QJM Advance Access originally published online on February 9, 2006
QJM 2006 99(3):171-175; doi:10.1093/qjmed/hcl012
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Quality assurance of neuroradiology in a District General Hospital
From the 1Altnagelvin Neurological Centre, Londonderry and 2Department of Epidemiology and Public Health, Queen's University, Belfast, UK
Address correspondence to Dr Mark Owen McCarron, Altnagelvin Neurological Centre, Altnagelvin Hospital, Londonderry BT47 6SB. email: markmccarron{at}doctors.org.uk
Received 6 September 2005 and in revised form 1 January 2006
 |
Summary |
|---|
 Top Summary IntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Background: An increasing number of neurologists in District General Hospitals (DGHs) rely on local neuroimaging reports from general radiologists.
Aim: To determine the level of disagreement between general radiologists and neuroradiologists in reporting neuroimaging from patients referred to a neurologist.
Design: Prospective observational study.
Methods: We studied 232 patients referred for a neuroradiologist's report on neuroimaging over a 17-month period. Pre-planned comparisons included primary and secondary report findings, length of report and suggestions for additional investigations.
Results: Of the 593 patients assessed during the study period, a neuroradiologist's report was sought for 232 (39%): 119 men, 113 women, mean age 46.1 (SD 17.6) years. Primary findings differed in 37 patients (15.9%) (95%CI 11.5–21.3). Reports from neuroradiologists changed subsequent management in 31 (13.4%) (95%CI 9.3–18.4). Differences in secondary findings occurred in 52 (22.4%) (95%CI 17.2–28.3), and differences in either primary or secondary outcomes in 77 (33.2%) (95%CI 27.2–39.6). The level of disagreement in primary findings was as frequent among patients investigated with magnetic resonance imaging as among computerized tomogram-only patients (p = 0.13). Neuroradiologists recommended additional investigations for 24 patients (10.3%) (95%CI 6.7–15.0) and provided longer reports than general radiologists (p < 0.001).
Discussion: Neuroimaging reports of some patients differ substantially between general radiologists and neuroradiologists. Optimal management of neurological patients in DGHs may require timely access to neuroradiologists.
|
Introduction |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Neurology services are expanding within the UK. Goals and standards are being established for neurological emergencies, diagnostic services, and successful interaction with primary and secondary care.1 UK neurologists have an increasing and important role in managing patients in District General Hospitals (DGHs): approximately 19% of medical admissions have a neurological complaint.2,3 There is emerging evidence from different medical disciplines that specialist care may improve clinical outcomes for a number of diseases.4 Neurologists can make a valuable contribution to diagnosis and management,5–7 either amending an existing diagnosis or providing a diagnosis in undiagnosed patients. A liaison neurology service not only yields more diagnoses, but also shortens lengths of hospital stay,8 and such a service has been proposed as one way of improving access to specialist neurology advice.9
The increase in the number of consultant neurologists in the UK has coincided with the arrival of magnetic resonance imaging (MRI) scanners in many DGHs. Local investigation potentially avoids the need for transfer to a tertiary neurology unit, thus facilitating the assessment and management of more neurological patients by DGH neurologists. This enables a greater range of diagnoses, such as multiple sclerosis, stroke due to arterial dissection, or cerebral venous sinus thrombosis, to be made in a DGH. General radiologists of varying neuroradiological experience increasingly perform brain and spinal MRI scans and issue reports in DGHs. However, there are no measurements of quality assurance, and in Northern Ireland there was until recently only an informal means of accessing a neuroradiologist for a second opinion or report. The Association of British Neurologists document, UK Neurology—the next ten years: putting the patient first1, has highlighted that there are few standards for neurological services which can be used as audit tools to evaluate and improve services. Assessment of neuroimaging standards has largely relied upon virtual reporting, in which hypothetical case scenarios or scans are presented to clinicians who have agreed to participate in a study,10,11 and therefore may not reflect real practice.
