NHS database of reference costs is severely flawed

Editor—The recent and growing emphasis on gaining a more rounded view of performance assessments in health and social care is to be welcomed. Assessments of efficiency (one of the dimensions of The NHS Performance Assessment Framework¹) should take into account aspects of both quality and cost. Leaving aside the debate about how to assure and improve quality over time—the debate over clinical governance—there are sufficient difficulties in using routinely available cost information.

The reference costs database is one of the key sources of information. It is based on healthcare resource groups, which are the result of a systematic classification of acute care interventions into categories that are clinically distinct and have similar implications for resources. The intention is to provide a standardised method for comparing costs across hospitals: the more effective the standardisation process, the more likely it is that differences in costs reflect true differences in resource use. In theory, the reference costs provide information that permits management to make judgments about trusts' financial performance, or efficiency.²

Although, as with most of the centrally produced performance indicators, the publication of such information is surrounded by caveats about interpretation and an emphasis on statistical measures that take account of outliers, concern has to be expressed about the quality of data when the minimums and maximums for key healthcare resource groups are examined. The table has been constructed from the Reference Costs 2000 database,² and it shows the minimum and maximum costs quoted for healthcare resource groups that account for ⩾25 000 finished consultant episodes (for elective inpatients only).

Even a cursory examination of the table highlights the ridiculous nature of the data that have been submitted by trusts (presumably after internal validation). For example, one trust has quoted a cost per healthcare resource group for a primary hip replacement of £480 ($672); doubtless patients were handed a prosthesis to take home to fit for themselves.

The pitfalls of performance indicators have been well documented.³ One is the inevitable tendency to focus on the average. It is pertinent to ask what those who commission healthcare services are doing with data that lie at the extremes. If the data were taken at face value, the total cost of these 363 796 procedures would vary 42-fold depending on whether the minimum or maximum values were used. If such data are subjected to serious inquiry by the NHS, then will the distribution of costs quoted per healthcare resource group have narrowed considerably by the time the next database of reference costs is produced?

Table.

Sample of reference costs (£) submitted to NHS, 2000²

HRG code	HRG label	No of finished consultant episodes	Minimum cost	Maximum cost
B02	Phacoemulsification with lens implant	31 869	98	3 495
C22	Nose procedures (category 3)*	29 831	53	4 129
C24	Mouth or throat procedures (category 3)†	83 180	71	11 221
F74	Inguinal, umbilical, or femoral hernia repairs without complications for patients <70 years	26 006	112	4 524
H02	Primary hip replacement	33 625	480	9 337
H04	Primary knee replacement	28 808	695	12 921
H10	Arthroscopy	29 487	167	6 650
M06	Upper genital tract (intermediate procedures)‡	39 309	164	10 731
M07	Upper genital tract (major procedures)§	61 681	142	5 531

HRG=healthcare resource group. ^*Category 3 procedures include septoplasty, nasal polypectomy, and submucous resection of nasal septum. ^†Category 3 procedures include tonsillectomy and adenoidectomy. ^‡Intermediate procedures include culdotomy, drainage or aspiration of pouch of Douglas, endoscopic procedures on uterus and fallopian tubes, and ligation and clipping of fallopian tubes. ^§Major procedures include abdominal and vaginal hysterectomy and excision of the uterus.