Factors Influencing the Detection of Early CT Signs of Cerebral Ischemia
An Internet-Based, International Multiobserver Study
Background and Purpose— Early CT signs of cerebral ischemia are subtle. Little is known of which factors influence the detection of infarct signs. We compared neuroradiologists’ scan readings with those of other specialists involved in the care of stroke patients.
Methods— We used the Internet to show 63 CT scans, all acquired <6 hours after stroke and representing different patient ages, times to scanning, stroke severity, and early CT signs of ischemia to physicians involved in stroke care. They completed a structured scan interpretation proforma over the Internet. We compared the detection of early ischemic signs stratified by severity and the effect of prior stroke between different specialties.
Results— Among 207 observers, neuroradiologists saw significantly more of “any early ischemic changes” than did stroke physicians, general radiologists, geriatricians, or neurologists (all P<0.0001), predominantly due to neuroradiologists’ greater detection of “mild” hypoattenuation or swelling. Detection of “severe” hypoattenuation or swelling, and hyperattenuated arteries did not differ between specialties. Old infarcts impaired recognition of early ischemic signs. Nonneuroradiologists did not “overcall” signs. Years of scan-reading experience did not account for these differences, but neuroradiologists took, on average, 30 seconds longer to read each scan than did most other specialists (P<0.0001).
Conclusions— Nonneuroradiologists should realize that they are unlikely to overcall signs, that old infarcts may distract them from seeing early ischemic signs, and read stroke CT scans more slowly, as these factors may help them perform more like neuroradiologists.
CT scanning early after stroke can make a positive diagnosis of ischemic stroke and exclude other causes of acute focal neurology. CT is widely available and can be used in virtually all stroke patients. However, early CT signs of ischemia are subtle, and physicians may lack confidence in their ability to detect or exclude early CT signs of ischemia, especially if thrombolysis is being considered.
Detection of CT signs of early ischemia is not good.1 In a recent systematic review, the interobserver agreement for “any early CT sign” was 0.14 to 0.78 (κ), with sensitivities and specificities for infarct sign detection of 20% to 87% and of 56% to 100%, respectively. Most studies had relatively few observers who were not categorized by experience or training, tested few scans, compared observers with only 1 or 2 experts, and gave little information on how the scans had been selected. Therefore, it was difficult to generalize from those studies to routine acute stroke diagnosis, determine which early ischemic changes were best recognized, or determine which factors (eg, observer experience, knowledge of symptoms, and patient-related factors like previous stroke) might influence scan interpretation. The comparison with only 1 or 2 experts may have made the observers’ recognition of early CT signs appear relatively poorer than it actually was in day-to-day practice. Comparison with a broader group of neuroradiologists (whose day-to-day role is to read CT brain scans but who may not have had the advantage of reading very large numbers of hyperacute stroke CT scans) might provide a more realistic estimate of how well neurologists, stroke physicians, and others involved in stroke patient care might perform on a daily basis.
We therefore performed a large observer-reliability study to determine which ischemic changes were most reliably detected, whether there were differences in observer performance attributable to their scan-reading experience, and which scan-related factors might impair detection of CT signs of early ischemia.
We established an Internet-based scan-reading tool (www.neuroimage. co.uk). To achieve as large a group of observers as possible, participants viewed CT scans and recorded their responses over the Internet. We applied STARD principles to the study design and report.2
We selected CT scans by (1) patient characteristics (age, duration of symptoms, and stroke severity) and (2) presence of specific signs. Scans chosen for specific patient characteristics came from 120 sequential patients admitted to a teaching hospital and who underwent CT scanning within 6 hours of acute stroke, from which we randomly chose 22 scans representing time to scanning (half <3 hours and half 3 to 6 hours after stroke); patient age (half <70 years and half >70 years); and stroke severity (National Institutes of Health Stroke Scale score; half ≤12 and half >12), independently of which signs the scans showed. Scans chosen for specific early ischemic changes (n=10) independent of patient characteristics came from previous3 and ongoing (www.IST3.co.uk) trials of thrombolysis in stroke.
