A critical theme in South Africa’s National Health Insurance initiative is changing the way that hospital services are financed. Switching to Diagnosis Related Groups (DRG) was part of the original Green Paper in 2011, and remains firmly in place in the updated version released in December 2015. It’s not a simple switch, and needs several building blocks to work effectively.
It’s also a proven financing methodology. Many countries use DRGs, and have developed their bespoke versions. Originally designed as a quality assurance tool in the USA, DRGs were used by the Medicare Program to reimburse hospitals using prospective prices from 1983. It had 470 DRGs across 23 Major Diagnostic Categories (MDC), each of which can include both surgical and medical services. As combinations of WHO’s International Classification of Diseases (ICD), currently ICD-10, they initially didn’t reflect the severity or complexity of the conditions, and adjustments for these came later. There are now over 750 DRGs in use for 26 MDCs, depending on the version and country. Some DRG prices distinguish day case from inpatients.
About half the DRGs are medical, 47% for surgery in MDCs, plus 2% for pre-surgery and 1% for surgical cases not in MDCs and two DRGs for an invalid principal diagnosis and ungroupable workload. About 3% of DRGs aren’t in an MDC. It’s roughly a 50:50 split between medical and surgical cases.
Prospective DRG prices are seen as a way to control healthcare costs. It requires hospitals to manage their unit costs and annual expenditure within these. So why’s it not that simple? DRGs don’t extend across all hospital activity and need time to catch up with new medical techniques. These can either have their own reimbursement rates, or take so long to set that they can be a brake on investment, a claim for a slow take up of telemedicine. A recent example’s the proposal in 2015 from the American Association of Hip and Knee Surgeons (AAHKS) to the Centers for Medicare & Medicaid Services (CMS) to modify or establish a new MS-DRG for total hip arthroplasty (THA) cases involving patients with hip fracture. For 2016, CMS has released DRG version 33, one a year since 1983, including eight changes to existing Medicare Severity (MS) DRGs.
Since 1983, DRGs have almost gone global. As a proven reimbursement method, many countries have refined them to match their health systems. The UK and Ireland have Healthcare Related Groups (HRG). The Nordic countries have their NordDRG. These reflect health systems’ bespoke characteristics. This includes their different healthcare models and their specific cost structures.
Setting prices for DRGs often starts with existing unit costs. This needs a costing system with two main methodologies. One’s Total Absorption Costing (TAC) to allocate and apportion expenditure to unit costs. The others a variable and semi-variable costing methodology to identify expenditure changes arising from changes in workloads. Semi-variable costs are the most challenging to compile. Both methodologies rely on sound workload data and ICD 10 coding.
Using TAC throws up a specific challenge for healthcare. Direct costs that can be allocated to specific patients are a small proportion of total costs, maybe less than 10%. This creates the risk of large skewed unit costs derived mainly from using formulae for apportionments. One way to minimise these inconsistencies it to create cost pools for the MDCs or specialties where direct costs can be increased significantly. Each MDC can have its own apportionments on to workloads. This minimises skewed results, but doesn’t avoid them, so comparisons between hospitals’ will still reveal some odd numbers.
For South Africa’s switch to DRGs, some important activities are needed that apply lessons learned from other health systems. They include:
- Test the accuracy and completeness of hospital’s workload data, including duplication challenges as reported in eHNA, and minimising the workload assigned to DRGs for invalid principal diagnoses as discharge diagnoses and ungroupable workload
- Test hospitals’ ICD-10 readiness to avoid the USA’s unhappy experiences reported in eHNA
- Design the DRG grouper needed for reimbursement, including DRG choices for patients with more than one DRG for their hospital stay
- Set up reimbursement models for hospital services that don’t fit the DRG model
- Design and trial a national hospital costing methodology that reconciles workloads x unit costs to total expenditure
- Refine the costing methodology
- Run a parallel DRG financing model alongside the existing model to identify large swings
- Design an interim DRG financing model to minimise financial risks
- Estimate and monitor changes to hospital costs and the impact on total health service spending and finance.
With the NHI’s fourteen-year development timescale, now’s the time to start these. An early start provides time to iron out data idiosyncrasies and will help to minimise risks, which are always prevalent when financing models change. eHNA’s looking forward to reviewing these soon.