We sought to determine the level of disagreement between reports from general radiologists and neuroradiologists in a selected group of neurological patients in a DGH and to assess whether any differences were of clinical relevance.
|
Methods |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Participants
In-patients and out-patients referred to a single consultant neurologist in a DGH were eligible for study. The DGH provided the diagnostic neuroradiological investigations: computerized tomography (CT) and magnetic resonance imaging (MRI). The study sample comprised consecutive patients for whom the neurologist sought a neuroradiologist's report on CT or MRI scans over a 17-month period (1 August 2003 to 31 December 2004). The neuroradiologists were given the same clinical information as provided on the initial referral forms (but additionally had the report from the general radiologist). Selection criteria included: advice from a general radiologist to obtain a neuroradiologist's report, concern on the part of the neurologist such as the presence or nature of a structural abnormality (as in patients with partial seizures or persisting neurological deficits), and doubt about the accuracy or the differential diagnosis of the general radiologist's report when neuroimaging was reviewed by the neurologist.
Outcome measures
Pre-planned outcome measures were determined. The primary outcome was the frequency of disagreement between general radiologists and neuroradiologists in the primary diagnosis or finding among the patients. Secondary outcomes included: (a) the frequency of disagreement in the secondary diagnosis/finding or the differential diagnoses; (b) the frequency of the composite of disagreements in either primary or secondary diagnoses/findings; (c) comparison of the length of reports from general radiologists and neuroradiologists; (d) the number (proportion) of patients who had subsequent investigations suggested by the neuroradiologists; and (e) the frequency of disagreement in the primary finding in patients who had only CT imaging compared to the primary finding in patients who underwent MRI.
Statistical analyses
The number and proportion of patients in whom neuroimaging reports differed between general radiologists and neuroradiologists were calculated, and 95%CIs for the proportions were computed using the standard formula for generating a confidence interval around a proportion: p (proportion) ± 1.96 x standard error. The 
2 test was used to compare the disagreement frequency between patients with MRI scans and patients with only CT scans, and to compare the proportionate disagreements of individual neuroradiologists with general radiologists in primary and secondary findings. As the number of words used in reports was positively skewed, the Wilcoxon matched pairs signed rank test was used to investigate differences in length of report.
|
Results |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
During the 17-month period of the study, the neurologist assessed 593 patients in whom neuroimaging had been performed because of the neurological complaint. Of these, 232 (39%) were selected for a neuroradiologist's report, comprising 119 men and 113 women, with an overall mean age of 46.1 (SD 17.6) years. Thirty-seven patients had only CT imaging, the remaining 195 patients having MRI with or without CT scanning. Differences in the reporting from general radiologists and neuroradiologists are shown in Table 1. The primary finding or diagnosis differed in 37 patients or 15.9% (95%CI 11.5–21.3). (Assuming no further primary outcome differences in the neuroradiological reports of the remaining 361 patients assessed by the neurologist, the minimum discrepancy proportion of all the DGH patients with neuroimaging during the study period would be 6.2%, 95%CI 4.4–8.5%.) Particular difficulty appeared to arise in distinguishing infarct from tumour, perivascular spaces (Virchow-Robin) from infarcts, and in determining the absence or presence of hippocampal sclerosis. Thirty-one (84%) of the 37 patients who had a primary finding difference had a change in their subsequent clinical management. Differences in the primary outcome measure were observed not only in patients with MRI scans in whom disagreements were recorded in 28/195 patients (14.4%), but also in patients who had only CT scans performed, in whom 9/37 patients (24.3%) had a different primary finding recorded. There was no evidence that the frequency of primary disagreement in MRI-investigated patients differed from that observed in CT-only patients (p = 0.13). In the group of 37 patients in whom the primary finding/diagnosis differed, the general radiologists had suggested a report from a neuroradiologist in five patients (13.5%).
