The Effect of Payment By Results on the HRG Creep in English Hospitals: An Empirical Investigation

Abstract

Rationale: Healthcare resource groups (HRGs) is UK version of diagnosis related groups (DRGs) and has been used for the development of hospital prospective payment systems. A new policy of Payment by Results (PbR) was introduced into the English NHS in 2003. There is evidence from other countries that unintended effects have occurred when these systems are implemented, such as a reduction in the quality of care or gaming.

Objectives: This study investigates whether hospitals in the English NHS are altering their HRG coding behaviour, specifically with respect to HRGs that differentiate a procedure as being with or without complications. An incentive exists to code patients as complicated, as these HRGs tend to attract a higher tariff than the without complications counterpart. This study also examines if Foundation Trust hospitals have different behaviours than most hospitals (non- Foundation Trusts) in coding patients.

Data: We used hospital level data for the period 2001 to 2005 to construct a panel dataset from the Hospital Episode Statistics (HES) data. HES data provide details on hospital admissions within England on an annual basis.

Methodology: We analysed 4 year panel data set using a difference-in-differences analysis, which was safely utilized in this case because the new policy of Payment by Results were phased in; In 2003/04, tariffs were used for all hospitals for changes in activity in 15 Healthcare Resource Groups (HRGs) from a possible 550. In 2004/05, for most hospitals (non-Foundation Hospital Trusts - NFTs) the system was extended to 48 HRGS, whilst a self-selected group of Foundation Hospital Trusts (FTs) moved to predominantly tariff-based funding. Our models compared pairs of HRGs that were paid at tariff to a control group of pairs of high volume non-tariffed HRGs and compared HRGs between Foundation Trusts and non-Foundation Trusts.

Results: There was an evidence of HRG creep following the introduction of PbR. The number of spells for complicated tariffed HRGs increased and there was a corresponding reduction for this group in the average length of stay. The percentage of complications within pairs of tariffed HRGs also increased. In addition, how Foundation Trusts behave following the introduction of PbR was investigated. The results find no evidence to suggest that FTs behave any differently than non-FTs over this time period.

Conclusions: The new reimbursement policy for English hospitals is based on a prospective case mix payment system, using cost per case funding. Results indicate PbR is affecting coding behaviour of hospitals, but this is not necessarily due to unintended effects such as gaming, but may be due to more accurate coding procedures. To test if this is the case, further analysis looking for an increase in the number of secondary diagnoses would be required.

Keywords: HRG creep, hospital behaviour, prospective payment system