The scans were shown anonymized in batches of ≈10, the first 5 batches without any clinical information and the last batch with the side of brain symptoms. The scans were managed electronically with a DICOM reader Digital Jacket (DesAcc, Inc, Chicago Ill) and were optimized for gray/white matter differentiation.
We sought as many observers as possible through relevant conferences, newsletters, professional organizations, and journal articles.4,5 Participation was encouraged by awarding 5 Continuing Professional Development credits from the UK Royal Colleges for reading all scans and several monetary prizes.
Observers logged onto the Web site (www.neuroimage.co.uk), registered which specialty (chosen from a list, Table 1) most closely represented their work (Table 1), years of training in that specialty, and country of origin. They optimized monitor settings (contrast/brightness) and ambient light for detecting subtle gray-scale differences with a Society of Motion Picture Television Engineers test on the www.neuroimage.co.uk Web site. They read a test scan to familiarize themselves with the Web scan-viewing tool (the Figure) and the structured questionnaire. Thereafter, study scans were assigned in batches. Each scan could be viewed in “light box” or “stack” mode (Figure). The time taken to complete the questionnaire was recorded.
The questionnaire was developed and tested on acute stroke CT scans outside the study cohort. The signs of early ischemia on CT are (1) decreased parenchymal x-ray attenuation, (2) tissue swelling (mass effect), and (3) hyperattenuated artery sign (arterial occlusion).6 The final questionnaire recorded scan quality; any change in attenuation or swelling (and whether mild or severe); the arterial territory(s) affected; tested 3 scoring systems (1/3 MCA rule,7 IST3 method,8 and the Alberta Stroke Program Early CT Score [ASPECTS]9); and whether there was any hyperattenuated artery or other abnormality (atrophy, tumor, hemorrhage, or old infarct). We defined mild hypoattenuation as gray matter reduced to that of normal white matter attenuation; severe hypoattenuation as gray or white matter attenuation less than normal white matter; mild swelling as effacement of the ipsilateral cortical sulci or slight effacement of the lateral ventricle, and severe swelling as complete effacement of the lateral ventricle or midline shift.
We analyzed data from readers who completed all 63 scans. Data were normally distributed. The readings of 1 observer, a neuroradiologist internationally acknowledged to be very experienced in interpretation of CT in acute stroke, were used to independently determine the proportion of scans with each CT sign. This was to characterize the scans, not to compare other readers with this 1 expert, as the latter analysis would not be generalizable to routine practice: 1 expert’s CT scan readings are not a perfect “gold standard,” so sensitivity/specificity analyses would not be appropriate, nor would a κ of each individual versus the expert address day-to-day practice. Therefore, for each scan, we calculated the proportion of neuroradiologists who recognized any sign and each individual sign and compared that with the proportion of each other specialty who recognized those signs. We then compared the proportion of neuroradiologists to the proportion of each other specialty (ie, continuous variables) by Bland-Altman analysis, with probability values calculated from 1-sample t tests.10 We also examined whether the presence of old infarcts, atrophy, or periventricular white matter lesions affected the proportion of each specialty’s ability to recognize signs by χ2 tests. Significance was set at the P<0.05 level.
Demographics of Scan Readers
Two hundred seven readers from 36 countries, mostly the UK (40%), Germany (18%), and Italy (14%), completed all 63 scans (Appendix). Nearly half of the 207 participants classified themselves as neurologists (99, 48%), then general radiologists (25, 12%), neuroradiologists (24, 12%), geriatricians (21, 10%), stroke physicians (16, 8%), and “others” (22, 11%; general practitioners, emergency medicine physicians, miscellaneous; Table 1).
Approximately one third (63/207) of readers had been reading CT scans for stroke as part of their routine daily work for <5 years, 77 of 207 (37%) for between 5 and 10 years, and 66 of 207 (32%) for 10 or more years (Table 1). Excluding others, general radiologists had the largest proportion of experienced readers (44% with >10 years’ experience), and stroke physicians had the largest proportion of least-experienced readers (44%). Neuroradiologists, neurologists, and geriatricians had similar experience to each other (Table 1, P=NS).
