British Journal of Radiology 75 (2002),136-139 © 2002 The British Institute of Radiology
Assessment of agreement between general practitioners and radiologists as to whether a radiation exposure is justified
R Dhingsa, FRCR
D B L Finlay, FRCR
G D Robinson
and
A J Liddicoat, FRCR
University Hospitals of Leicester NHS Trust, Leicester Royal Infirmary, Leicester LE1 5WW, UK
 |
Abstract
|
|---|
The objective of this study was to assess agreement between General Practitioners (GPs) and Consultant Radiologists as to whether a radiation exposure is justified and whether a request conforms to the Royal College of Radiologists (RCR) guidelines. Three GPs and three Consultant Radiologists were asked to review 100 requests for plain film imaging from GPs and to state whether the request justified a radiation exposure and whether the request conformed to RCR guidelines. It was discovered that there is greater agreement between radiologists than between GPs; this is a consistent pattern. The best agreement was between two Consultant Radiologists using the RCR guidelines. The poorest was between GPs using the request form details. It is suggested that the guidelines should be symptom-based to improve efficacy.
|
Introduction
|
|---|
General Practitioners (GPs) have enjoyed "open access" for plain film imaging in many hospitals, the aim of which is to shorten investigation time and to improve the quality of service offered to GPs [1]. As a group, GPs have little opportunity to discuss clinical cases and their management with Consultant Radiologists. The request card is usually their only means of communication. Additional information can be sought by radiologists or radiographers from the patient themselves or by contacting the GP.
Guidelines from the Royal College of Radiologists (RCR) [2] that aim to encourage more appropriate use of diagnostic radiology and so reduce the number of clinically unhelpful radiographic examinations have been available for 10 years. Previous studies have found that up to 20% of examinations are unhelpful [3–5].
In this study we have assessed agreement between three Consultant Radiologists and three GPs as to whether a radiation exposure is justified on the basis of information given on a request card and, second, whether the request conforms to RCR guidelines. We also assessed agreement between GPs and radiologists as a group.
|
Method
|
|---|
The study group comprised 100 consecutive requests from GPs for plain film imaging. The request forms were reviewed independently by three Consultant Radiologists and three GPs. The Consultant Radiologists, who regularly report plain films, consisted of a musculoskeletal radiologist, a cross-sectional radiologist and a senior lecturer. The GP group consisted of a senior lecturer and two senior GPs. All reviewers were given a copy of the RCR guidelines and were asked to state (1) whether they felt the request justified a radiation exposure and (2) whether the request conformed to the RCR guidelines.
The "majority decision" of the three radiologists and the three GPs was recorded. Agreement between the individual radiologists and between the individual GPs as well as between the "majority decision" of each group was assessed using the kappa statistic, which is a measure of agreement beyond the level of agreement expected by chance alone. It has a maximum of 1.00 when there is perfect agreement, a value of zero indicating no agreement better than chance; negative values show worse than chance agreement (see Table 1
) [6].
|
Results
|
|---|
Examination types and number of requests are shown in Table 2.
Only one request did not contain enough clinical information for all six reviewers to decide whether an exposure should be made.
Only 16 of the 100 requests were justified by all six reviewers. 12 of these requests were for chest radiography, 5 for suspected cancer, 3 for possible pleural disease, 2 for shortness of breath, 1 for suspected tuberculosis and 1 for investigation of haemoptysis. The remaining four justified requests consisted of two knee radiographs, for possible loose body and osteoarthritis, respectively, a pelvic radiograph for sacroilitis and a lower limb radiograph for suspected osteomyelitis.
4 of the 100 requests were considered "not justified" by all six reviewers. This group consisted of two cervical spine X-rays, one for investigation of vertigo and the other for neck pain. The remaing two comprised a lumbar spine radiograph for backache and a sinus view for sinusitis.
50 of the 100 requests were justified by all three GPs, however only 16 of these were justified by all three radiologists, with 6 not justified by any of the three radiologists. These six consisted of two foot radiographs and two knee radiographs for possible osteoarthritis, a chest radiograph for follow-up of shortness of breath and a lumbar spine radiograph for pain.
All six reviewers felt that 11 of the 100 requests conformed to RCR guidelines and that 9 requests did not. The 9 that did not conform consisted of six of the nine requests for imaging the cervical spine, (three for investigation of neck pain, two for vertigo and one to investigate dizziness), a chest radiograph to exclude a rib fracture, a knee radiograph for possible osteoarthritis and a lumbar spine radiograph for backache.
Importantly, in 40 of 100 cases at least one of the six assessors felt the request was not covered in the RCR guidelines. Of these, 22 were chest radiographs, 10 of which were requested to investigate shortness of breath, 4 for cough, 3 for infection, 2 for pleural disease and one each for investigation of an abnormal chest shape, suspected rib fracture and chest pain.
Agreement between the radiologists (R1, R2 and R3) and GPs (GP1, GP2 and GP3) and the request form details are shown in Table 3. There is moderate agreement between R1 and R2 and between R1 and R3, and fair agreement between R2 and R3. There is fair agreement between GP1 and GP2, GP1 and GP3, and GP2 and GP3. Agreement between radiologists and GPs ("majority decision") as a group is fair (kappa = 0.29).
Agreement between radiologists and GPs and the guidelines are shown in Table 4. There is good agreement between R1 and R2, and moderate agreement between R1 and R3 and between R2 and R3 as to whether a request conformed to RCR guidelines. With GPs there is fair agreement between GP1 and GP2, GP1 and GP3, and GP2 and GP3. Agreement between radiologists and GPs ("majority decision") is moderate (kappa = 0.47).
