Family Practice Vol. 18, No. 6, 586-589
© Oxford University Press 2001
Original Paper |
Chest pain in general practice or in the hospital emergency department: is it the same?
a Department of General Practice, Katholieke Universiteit Leuven,
b Department of Emergency Medicine, University Hospital Gasthuisberg, Leuven,
d Department of Statistics, University Centre of Limburg, Diepenbeek, Belgium and
c Department of General Practice, University of Maastricht, The Netherlands.
Professor Dr Frank Buntinx, Department of General Practice, Katholieke Universiteit Leuven, Kapucijnenvoer 33, blok J, B-3000 Leuven, Belgium.
Buntinx F, Knockaert D, Bruyninckx R, de Blaey N, Aerts M, Knottnerus JA and Delooz H. Chest pain in general practice or in the hospital emergency department: is it the same? Family Practice 2001; 18: 586–589.
Received 5 January 2001; Accepted 9 July 2001.
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionReferences |
|---|
Objective. The aim of the present study was to provide a description of the impact of setting on the diagnostic case mix that is identified in consecutive patients presenting with chest pain.
Methods. A cross-sectional study was carried out of patients presenting with chest pain according to setting: general practice, self-referred, referred or arriving by ambulance at the hospital emergency department (ED). GPs from 25 general practices situated in the Flemish part of Belgium were recruited, and the hospital involved was a major teaching hospital in the same area. A total of 320 patients in general practice and 580 patients in the hospital ED were studied. The difference in prevalence rates for the major diagnostic categories was the main outcome measure.
Results. Gastrointestinal disorders, musculoskeletal problems and psychopathology are identified more frequently in general practice; and serious lung diseases and cardiovascular diseases in the hospital ED. Within the hospital, there is a strong trend towards increasing frequency of serious cardiovascular diseases including unstable angina (P = 0.01) from self-referred to referred patients and those rushed in by ambulance. The opposite trend was identified for respiratory (P = 0.02) and musculoskeletal (P = 0.07) diseases. The diagnostic case mix in self-referred patients tends to be more similar to the other groups of hospital patients than to patients in general practice.
Conclusions. There is a large difference between the diagnostic case mix presented in general practice compared with the ED and among referral-related subgroups within the hospital emergency department.
Keywords. Chest pain, general practice, hospital emergency department, referral.
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
It has been suggested in the past that indicators of diagnostic test accuracy are not only related to test characteristics, but also to patient and setting characteristics.1–4 To a large extent, this would be the case when comparing primary care with secondary care patients since different selection mechanisms, including differences in diagnostic case mix, are supposed to influence both populations. Empirical evidence with respect to the differences in case mix among patients presenting with similar complaints in different settings, however, is scarce.
We therefore tried to study the impact of setting on the diagnostic case mix that is identified in consecutive patients with a seemingly identical situation: we compared all patients presenting with chest pain to either a GP (in the surgery or via a house-call) or the emergency department of a large university teaching hospital in the same region from which the GPs were recruited. Within the hospital group, setting was defined as either self-referred, referred by another doctor (generally a GP as transfers from other hospitals were excluded) or brought in by ambulance. We looked for the proportion of patients finally classified within one of 11 diagnostic categories.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Patients
Included were all consecutive patients presenting with a new episode of chest pain, chest discomfort or tightness. In the general practice registration, patients were included during two consecutive months (February and March 1988) when presenting to collaborating GPs. Practices were spread over the Flemish part of Belgium. A detailed analysis of this registration has been published before.5,6 In the hospital-based registration, patients were included consecutively during a 7-month period in 1993–1994.
Hospital patients referred by their GP, self-referred or admitted by ambulance were classified separately. Patients admitted by ambulance after referral by their GP were classified as referred. Excluded were patients referred by another hospital, patients admitted after trauma or directly to the surgical department, and patients with acute life-threatening signs and symptoms such as shock or focal neurological symptoms.
Diagnoses
Diagnoses were classified according to the International Classification of Primary Care (ICPC),7 including a third numerical digit whenever subclassification was needed. For the analysis, only final diagnoses were used. For hospital patients, these were decided upon by an independent researcher after discharge of the patient and were based on all information and test results available. For GP patients, the final diagnoses were stated after a follow-up period of 2 months, and were based upon all available information at that time.
Procedure
After each contact with a patient meeting the inclusion criteria, the physician completed a questionnaire providing data on patient characteristics, signs and symptoms from the files. Follow-up data including the final diagnosis were added after discharge of the patient (hospital) or after the end of the follow-up period (general practice).
Analysis
Prevalences were estimated as the proportion of patients finally diagnosed with a disease, using the total group of patients presenting with chest pain in each setting during the registration period as the denominator. Differences between GP patients and hospital patients as well as among the three subgroups of hospital patients were tested using a chi-square test for independent proportions. Additionally, in hospital-based patients, a possible linear trend in the proportion of patients with a disease category from self-referred patients over referred patients to patients who arrived by ambulance was tested using a chi-square test for linear trend. We controlled detected differences in the prevalence of each diagnostic category between settings for possible confounding by age and sex by comparing odds ratios based on crude data with odds ratios adjusted for age and sex. These were calculated by logistic regression analysis.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Characteristics of the study population
Complete data were available for 900 patients, 320 in general practice and 580 in the emergency department. A final diagnosis was missing in 1.3 and 4.5%, respectively. Patients presenting with chest pain in general practice tend to be younger (mean = 45 years, SD = 19) compared with those at the emergency department (mean = 60 years, SD = 18). Within the latter group, self-referred patients are younger (mean = 52 years, SD = 17) than referred patients (mean = 62 years, SD = 17) and ambulance patients (mean = 64 years, SD = 18). There is no significant difference between the sex distributions. Chest pain is rare in children and young people. Only 1.6% of the patients were younger than age 16.