The average time taken to read a scan and complete the questionnaire was 3.4 minutes (2SD=10 minutes; range, 6 seconds to 10 minutes). The mean time to read a scan was 222 seconds/scan for neuroradiologists versus general radiologists 193 seconds, neurologists 193 seconds, and geriatricians 189 seconds; ie, neuroradiologists took 29 seconds longer on average than both general radiologists and neurologists and 32 seconds longer than geriatricians (all P=0.0001); stroke physicians’ times were not significantly different from those of neuroradiologists (P=0.8).
Detection of Early CT Signs of Ischemia
There were early CT signs of ischemia in 43 of 63 (68%) scans according to the reference expert. The proportion of readers who identified any acute ischemic change per scan was highest among neuroradiologists. More than 50% of neuroradiologists identified signs of acute ischemia in 43 of 63 scans. Compared with neuroradiologists, other specialties combined identified any early signs of ischemia less often, on average by 12.3±1.4% (P<0.0001, Table 2). Compared with neuroradiologists (and excluding others), neurologists recognized the least of any acute ischemic change (14.6±1.3% less often than neuroradiologists), followed by geriatricians (13±1.8% less often), general radiologists (6.8±2.2% less often), and stroke physicians (6.4±1.7% less often), all P<0.005.
A pattern similar to that for any early signs of ischemia was seen for neuroradiologists versus the other specialties in recognizing mild hypoattenuation or mild swelling. On average, neuroradiologists saw mild hypoattenuation (13.5±1.7%) and mild swelling (13.4±1.4%) more often (P<0.0001) than other specialties (Table 2). There was no difference between specialties in recognition of severe hypoattenuation or swelling (Table 2). There was no difference between neuroradiologists and most other specialists’ recognition of the hyperattenuated artery sign. Only general radiologists saw hyperattenuated arteries significantly less often (8.7±1.8%, P=0.0001) than neuroradiologists (Table 2).
There was little “overcalling” of any signs by any specialty compared with neuroradiologists (minus numbers in Table 2) and none that reached statistical significance. Because the minus numbers that do occur for parenchymal signs are in the detection of severe hypoattenuation or swelling (not mild), this may simply reflect use of terminology in calling a finding severe rather than mild, rather than in seeing a sign that was not there.
Effect of Old Infarcts
Neuroradiologists were less likely to be distracted by old infarcts from seeing early ischemic changes than were those in other specialties. Overall, neuroradiologists identified the largest proportion of acute ischemic changes, whether old infarcts were present or not (Table 3). When an old infarct was present (Table 3), neuroradiologists, general radiologists, and stroke physicians were not particularly distracted from identifying early signs of ischemia, but neurologists, geriatricians, and others classified more scans as having only an old infarct and missed the presence of acute ischemic changes significantly more often than did neuroradiologists (χ2 P<0.05).
This large study of reader reliability of early ischemic stroke signs on CT has demonstrated that neuroradiologists identified early ischemic signs more often than did other specialists, but most of this difference was due to the neuroradiologists’ greater detection of mild hypoattenuation and tissue swelling. Other specialists saw hyperattenuated arteries and severe parenchymal changes as often as neuroradiologists. Parenchymal hypoattenuation and tissue swelling were recognized equally. There was little if any overcalling of early ischemic changes by other specialists compared with neuroradiologists. The chances of missing early ischemic changes were increased by the presence of an old infarct.
These differences between specialties were not simply due to years of observer experience, as stroke physicians (who performed most like neuroradiologists) included more observers with fewer than 5 years’ experience of scan reading than most other groups; neurologists (who saw fewer mild signs) had very similar years of experience in acute stroke CT reading to neuroradiologists.
It may be that improved detection of early ischemic changes was related to the time taken to read a scan, as neuroradiologists took significantly longer, even though by only half a minute, to read scans than did all other groups except stroke physicians. Perhaps the advice from this study is “slow down, and don’t be put off by old infarcts—keep looking.” Whether or not the neuroradiologists and stroke physicians also approached the scan reading in some more organized way that improved detection of CT early ischemic changes or not we are unable to say, but the structured questionnaire used in the study encouraged all observers to examine the scan in the same structured way.
