure-driven HRG cannot be generated. In these situations, assuming minimum mandatory information has been recorded, one of the default non-admitted Uni-disciplinary HRGs within Subchapter WF Non-Admitted Consultations will be assigned. For further information, please refer to the chapter summary for Subchapter WF. Chapter summaries are available for every HRG subchapter and provide an overview of the HRGs within that subchapter, details of changes made from previous Grouper releases and a brief description of the design concepts utilised in the development of the HRGs in the subchapter.

Each procedure is assigned a hierarchical value associated with its expected resource consequences. These hierarchical rankings are intended to reflect the expected relative costs of individual procedures and provide a mechanism by which the relative complexity of procedures can be compared across HRG chapters.

The range of procedure hierarchy (PH) values was expanded as part of the move to HRG4+ to enable the design to more appropriately reflect the expected resource use of procedures across all subchapters, particularly when differentiating between low-cost high-volume procedures. As part of this change, each OPCS-4 procedure code valid for driving grouping was reassigned a PH value. The logarithmic hierarchy range runs from 3 to 41, with a lower resource difference expected between the values at the lower end of the range than those at the higher end. PH values were also amended to eliminate overlap between HRG complexity categories.

If a single procedure code is recorded for a patient and its hierarchy value is equal to or greater than 3 (5 for admitted patient care), it will be used for grouping (subject to length of stay criteria for minor procedures).

If multiple procedures are recorded, the procedure code classifying the dominant procedure is identified using the procedure hierarchy. In the event of two (or more) procedures being recorded within a single patient record with the same procedure hierarchy value, the first of these procedures recorded in the patient record will drive HRG grouping at both the episode and spell level.

Event-based unbundled HRGs have a hierarchy value of 2 and are output based on each instance of an OPCS-4 code being recorded. In the absence of any procedures, or where the only procedure code recorded has a hierarchy value of 1 or 2, grouping will flip to using the primary diagnosis of the episode, or the derived primary diagnosis of the spell, to determine the HRG.

In summary, each procedure code has an associated value reflecting its relative expected resource use:

The following table provides additional examples of the types of OPCS-4 codes and their hierarchies.

Each Admitted Patient Care Finished Consultant Episode (FCE) will have a primary diagnosis recorded, reflecting the primary reason for care and as determined by the clinical record for the patient.

Primary diagnosis is used to drive grouping when there are no significant procedures in the patient record suitable to drive grouping, or where procedure-driven grouping has effectively flipped to diagnosis-driven grouping as a result of exceeding maximum length of stay criteria for the dominant procedure.

Each diagnosis that is valid as a primary diagnosis in a patient record is assigned a hierarchical value associated with its expected resource consequences. These hierarchical rankings reflect the expected relative cost of admissions for each primary diagnosis.

The range of diagnosis hierarchy (DH) values was expanded as part of the move to HRG4+ to enable the design to better reflect the expected resource use associated with specific diagnoses across all subchapters. As part of this change, each ICD-10 code valid for driving grouping was reassigned a DH value. DH values are used to determine the primary diagnosis of a multi-episode spell with multiple different primary diagnoses across the episodes. The logarithmic DH range runs from 5 to 25, with a lower resource difference expected between the values at the lower end of the range than those at the higher end. This change also provided improved foundations on which to implement Interactive CC logics.

DH values are not used to determine diagnosis-driven FCE HRGs, as the primary diagnosis for an FCE is determined by the admitting clinician.

Where a patient has more than one primary diagnosis in a spell, because that patient spell contains more than a single FCE (as a result of a transfer of consultant responsibility), and the primary diagnoses of the FCEs within that spell differ, it is necessary to determine the primary diagnosis of the spell before the spell activity can derive an appropriate spell HRG.

For HRG grouping purposes, the primary diagnosis of a spell is therefore deemed to be:

The primary diagnosis of the episode containing the dominant procedure (the latter as determined by the PH value), irrespective of whether that dominant procedure has a maximum length of stay check that results in the record effectively flipping to group off the primary diagnosis of that episode

Or, where no dominant significant procedure exists within the patient record, the primary diagnosis spell is deemed to be:

The primary diagnosis with the first highest DH value in the patient record

In summary, within the DH there are 21 bands running from 5 to 25, where 5 represents the least resource-intensive primary diagnosis and 25 represents the most resource intensive primary diagnosis.

The following table provides additional examples of the types of ICD-10 codes and their hierarchies.

Complication and comorbidity (CC) splits are a way of incorporating and recognising varying levels of patient severity and complexity within the HRG design.

Dual-coded diagnoses often provide a way of describing the severity of a condition and are a principle used in disease staging. CC splits are used in particular in the diagnosis-driven HRGs as a way of indicating varying illness severity for patients with the same primary diagnosis.

The coding of multiple morbidities and complications describes one aspect of patient complexity. The ICD-10 diagnosis coding classification also includes a number of social factors and proxies that may help to describe the wider health needs of a patient. These may also reflect additional resource usage and will be on CC lists where clinically appropriate.

The majority of HRGs employing CC splits rely on a subchapter-specific CC list to separate activity. The purpose of each CC list is to identify unique secondary diagnoses that are expected to result in additional resources being used by patients.

It is important to note that a particular secondary diagnosis may be a major complication for some procedures or conditions while not being a relevant complication for others. The relevance and ranking of CCs are assessed at subchapter level by individual EWGs to ensure that the CCs are appropriately acknowledged. For secondary diagnoses to be recognised in HRG derivation terms, they must therefore be both unique, and clinically relevant.