|
Differences in secondary findings or differential diagnoses were found in 52 patients or 22.4%, (95%CI 17.2–28.3). Cortical or cerebellar atrophy was frequently identified by general radiologists but not confirmed by neuroradiologists (n = 11). Differences in secondary findings did not usually alter clinical management, although in one patient an incidental and small middle cerebral artery aneurysm, identified by a neuroradiologist, was investigated with a catheter angiogram. The disagreement rates for both primary and secondary findings were similar among the three neuroradiologists providing the second reports on each of their separate patient groups (Table 2). A composite of the differences in primary or secondary finding/differential diagnoses was recorded in 77 patients (33.2%) (95%CI 27.2–39.6%). Neuroradiologists contributed to subsequent management with advice on further neuroimaging in 24 patients (10.3%) (95%CI 6.7–15.0%). All reports were written 'from scratch', and neuroradiologists, on average, provided more detailed reports: median 102.5 (IQR 52–156.6) vs. 39 (IQR 17–68) words (p < 0.001). This difference remained even after removing the 77 patients in whom disagreements were identified: 88 (IQR 46–140) vs. 35 (IQR 15–69) words (p < 0.001).
|
|
Discussion |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Neuroradiologists are thought to provide the best available reports for neuroimaging investigations. Although such reports cannot be reliably taken as a 'gold standard', specialists generally tend to show higher levels of agreement than non-specialists.12 Progress in neuroimaging, particularly with MRI scanning, has had a profound impact on the management of neurological patients in the last 15–20 years, and continues to evolve. Our study has provided evidence that, without timely input from neuroradiologists, the increasing numbers of neurologists in DGHs may not fully realize the potential benefits of managing neurological patients locally. The disagreements in neuroradiological investigations highlighted in previous virtual studies10,11 have now been replicated in real practice. UK neurologists who have trained mainly in tertiary referral neurology units have been used to timely reports from neuroradiologists.13 As these neurology trainees are increasingly being appointed to do most of their clinical work in DGHs, there is an increasing need for a quality-assured neuroimaging service.
Our study also suggests that general radiologists alone are unable to recognize when a neuroradiologist's report would be useful. Conversely, a neurologist may seek a formal neuroradiologist's report too frequently, in that many such requests may not change clinical management. In previous work in patients with epilepsy, the sensitivity of the detection of focal lesions was higher when MRI scans were analysed by 'expert' neuroradiologists compared to non-expert radiologists,14 suggesting that particular subspecialist skill and experience are required to recognize focal lesions.
To date, most research on interobserver reliability in neuroimaging has focused on detection of early infarction on CT scanning prior to thrombolytic therapy.15 It is still uncertain whether signs of early infarction should influence clinical decisions concerning thrombolysis. However, in our selected group of patients, the majority of primary outcome disagreements had a clinical impact.
Study limitations
Our study has a number of limitations which merit consideration. Firstly, the patients were selected mainly by a single neurologist, and the results may have differed had all patients with neuroimaging in the DGH for the period of the study been enrolled. Because 61% or 361 patients did not have a neuroradiologist's report, the disagreement frequency of approximately 1:20 patients can only be regarded as a minimum among neurology patients with neuroimaging. Secondly, it may be argued that there is a learning curve for general radiologists while MRI scanners are installed and DGH neurologists are appointed. However, the MRI scanner in our DGH was installed more than a year before this study began, and neurologists had already been working in the DGH. In addition, the study demonstrated a similar frequency of disagreement in patients with CT brain scans only. There is already published evidence that 'non-expert' reporting may not improve despite advances in technical quality.14 Thirdly, as the study was not double-blinded (the neuroradiologists being aware of the reports from general radiologists), the difference in the length of the reports needs to be interpreted cautiously, even after removing reports with disagreements, as the differences may have resulted from the neuroradiologists directing their reports to both the neurologist and general radiologists: both were sent a copy of the reports. Fourthly, the study was carried out in one DGH; further work is required to investigate whether a similar level of disagreement occurs in other DGHs and whether the discrepancies persist when general radiologists and neuroradiologists are unaware of any ongoing study. Finally, no gold standard or independent adjudicator existed in the study to determine the correct report for those patients in whom there was disagreement. However, the study is an accurate representation of real neurology practice.
