 
|
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Journal of the New Zealand Medical Association,
21-February-2003, Vol 116 No 1169
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
General practitioners’ assessments of the primary care
caseload in Middlemore Hospital Emergency Department
Barry Gribben
Over the past three years (1998–2000), patient
attendances at Middlemore Hospital (MMH) emergency department have been
increasing at 8.5% per annum. The growth in self-referrals is 12.5% per annum.
Over this same period, there has been only a 2% annual increase in population of
the Counties Manukau District Health Board (CMDHB) area. This disproportionate
growth in emergency department (ED) attendances reflects an international
phenomenon that has been observed in the UK, US and
Australia,1,2,3 although its meaning and
interpretation have been debated.4 This growth
has significant resource implications for CMDHB, and for this reason, as part of
a wider study into ED attendance patterns,5
CMDHB commissioned an investigation into the proportion of the ED caseload that
could potentially be handled in primary healthcare. This paper describes the
results of a retrospective review of 300 random ED discharge summaries conducted
by a panel of 12 GPs, designed to estimate this proportion.
MethodsSince January 2001, MMH has
produced electronic discharge summaries for all ED patients who are not
subsequently admitted to a hospital ward, regardless of original presentation.
Discharge summaries are produced for self-referred patients, GP referrals, and
ambulance cases. Patients who are kept in ED for short-stay observation also
receive a discharge summary. These discharge summaries are then faxed or emailed
to the patient’s GP, if known. In total, 61% of all attendances at ED over
the study period had such a discharge summary, the remaining 39% being admitted
to a ward at MMH.
We randomly selected 300 anonymous cases from all the ED discharge summaries produced between 1 January 2001 and 30 June 2001. We chose 300 cases to provide a small enough confidence interval around the estimated proportions to justify any management decisions (worst case 50% +/- 5%, α = 0.05, β = 0.2). Every 69th case of the 20 961 cases seen but not admitted over this period was selected. The cases selected were representative of the age, gender, ethnicity and time-of-attendance of all cases. We constructed a report of each attendance, which included age, gender, day of the week, time of attendance, and the “clinical management” data, in which the presenting complaint was described. No identifying data were supplied. In 26 cases there were no data in the “clinical management” field. We asked 12 GPs to assess these cases. Ten of the GPs came from a range of locations within CMDHB (Papakura, Takanini, Manurewa, Papatoetoe, Otahuhu, Mangere and Howick). The other two were lecturers in General Practice from the Department of General Practice and Primary Health Care. Two of the ten CMDHB GPs worked in after-hours clinics. There were four female GPs in the group. The assessment of the cases took place on a Saturday morning, with a meal break. GPs were paid for their time. GPs were provided with sheets that described each case. All assessments were completed by each GP independently. The sets of cases were presented in different order to each GP. The cases were assessed using a form that attempted to capture some of the uncertainty involved in performing these assessments. For each case, each GP was asked to choose from a number of options:
Table 1 provides an example of the
coding form. It should be noted that options 3 and 4 (above) were not mutually
exclusive; GPs could tick as many boxes as applied.
Given the data available to the GPs, we tried to set an accurate lower bound on the proportion of cases that they thought could have been handled in primary care by:
ResultsThe 12 GPs considered that an
average of 56% of the cases presented could be managed in primary care. The
range was large, however, as shown Figure 1.
In addition, the distribution of these percentages is not
normal, suggesting real variation in the ability or willingness of GPs to handle
ED cases in primary care. In 50 of the 274 cases that were assessed, all 12
doctors agreed that the specific case could be handled in primary care. The
distribution of the scoring of cases by the 12 doctors across each possible
classification on the scoring form is shown in Table 1. The agreement between
GPs was “fair” with kappa = 0.34, when coding with three categories:
1 = surgery, 3 = referral, 2 = any other response. As these data are ordinal, a
Kendall’s W (coefficient of concordance) was also calculated, which showed
(as expected) slightly greater agreement (W = 0.48).
Figure 1. Assessment of proportion of cases that could
be handled in primary healthcare (PHC), based upon assessment of discharge
summary
 GP=general practitioner, AM=Accident and Medical doctor, AC=academic general practitioner Table 1. GP scoring of ED cases (NB groups are not
exclusive, so percentages do not add to 100%)
u/s=ultrasound; other Ix=other investigations
To illustrate the types of cases that were considered, and
the data with which the panel were provided, the management of the first eight
cases assessed by the GPs is shown in Table 2. The numbers of GPs who indicated
that their action would depend upon further tests, and have indicated a
“possible” or “probable” admission, are not displayed in
the table.
The proportion of the total ED caseload that could be
handled by primary care may be estimated by assuming that no admitted cases
could have been handled in primary care, a very conservative assumption. Under
this assumption, we can estimate that GPs would assess that 34% of the total ED
case load (56% of 61%) could be handled in primary care. An alternative analysis
is to count only those cases for which all GPs agreed that primary care
treatment was possible. The range of assessments given by the GPs was, as
mentioned, very variable. The 12 GPs agreed unanimously that 10% (50/300 * 61%)
of ED cases could have been treated in primary care, and in only 1% (2/300 *
61%) of cases did all GPs agree that a case should definitely be referred to
ED.