There are specific exceptions to the use of a CC list to determine a CC value. For example, where a patient’s primary diagnosis has an inherent CC explicitly stated in the ICD-10 code (such as K43.1 Incisional hernia with gangrene) or where the presence of multiple secondary cancers and infections are used as a proxy CC score, such as when generating the HRG root PM45 Paediatric Febrile Neutropenia with Malignancy.

Standard CC splits were replaced with interactive CC splits in HRG4+. Interactive CC splits rely on summed scores and more appropriately reflect the expected additional resource use of treating patients with multiple complications and/or comorbidities.

To understand how the Grouper determines CC scores, please see section 5 of this document.

This grouping mechanism has been defined to identify high resource, complex treatments associated with admissions for multiple trauma cases - simultaneous traumatic injuries involving more than one body site for example. These injuries are coded in accordance with ICD-10 Chapter XIX, Injury, poisoning and certain other consequences of external causes (S00 – T98).

Body sites have been defined, and a table containing non-superficial trauma injuries relating to each specific body site has been compiled. List of these injuries can be found in the Comp_VA (such as Comp_VA_Upper) lists in the Other Lists worksheet of the Code to Group Excel workbook).

The body sites are:

• Abdominal
• Chest
• Head
• Kidney
• Lower Limb
• Upper Limb
• Pelvis or Spine
• Urinary
• Other

If a patient is recorded as requiring treatment for traumatic injuries to two or more different body sites (and one of these is the primary diagnosis), a multiple trauma HRG will be generated for that episode of care. Multiple Trauma is a separate concept to Major Trauma: while Major Trauma may involve a single body site, a minimum of two different body sites is required for Multiple Trauma HRG derivation.

Once a patient is determined to be a Multiple Trauma patient in HRG design terms, the concepts of primary diagnosis and dominant procedure are no longer relevant. The HRG design effectively acknowledges all distinct diagnoses and all procedures as being relevant to the resource impact of the healthcare provided, and HRGs are assigned via a matrix scoring system that reflects the breadth of what is clinically wrong with the patient and the range of procedures undertaken on that patient.

A multiple trauma HRG will be generated for a spell where the HRG of the first episode of a multi-episode spell is a multiple trauma HRG. For multi-episode spells where the first episode is not assigned a multiple trauma HRG but a later episode is, the spell HRG will not be a multiple trauma HRG.

Intervention splits are included for a number of diagnosis-driven HRGs in various subchapters. This split acknowledges that 'minor interventions' have been undertaken during a patient admission. The benefit of this approach is twofold: these HRGs will not only include the additional cost/resources associated with performing these relatively minor procedures, but they may also provide an indication that the patient’s condition was more severe, often resulting in more resource-intensive treatment.

The design includes 'with Multiple Interventions' and 'with Single Intervention' HRGs to more appropriately capture the additional resource use of patients who have multiple minor interventions during their episode or spell.

HRG4+ introduced HRGs specific to specialised activity, such as those for congenital cardiac surgery. HRG4+ also expanded the concept of age splits by introducing paediatric age splits that enable the creation of HRGs specific to a given subset of patients within the child population. Paediatric age splits can be employed to separate activity where there is a significant difference in expected resource use, for example between treating infants and treating older children.

A significant number of HRGs continue to have a Paediatric (18 years and under)/Adult (19 years and over) age split to recognise the significant resource difference that can occur when treating children rather than adults, where greater subdivision within the child population is not clinically relevant.

The majority of minor procedure HRGs across all subchapters have maximum length of stay checks. Where the length of stay is longer than the set maximum, the primary diagnosis will be used to derive the HRG rather than the minor procedure. This approach is intended to ensure that HRG grouping accurately reflects the primary reason for the patients’ admission. It reduces the likelihood that procedure-driven HRGs will be derived for patients with long lengths of stay undergoing a relatively minor procedure during that admission, where the length of stay is more reflective of the treatment for their condition.

As previously mentioned, however, these relatively minor procedures may themselves be acknowledged as interventions for a number of diagnosis-driven HRGs whose grouping has effectively flipped from procedure-driven to diagnosis-driven as a result of exceeding maximum length of stay criteria for the procedure.

To improve the performance of HRGs and to better represent activity and costs, some significant elements of cost and activity are identified separately, that is they are 'unbundled' from the core HRGs that reflect the primary reason for a patient admission or treatment. These unbundled HRGs therefore better describe the elements of care that comprise the patient pathway within a hospital admission or outpatient attendance.

In previous HRG designs (that is, up to HRG v3.5), each episode of care would derive a single HRG. However, from HRG4 onwards, some significant elements of cost and activity were 'unbundled' from core HRGs. The impact of this is that a single patient record is assigned more than one HRG if it includes any 'unbundled' elements. The 'unbundled component' becomes an HRG in its own right and is generated in addition to a core HRG for the episode or spell of care, or attendance.

An unbundled HRG may be event-based, and thus derived from the presence of a specific OPCS-4 or ICD-10 code in the patient record, or duration-based, in which case it is generated on a per diem basis.

Unbundled HRGs have been developed for:

• Chemotherapy – Regimen Procurement and Delivery
• Radiotherapy – Planning and Treatment
• Diagnostic Imaging and Nuclear Medicine (such as MRI/CT/SPECT-CT)
• Rehabilitation
• Renal Dialysis for Acute Kidney Injury
• Critical Care – Adult, Paediatric and Neonatal (derived from the Critical Care Minimum Data Sets)
• Specialist Palliative Care
• High Cost Drugs