|
Conclusion |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Neurologists do not always rely on a written report of a neuroimaging scan to make a clinical decision.16 However, in selected patients, such as some of those in our study, a neuroradiologist's opinion can be pivotal. Our findings have important implications for general radiologists and all clinicians looking after neurological patients, including the increasing numbers of neurologists appointed to DGHs with neuroimaging facilities.13,17 The results support the assertion of the Association of British Neurologists that DGH neurologists require access to neuroradiologists.1 Quality-assured neuroimaging reports in DGHs within managed clinical networks or regular multidisciplinary neuroradiology meetings13 integrated into consultant job plans may improve the management of neurological patients. There is emerging evidence that neuroradiologists are forging important links with DGH neurologists and general radiologists through regular meetings.13 In addition, teleradiology has worked well for neurosurgeons, and the imminent arrival of a picture archiving and communication system in Northern Ireland offers opportunities for developing effective managed clinical networks for neuroimaging. Neurologists, who have suggested that their specialist care can improve patient management5–8,17,18 need to continue to explore areas of their work such as neuroimaging to optimally manage neurological patients in local hospitals.
|
Acknowledgments |
|---|
P. McCarron is supported by a career scientist award funded by the Research & Development Office for Health and Personal Social Services in Northern Ireland. We thank Jonathon Sterne for comment on the statistical analyses.
|
References |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
1. The Association of British Neurologists. UK Neurology-the next ten years: putting the patient first. 2003; [http://www.theabn.org/education/publications.html] (accessed December 2005).
2. Morrow JI and Patterson VH. The neurological practice of a district general hospital. J Neurol Neurosurg Psychiatry 1987; 50:1397–401.
3. Payford ED, Crawford P, Munro PS. A study of neurological disability at a District General Hospital. Brit J Clin Pract 1994; 48:304–6.[Web of Science][Medline]
4. Harold LR, Field TS, Gurwitz JH. Knowledge, patterns of care, and outcomes of care for generalists and specialists. J Gen Intern Med 1999; 14:499–511.[CrossRef][Web of Science][Medline]
5. Steiger MJ, Enevoldson TP, Haqmmans SR, Ginsberg L. Influence of obtaining a neurological opinion on the diagnosis and management of hospital in-patients. J Neurol Neurosurg Psychiatry 1996; 61:653–54.
6. Hillen ME and Sage JL. Proving the worth of neurologists. Neurology 1996; 46:276–7.
7. Nixon J, Bateman DE, Malthouse S. Computerised tomography scans of the head in a District General Hospital. J Roy Coll Phys Lond 1996; 30:547–50.[Web of Science][Medline]
8. Forbes R, Craig J, Callender M, Patterson V. Liaison neurology for acute medical admissions. Clin Med 2004; 4:290.[Web of Science][Medline]
9. Association of British Neurologists. Acute neurological emergencies in adults 2002; [www.theabn.org/education/publications.html].
10. Wardlaw J. ACCESS: the acute cerebral CT evaluation stroke study. Emerg Med J 2004; 21:666.
11. Wardlaw JM, Dorman PJ, Lewis SC, Sandercock PA. Can stroke physicians and neuroradiologists identify signs of early cerebral infarction on CT? J Neurol Neurosurg Psychiat 1999; 67:651–3.
12. Balabanova Y, Coker R, Fedorin I, et al. Variability in interpretation of chest radiographs among Russian clinicians and implications for screening programmes: observational study. BMJ 2005; 331:379–82.
13. Teasdale E. The neurologist in the DGH: how to survive without on site neuroradiology. J Neurol Neurosurg Psychiat 2005; 76:(suppl. III), 39–47.[CrossRef]
14. Von Oertzen J, Urbach H, Jungbluth S, et al. Standard magnetic resonance imaging is inadequate for patients with refractory focal epilepsy. J Neurol Neurosurg Psychiat 2002; 73:643–7.
15. Wardlaw JM and Mielke O. Early signs of brain infarction at CT: observer reliability and outcome after thrombolytic treatment—systematic review. Radiology 2005; 235:444–53.
16. Masdeu JC. What do neurologists think about their role in neuroimaging training and practice? J Neuroimaging 1999; 9:39–42.[Web of Science][Medline]
17. Warlow C, Humphrey P, Venables G. UK neurologists and the care of adults with acute neurological problems. Clin Med JRCPL 2002; 2:436–9.
18. Moulin T, Sablot D, Vidry E, et al. Impact of emergency room neurologists on patient management and outcome. Eur Neurol 2003; 50:207–14.[CrossRef][Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||