Influence of setting on prevalence
There is a strong and statistically highly significant difference between the diagnoses presented in general practice compared with the emergency department and among the three subgroups within the hospital emergency department (P < 0.01 in both comparisons) (Table 1
).
|
Gastrointestinal disorders, musculoskeletal problems and psychopathology are detected more frequently in general practice. Serious lung diseases and especially cardiovascular diseases were far more frequent in the hospital emergency department. Within the hospital, there is a strong trend towards increasing frequency of serious cardiovascular diseases including unstable angina (P = 0.01) from self-referred to referred patients and those brought in by ambulance. An opposite trend was identified for respiratory (P = 0.02) and musculoskeletal (P = 0.07) diseases. The diagnostic case mix in self-referred patiens tends to be more similar to the other groups of hospital patients than to patients in general practice. The exceptions are angina pectoris and heart neurosis. These findings remained stable after adjusting for age and sex using logistic regression analysis.
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
This study quantifies differences in diagnostic case mix in patients with the same reason for encounter, according to the medical setting and the method of referral. Differences between referred and self-referred patients can only be studied in countries in which self-referral of patients to the hospital emergency department is common, as is the case in Belgium. The study design is weakened by a difference between the registration period of the general practice and the hospital part of the study. Apart from unstable angina, which was not classified separately during the GP study period, there is no reason, however, to expect that prevalence results on the level of aggregation that was used have changed significantly within this period of time. Final diagnosis was based upon follow-up in both registrations. However, more technical tests certainly were performed and more patients were hospitalized in the emergency department patients, where specialized consultants were readily available. This may result in some degree of unavoidable verification bias.4 In both groups there was no pre-defined list of technical tests to be performed. The registration form and all other procedural details were identical between the groups and, within the emergency department, the procedure was identical for patients of all three subgroups. Musculoskeletal disorders and psychopathology are essentially clinical diagnoses. They are always difficult to confirm and susceptible to interobserver variation. This may have influenced our data with respect to these diagnostic categories. It may also influence comparison with other registration data.
A strong and statistically significant relationship has been identified between setting and the diagnostic yield in patients presenting with chest pain. This is the case when comparing GP and hospital patients as well as when comparing the three subgroups of hospital patients, differing according to the way in which they were referred to the hospital. Although some age differences exist between the different groups, these differences could not explain our findings. The number of undefined diagnoses (‘unknown’) differs between the study groups and is highest in patients who arrived by ambulance. It is therefore possible that more severe cases occurred in this group than estimated in this study. This would only increase the magnitude of the differences we detected.
The differences between the primary and a secondary care setting are not only related to selection by referral by the primary care physician. There is a larger difference in the proportion of patients with a certain disease (e.g. cardiovascular disease) between GP patients and self-referred hospital patients, than between self-referred and referred patients within the hospital emergency department. Self-referral does not seem to be used inadequately.
This study confirms the impressions of many clinicians that the morbidity spectrums in general practice and in hospital are related to both the setting and, within the hospital, the method of referral. This can be expected to have consequences when studying the diagnostic accuracy of tests in this field of care. It may also lead to a more efficient and less costly assessment of non-cardiac chest pain patients, both in general practice and within the hospital.
|
Acknowledgments |
|---|
We thank the GPs and the hospital staff who registered the data that were used for this analysis, and Marina Devis for assistance in preparing both the data set and the final manuscript.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
1 Knottnerus JA, Leffers P. The influence of referral patterns on the characteristics of diagnostic tests. J Clin Epidemiol 1992; 45: 1143–1154.[Web of Science][Medline]
2 Ransohoff DF, Feinstein AR. Problems of spectrum and bias in evaluating the efficacy of diagnostic tests. N Engl J Med 1978; 299: 926–930.[Abstract]
3 Begg CB, Greenes RA. Assessment of diagnostic tests when disease verification is subject to selection bias. Biometrics 1983; 39: 207–215.[Web of Science][Medline]
4 Knottnerus JA. The effects of disease verification and referral on the relationship between symptoms and diseases. Med Decision Making 1987; 7: 139–148.
5
Buntinx F, Truyen J, Embrechts P, Moreel G, Peeters R. Chest pain: an evaluation of the initial diagnosis made by 25 Flemish general practitioners. Fam Pract 1991; 8: 121–124.
6
Buntinx F, Truyen J, Embrechts P, Moreel G, Peeters R. Evaluating patients with chest pain using CART. Fam Pract 1992; 9: 149–153.
7 Lamberts H, Wood M. International Classification of Primary Care. Oxford: Oxford Medical Publications, 1987.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  M. A. Hani, H. Keller, J. Vandenesch, A. C Sonnichsen, F. Griffiths, and N. Donner-Banzhoff Different from what the textbooks say: how GPs diagnose coronary heart disease Fam. Pract., December 1, 2007; 24(6): 622 - 627. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||