Explaining variation in hospital admission rates between general practices: cross sectional studyBMJ 1999; 319 doi: https://doi.org/10.1136/bmj.319.7202.98 (Published 10 July 1999) Cite this as: BMJ 1999;319:98
All rapid responses
Rapid responses are electronic comments to the editor. They enable our users to debate issues raised in articles published on bmj.com. A rapid response is first posted online. If you need the URL (web address) of an individual response, simply click on the response headline and copy the URL from the browser window. A proportion of responses will, after editing, be published online and in the print journal as letters, which are indexed in PubMed. Rapid responses are not indexed in PubMed and they are not journal articles. The BMJ reserves the right to remove responses which are being wilfully misrepresented as published articles or when it is brought to our attention that a response spreads misinformation.
From March 2022, the word limit for rapid responses will be 600 words not including references and author details. We will no longer post responses that exceed this limit.
The word limit for letters selected from posted responses remains 300 words.
Fiona Reid et al’s observation that variation in hospital admission
rates between general practices is related to patients’ socio-economic
factors rather than practice characteristics comes as little surprise
since the study was conducted across an entire health authority with
considerable social diversity.
Likewise the lack of correlation between proxies of quality in
practices – cervical screening uptake rates etc – and hospital admissions
is easily explicable in that the former directly enhanced the doctors
earnings while referral and admission rates reflect in part the burden of
clinical load primary care doctors are willing to accept.
With the advent of Primary Care Groups evidence is emerging that
individual practices within a locality may vary greatly in both the
baseline budget they enjoy for practice staff and the support they receive
from community based staff. Such variation probably reflects the historic
enterprise of individual practices in maximising opportunities to improve
patient care both through General Medical Services monies and fund-holding
It follows, therefore, that to evaluate performance indicators it
might prove more productive to study populations with uniform socio-
economic circumstances and the practices that serve them. Once the
population is standardised then it should prove possible to examine
suitably selected practice characteristics, eg baseline staff budgets and
community staff, in relation to referral rates and admission rates. For
the purposes of resource allocation within individual PCGs this work needs
to be urgently undertaken.
Competing interests: No competing interests
Rates of general practitioner referrals to hospital outpatients have
been extensively studied, but emergency referrals, which are potentially
far more disruptive to patients, as well as being very expensive, have
Reid et al in their study published 10 July (1) address this issue
indirectly by studying hospital admissions in a large well designed study.
They acknowledge, and the accompanying editorial (2) also highlights, that
admission rates do not necessarily reflect referral rates.
This is not just a theoretical point. In a pilot prospective study of 760
emergency referrals to Northwick Park Hospital, London, in 1993-4 there
was a very wide variation in referral patterns. The highest referring
practice, which referred over 4 times the median number, had under a
quarter of all referrals admitted, the rest being sent home after
assessment. The average practice had two-thirds of their referrals
admitted. This does not necessarily mean that any referrals were
inappropriate, but if emergency admission rates had been studied the
difference in referral rates would have appeared much smaller than it was.
In this small sample 20% of all emergency referrals made to the hospital
were made by 10 GPs, out of over 200 GPs in the hospital catchment area.
Trying to determine emergency referral rates accurately is fraught with
methodological problems, but wide differences in admissions seen in
studies of emergency admissions may well be a significant underestimate of
the differences in true rates of referral.
Christopher JM Whitty
Department of Medicine,
College of Medicine,
University of Malawi,
Private Bag 360, Chichiri,
1) Reid FDA, Cook DG, Majeed A. Explaining variation in hospital
admission rates between general practices: cross sectional study. BMJ
1999; 319: 98-103
2) Jankowski R. What do hospital admission rates say about primary
care? BMJ 1999; 319: 67-8.
Competing interests: No competing interests
Editor - With reference to the articles referring to the use of HES (Hospital
Episode Statistic) data in this week’s BMJ1,2, we would like to inform you
of some of the difficulties we have encountered in the use of this data in
We have recently undertaken a study comparing rates of hospital
admission for elective surgery for residents of this district with rates
for residents of similar districts. This study used HES data grouped into
HRGs (healthcare resource groups) for contracting purposes .
