• 
• 

Three Part Question

In [interpretation of CT head for trauma and non-trauma] are [staff in the Emergency Department] able to [interpret results accurately and safely]?

Clinical Scenario

A 21-year old man attends the emergency department after a night out. He is intoxicated and has an occipital head injury. He apparently lost consciousness for 10 minutes and has vomited 4 times since arriving in the department. You decide to request a CT scan of his head. Local guidelines allow you to interpret this yourself. You wonder how robust this is compared to the old system of requesting the scan through the radiologist on call

Search Strategy

Medline 1966-2008 using the OVID interface.
[cranial computed tomography or CT head] AND [emergency]
LIMIT to human and English
Search repeated in EMBASE, CINAHL and Google Scholar

Search Outcome

191 papers were found of which 5 were relevant. Search of references revealed another 3 papers

Relevant Paper(s)

Author, date and country	Patient group	Study type (level of evidence)	Outcomes	Key results	Study Weaknesses
Mucci, Brett, Huntley, Greene 2005 UK	100 consecutive CT head scans requested by Emergency Department and reviewed by senior Emergency Department staff	Retrospective case series. Single centre	Concordance between ED report and consensus opinion of 2 radiologists	86.6% agreement. No findings missed that would change the overnight management of patient	Small study. Scans not anonymised
Peron, Huff, Ullrich 1998 USA	83 Emergency Medicine residents at 5 institutions	Prospective multi-centre study	Participants underwent exam of 12 CT scans. Then 2 hour course. Re-examined 3 months later. Comparison of pre and post test scores	Mean score in initial test was 60%. Retest 3 months later had mean score of 78% (p<.001 paired t test)	Only 63 residents completed post test so may be selection bias
Mehta, Mills, Jones 2005 UK	212 emergency cranial CTs over 5 months reported by senior ED doctors	Prospective single centre cohort study	Concordance between reports by senior ED staff and consultant neuroradiologist and between radiology residents and consultant neuroradiologist	Concordance between radiology residents and neuroradiologist was significantly higher (93.9%) than between ED staff and neuroradiologist (78.3%). 6.6% of discordant ED reports could have had adverse clinical outcome	Abstract only
Alfaro, Levitt, English 1994 USA	555 patients undergoing CT scanning in ED for trauma and non-trauma. Scans reported by 14 ED staff and result compared with radiology report	Prospective single centre cohort study	Concordance between ED report and radiology report	Non-concordance rate 38.7% between radiology and emergency physician reports. Clinically significant misinterpretations found in 24.1% of scans	Not blinded. Some of the scans reviewed retrospectively by ED
Arendts, Manovel, Chai 2003 Australia	1282 patients undergoing non-contrast CT head for trauma and non-trauma. Reported by senior ED staff	Prospecive blinded cohort study	Concordance between ED and consultant radiology report	190 scans (14.8%) were misinterpreted. 78 (6%) of these had potential or actual consequences. No significant difference with varying level of experience or qualification of ED doctor reporting scan	Sample selection bias - 43.3% of scans could not be included as reported by radiology before ED, these included a high proportion of abnormal scans
Khoo, Duffy 2007 Australia	315 consecutive out of hours scans reported by 7 Emergency Physicians and 14 registrars	Retrospective single centre study	Concordance between ED and consultant radiology report	67% agreement. 63 false positives	24 films excluded as lost
Levitt, Dawkins, Williams, Bullock 1997 USA	324 scans reported by 14 Emergency medicine residents. Phase I (217 scans) then 1 hour educational session then phase II (89 scans) 10 days later	Prospective single centre interventional study	Accuracy rate of reporting compared pre and post test	Concordance rate between ED and radiology report improved significantly from 61.3% pre-test to 88.6% post-test. Major missed findings decreased from 11.4% to 2.8% (p<.0001). No clinically significant mismanagement	Single radiology opinion
Perron, Kline, Carolinas 1997 USA	30 Emergency Medicine residents tested on interpretation of 10 CT scans. Then 2-hour educational session. Retested after 3 months	Prospective cohort study	Comparison of pre and post test scores	Pre test mean score 7, post test score 8. Scores significantly improved between pre and post tests	Very small numbers. Only 18 completed post test - selection bias?

Comment(s)

The non-concordance rate between emergency physician and radiologist reporting varies between 14 and 33% in these studies. 3 studies showed a significant number of misinterpretations which could have had adverse clinical consequences. It is unclear from the above studies what level of accuracy is needed for safe clinical practice and whether this is different between scans for trauma and non-trauma. 3 studies have shown that a single intervention with a teaching session on CT interpretation leads to a significant improvement in recognition of scans. Currently, in the South West region, Emergency Medicine trainees receive an educational session on CT head in trauma with a senior Consultant Neuro-radiologist as part of their training. There was no significant correlation between level of experience and qualifications of the reporting emergency doctor and misinterpretation rates. One study showed a significantly higher rate of discordance for abnormal scans (27% compared to 5.7% for normal scans) and postulates that the higher the number of normal scans included in a study, the better the ED reporting performance is likely to appear.

Clinical Bottom Line

Reporting performance for CT heads by Emergency Physicians is varied but is shown to improve significantly with a short educational intervention.

References

1. Mucci B, Brett C, Huntly LS, Greene MK Cranial computed tomography in trauma: The accuracy of interpretation by staff in the Emergency Department Emergency Medicine Journal 2005; 538-540
2. Peron AD, Huff JS, Ullrich CG et al A multi-centre study to improve emergency medicine residents' recognition of intracranial emergencies on computed tomography Annals of Emergency Medicine 1998; 32 (5) 554-562
3. Mehta A, Mills TD, Jones JO et al Accuracy of interpretation of emergency cranial computed tomography scans by radiology residents and the senior attending member of the requesting clinical team Radiology 2005; 221 (RSNA abstract supplement); 440
4. Alfaro D, Levitt MA, English DK et al Accuracy of interpretation of cranial computed tomography scans in an emergency training programme Annals of Emergency Medicine 1995; 25: 167-174
5. Arendts, Glenn; Manovel, Alvaro; Chai, Alan Cranial CT interpretation by senior emergency department staff Australasian Radiology 2003; 47, 368-374
6. Khoo NC, Duffy M Out of hours non-contrast CT head scan interpretation by Senior Emergency Department Medical Staff Emergency Medicine Australia 2007; 19(2) 122-128
7. Levitt MA, Dawkins R, Williams V et al Abbreviated educational sessions improve computed tomography scan interpretations by emergency physicians Annals of Emergency Medicine 1997; 30: 616-621
8. Perron AD, Kline JA Blood Can Be Very Bad: A simple mnemonic to improve the accuracy of cranial CT interpretation by the emergency physician (abstract) Annals of Emergency Medicine 1997; 30:385