The best agreement is between two radiologists using the RCR guidelines. The poorest lies between GPs using the request form. The agreement between the "majority decision" is fair with the request form details and moderate if guidelines are used. There is greater agreement between radiologists than between GPs; this is a consistent pattern.
Chest radiography requests made up 36 of the 100 requests and this subset was analysed using the kappa statistic (Tables 5 and 6). Comparable with the total data set, the best agreement is between two radiologists using the RCR guidelines and the poorest is between GPs using the request form. Agreement between the "majority decision" is poor using only the request form and moderate with the guidelines. There is greater agreement as to whether a request falls within the guidelines than whether it should have been carried out at all.
|
Discussion
|
|---|
In May 2000 the Ionising Radiation (Medical Exposure) Regulations (IR(ME)R) were introduced, requiring justification of a request for imaging. Responsibility for justification of an exposure lies with a radiological practitioner. A practitioner must have adequate training to justify a request and therefore a GP is unlikely to be classified as a "practitioner" under IR(ME)R. Radiologists can only justify an exposure if sufficient relevant clinical information is provided on the request form. Therefore, we thought it useful to ask GPs to review request forms to consider if a request contained the relevant clinical information to warrant the radiographic examination. Guidelines are best devised in collaboration with all groups who will use and implement them. They are recommendations regarding when and how best to investigate clinical problems. Up to 20% of radiographic examinations are clinically unhelpful [3–5]. Previous studies have shown a decrease in referral rates following the introduction of guidelines in hospitals and general practice [1, 3], which in turn reduces the number of unhelpful radiographs. Guidelines offer structure and uniformity and we should aim to further decrease unnecessary exposures by improving guidelines. Three areas are highlighted as problem areas: chest, hand and cervical spine radiography. Chest radiographs have been reported to account for one-third of radiological examinations [7], which is also true of our group.
We have found that RCR guidelines are not symptom-based and are unhelpful in assessing which symptoms warrant imaging; for instance, neither "cough" nor "shortness of breath" are included in the guidelines, both of which may suggest underlying pathology, i.e. tuberculosis, bronchial carcinoma or pulmonary oedema. Hand radiography made up 7 of the 100 requests, 6 of which were thought not to be covered by guidelines by at least one of the six assessors. Four of these were for suspected fracture following no recent trauma. Six of nine cervical spine radiography requests did not conform to RCR guidelines.
Our results show that there is the greatest agreement between doctors when guidelines are used and least agreement using only the request form. Individual opinion as to whether a request is justified is based on an individual's experience and expectation of the result of imaging. Despite the small number of requests in our study, if extrapolated to a wider clinical setting we believe that this could be a major problem now that IR(ME)R is legally binding. We believe that modification of guidelines is required to make them more relevant to symptomatology. It has been noted that GPs use diagnostic radiology responsibly [8], and Winkens et al [9] found that feedback on diagnostic requests exerted a strong influence on GP request behaviour. GPs from our group commented that the request forms often contained inadequate information or were difficult to read. GPs felt that meeting with radiologists would be constructive and beneficial. This is especially important when introducing changes in practice. With the help of GPs we have formed our own local guidelines for chest, lumbar spine, paediatric and trauma radiography. We plan to repeat this study to see whether agreement improves and reduces the number of unhelpful examinations.
|
Acknowledgments
|
|---|
The authors acknowledge Dr N Dadge (GP), Dr P A Downs (GP), Dr S Longworth (Senior Lecturer, Department of General Practice, University of Leicester), Dr B Morgan (Senior Lecturer, Department of Radiology, University of Leicester) and Mr M Joshi (Department of Epidemiology and Public Health, University of Leicester).
Received for publication February 28, 2001.
Accepted for publication July 23, 2001.
|
References
|
|---|
-
Royal College of Radiologists Working Party. Influence of Royal College of Radiologists' guidelines on referral from general practice. BMJ 1993;306:110–1.
-
Royal College of Radiologists. Making the best use of a department of clinical radiology (4th edn). London: RCR, 1998.
-
Royal College of Radiologists Working Party. Influence of Royal College of Radiologists' guidelines on hospital practice: a multicentre study. BMJ 1992;304:740–3.
-
National Radiological Protection Board and Royal College of Radiologists. Patient dose reduction in diagnostic radiology. London: HMSO, 1990.
-
Chisholm R. Guidelines for radiological investigations. BMJ 1991;303:797–8.
-
Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977;33:159–74.[Medline]
-
Sandler G. Costs of unnecessary tests. BMJ 1979;2:21–4.
-
Williams JE. Making the best use of a department of radiology. In: Royal College of General Practioners 1990 members reference book. London; Sabrecrown, 1991.
-
Winkens RAG, Pop P, Grol RPT, et al. Effect of feedback on test ordering behaviour of general practitioners. BMJ 1992;304:1093–6.
This article has been cited by other articles:
|
|
|
|
 
Y Balabanova, R Coker, I Fedorin, S Zakharova, S Plavinskij, N Krukov, R Atun, and F Drobniewski
Variability in interpretation of chest radiographs among Russian clinicians and implications for screening programmes: observational study
BMJ,
August 13, 2005;
331(7513):
379 - 382.
[Abstract]
[Full Text]
[PDF]
|
|
|
Top
Abstract
Introduction