This study attempted to identify procedures where rates of surgery
for our residents were higher than those of the comparators. Amongst the
procedures with higher rates, there were four procedures where the
differences in rates were so large that we suspected that this might be
due to discrepancies in the coding of procedures between trusts. Further
enquiries revealed that our main provider was coding these procedures as a
day case, whereas the vast majority of other trusts, including our other
major local provider, appeared to be coding them as an outpatient
The numbers of procedures identified is as follows:
Procedure Number HRG Cost per day case of FCEs FCE from main provider Injection of haemorrhoids 151 F95 £390 Prostatic biopsy 285 L30 £243 Retinal coagulation 152 B05 £504 Hysteroscopy 726 M06 £455
Thus it can be seen that these day case procedures cost the health
authority a total of £535,083.
Discussions with this provider about the contracting arrangements
revealed that some of these reclassifications had arisen for historical
reasons where it had been advantageous to show a higher number of day
cases, and that financially the charges for these procedures had been
adjusted to ensure that these procedures would cost no more than they
would had they been classified as an outpatient procedure. In subsequent
years, however, this distinction was dropped and the full cost was
We have found no other examples in the literature of such problems in
Britain, although in the USA, incorrect coding of DRGs (diagnosis related
groups, the US equivalent) in the favour of health care providers, in
their case for financial gain, is a recognised phenomenon. This has become
known as ‘DRG creep’.
Hsia, Krushat et al3 showed that from October 1984 to March 1985
there was an error rate of 20.8% in DRG coding. Unlike previous studies
where errors had occurred randomly , this study showed that a
statistically significant 61 percent of errors were in favour of
hospitals, causing a 1.9% increase in their casemix (an index of
complexity of the cases treated), leading to an increase in hospital
reimbursement from Medicare of 308 million dollars. However, a follow up
study by Hsia et al4, looking at accuracy of data coding between 1985 and
1988, showed that coding errors had declined to 14%, and that only 50.7%
of these errors (a statistically non-significant figure) were in favour of
Goldfarb and Coffey5 analysed changes in casemix during the early
years of the Medicare PPS (prospective payment system) using data from
1980-86. They used statistical techniques, including regression analysis,
in order to assess how much of the increase in casemix was as a result of
the change in payment system, and how much was due to underlying changes
in medical practice. They estimated that half of the change was due to
each, but that the effect of coding was starting to decline by 1986. This
would concur with the findings in the second paper of Hsia4. This paper
cites the influence of stringent Medicare peer review organisations as the
main influence in reducing the rise in casemix.
A Medline search shows that little since then has been published in
relation to this issue. However, searching the Internet reveals that
careful coding may still be alive and well. Using the ‘Alta Vista’ search
engine, the terms ‘DRGs’ and ‘coding accuracy’ revealed a number of
commercial companies in the USA offering services and software to improve
hospitals’ accuracy of coding with the explicit aim of increasing
hospital income. Conversely another company offered software to reduce the
expenditure of healthcare purchasers, also by examining coding.
We wonder if there are any other examples of major HRG coding
discrepancies in Britain, which favour the interests of NHS trusts,
whether these interests are administrative or financial, and whether our
findings could reflect altered behaviour of trusts in response to the
purchaser-provider arrangements of the changed NHS. In addition it is
clear that such coding discrepancies affect the use of this data for
comparisons of hospital admission rates.
We wonder whether there are any examples in this country of miscoding
for explicit financial reward, and whether ‘HRG creep’ could become a
recognised entity in the UK.
Dr F.J.A. Perlman, Project Facilitator
Dr C.J. Watts, Director of Public Health
Public Health Department
Barking and Havering Health Authority,
The Clock House,
Barking. IG11 8EY
1. Giuffrida A, Gravelle H, Roland M. Measuring quality of care with
routine data: avoiding confusion between performance indicators and health
outcomes. BMJ, 1999; 319:94-98
2. Reid FDA, Cook DG, Majeed A. Explaining variation in hospital admission
rates between hospital practices: cross sectional study. BMJ, 1999; 319:98
3. Hsia DC, Krushat WM et al. Accuracy of diagnostic coding for Medicare
patients under the prospective payment system. N Eng J Med, 1998; 318(6):
4. Hsia DC, Ahern CA et al. Medicare reimbursement accuracy under the
prospective payment system, 1985-1988. JAMA, 1992; 268(7): 896-9
5. Goldfarb MG: Coffey RM. Change in the Medicare case-mix index in the
1980s and the effect of the prospective payment system. Health Services
Research, 1992; 27(3): 385-415
Competing interests: Procedure Number HRG Cost per day case of FCEs FCE from main providerInjection of haemorrhoids 151 F95 £390Prostatic biopsy 285 L30 £243Retinal coagulation 152 B05 £504Hysteroscopy 726 M06 £455