*number of GPs who felt case could have been managed in
primary care; †number of GPs who would
have referred the case to hospital
DiscussionOther studies have found large
variation in the proportion of ED cases that are assessed by doctors as being
“inappropriate”, from 19% to
87%.6,7 These studies, and others on ED
attendance, have been described in a very useful literature overview prepared by
New Zealand Health Technology Assessment
(NZHTA).8 The variation in the figures is
influenced by the national context and the definition of
“inappropriate” that is employed. In most studies,
“inappropriate” is taken to mean that a patient could have been
treated in primary care.9 In the UK, a study
that calibrated existing tools for measuring “inappropriateness”
(based on ICD9 classification and process measurements) against opinions of a GP
panel, estimated that 23% of cases in 16 English A&E departments could have
been handled by a patient’s GP.10 A
review of 2980 ED cases from Elche hospital (Valencia, Spain), using an explicit
review instrument applied by ED staff, estimated 29.6% of cases were primary
care cases.11
Features of ED attendance have also been investigated in
NZ,12,13,14 in particular for asthma
treatment,15 and in South Auckland
specifically,16 but these studies have not
attempted to measure appropriateness of utilisation, and may now not reflect the
contemporary situation. A report commissioned by South Auckland Health in 1995
describes a review in which different clinicians were asked to assess 78 ED case
notes to determine if the cases could have been managed in primary
care.17 The emergency medicine specialist
assessed that 12% were primary care cases; a GP, 33%; and a private Accident and
Medical (A&M) doctor, 37%. The present study is remarkably consistent with
these estimates, given the large increase in ED volumes that has occurred in the
interim.
The present study may be criticised from a number of
perspectives. Deciding what information to provide GPs from the discharge
summary was not straightforward. We wanted to provide the closest possible
approximation to real-life general practice, to allow GPs to make the most
natural decision about patient management. For example, in real-life general
practice the clinician will know what medications a patient is taking. However,
the medications list from the discharge summary may also provide clues as to the
discharge diagnosis, which of course is not known when the patient is triaged.
In this study we decided not to provide GPs with the medication list.
We also considered providing full case notes to GPs, but
decided that the logistics of deleting extra information not normally available
when the patient is triaged from clinical records, copying the notes for
off-site use, and maintaining confidentiality, made this approach very costly
for no obvious methodological advantage.
We asked each doctor to indicate their view of the validity
of the assessment process, given the retrospective nature of the study and the
limited clinical data presented – for example, in no case was a
temperature recorded in the clinical notes field.
The doctors felt that although there were certainly cases
for which more information was required, this was a valid process, and generally
were very surprised at the number of cases that, according to them, were
obviously general practice cases. This does not prove content validity, but the
lack of any discomfort amongst the GPs about the process lends some additional
weight to the results. To a large extent, discomfort was avoided by allowing the
assessment of a case as a “possible” or “probable”
referral.
The doctors themselves were not a random sample – they
agreed to give up a Saturday morning to code discharge summaries! They were
chosen by the researcher for geographic, institutional and gender spread. It
would thus be incorrect to extrapolate from these results to CMDHB general
practitioners as a whole. However, we were not trying to estimate the views of
all CMDHB doctors – we were trying to estimate what proportion of cases
could be managed in primary care. The group of doctors was our expert panel.
Their mean rate is our best estimate of the actual proportion of these 300 cases
that could be managed in primary care.
There was a surprisingly wide range of views on the
proportion of cases that the GPs thought could be completely handled in primary
care. We asked for personal views, and so naturally the skills and experience of
individual GPs will have contributed to the range. These assessments did not
appear to be related to age or gender and the scores of the A&M doctors and
academic GPs were distributed across the range.
In 19% of cases, GPs would have liked access to services
they do not currently have available. However, in 14% of cases they would have
possibly or probably referred. The interpretation of these data is not
straightforward (for example, GPs might still have referred once results were
available), but a fair interpretation would be that an extra 5–10% of
cases could probably be managed by GPs with additional access to
diagnostics.
A further criticism of this study is that GPs would have a
vested interest, both financial and in terms of professional standing, to score
cases as “GP” cases. We started from the premise that GPs are the
most qualified group to determine which cases a GP could handle. A randomized
prospective trial, in which patients that were assessed by GPs as being primary
care patients were treated in ED or at a GP’s surgery, would be a more
robust design, but poses major methodological and possibly ethical
problems.
Finally, it should be remembered that the cases assessed by
GPs as being primary care cases, did in fact attend Middlemore Hospital ED. The
reasons for this need to be fully understood before a coherent approach to the
primary care caseload at MMH ED can be developed. The unique population served
by MMH ED means that these findings might not be applicable to other emergency
departments in NZ. In a qualitative research programme commissioned by
CMDHB,5 many GPs felt that the lack of a
financial barrier at the ED is the most important reason that many patients
choose to attend the ED for primary care. However, patient interview data
suggest that issues of accessibility, familiarity and confidence are also
important factors.
Author information:
Barry Gribben, Senior Research Fellow, Department of General Practice and
Primary Health Care, University of Auckland, Auckland
Acknowledgements:
Thanks to the staff of MMH ED for their advice and feedback on the
reports from this research project, and the GPs that gave up their time to work
on assessing the ED cases.
Correspondence: Dr
Barry Gribben, Department of General Practice and Primary Health Care,
University of Auckland, Private Bag 92019, Auckland. Fax: (09) 373 7006; email:
barry.gribben@cbg.co.nz
References:
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| | Current
issue | Search journal |
Archived issues | Classifieds
| Hotline (free ads) Subscribe | Contribute | Advertise | Contact Us | Copyright | Other Journals |
|