How does the present study compare with previous work? We tested more scan-observer interactions (13 041), the previous largest study having had 530 observers (mostly neurologists) and 20 scans11 (10 600 scan-observer interactions). We also used scans chosen randomly from a defined population to be representative of characteristics of patients presenting within 6 hours of stroke to reduce bias.2 In previous studies, the basis on which scans were chosen was unclear.
The strengths of the present study are the large number of scans and observers, the careful method for choosing scans to represent real-life stroke patient demographics, the analysis by specialty and years of experience, the careful design of the scan interpretation questionnaire, and the analysis comparing a large group of neuroradiologists with other groups.
What are the limitations? Although the scans were presented on optimum window settings and readers were instructed to optimize their viewing conditions, we do not know whether readers actually followed these instructions. It is likely that we attracted interested persons who may be better at acute stroke CT scan reading than the average of their specialty. However, the relative conclusions (“hasty reading misses subtle signs, and beware of distracters”) are unlikely to be substantially different from real life. There is relatively little information on whether electronic scan viewing has different sensitivity than cut film, but the available information suggests that electronic scan viewing is as sensitive and specific as cut film, provided the general viewing conditions (ambient light, etc) are appropriate.12 Participants classified their specialties from a list of possible choices at registration, being instructed to “indicate the specialty that most closely describes their qualifications, training and current job.” However, stroke is cared for by different specialties in different countries; eg, mostly by geriatricians or stroke physicians in the UK, and not all countries have “stroke physician” as a specialty.
What are the implications? Differences in the detection of subtle early signs of ischemia might lead to differences in estimation of prognosis or in treatment decisions about whether to use thrombolysis, depending on whether a neuroradiologist was available or not, especially if a strict cutoff were used on a scoring system like ASPECTS. However, the available evidence suggests that the relation between ASPECTS scores and prognosis is not sufficiently precise for strict cutoffs to be valid. Physicians should not just rely on radiological findings when making treatment decisions.
The scans from the present study remain available for training in acute stroke to new readers (www.neuroimage. co.uk), with feedback from the reference standard, a panel of experts, all readers and readers grouped by specialty, and the 24-hour follow-up scan showing where the infarct developed.
Participants Who Read All 63 Study CT Scans
Argentia: Dr Claudia Gandolfo, Churruca Hospital, Buenos Aires; Dr Guillermo Povedano.
Australia: Prof Lesley Cala, Australian Research Centre for Medical Engineering, Perth; Dr Sandy Patel.
Austria: Dr Birgit Glawar; Dr Erich Gatterbauer, Vienna; Dr Klaus Hohensinner.
Belgium: Dr Andre Peeters, UCL St Luc, Brussels; Dr Philippe Demaerel, University Hospital, Leuven; Dr Vincent Thijs.
Brazil: Dr Luciano Farage, Hospital de Clinicas de Uberlandia UFU, Brasilia DF.
Canada: Dr Jessica Simon; Dr Michael Hill; Dr Nicole Isaac, Lakeridge Health Oshawa, Toronto; Dr Steve Phillips, Queen Elizabeth II Health Sciences Centre, Halifax.
China: Dr Henry Ka Fung Mak; Dr Ming Hua Wu, Stroke Center of Jiangsu TCM Hospital, Nanjing; Dr Xianrong Zeng, Chinese Medical Science Institution of Sichuan, Chengdu.
Denmark: Dr John Buch Larsen.
Finland: Dr Anna Saukkonen, North Karelian Central Hospital, Joensuu.
France: Dr Thierry Moulin, Neurologie CHU Besancon, Besancon.
Germany: Dr Dieter Hofler, Neurologische Klinik, Bamberg; Dr Hermann Weber, Neurologische Klinik, Bamberg; Dr Lutz Harms, Medizinische Fakultät der Humboldt-Universität, Universitätsklinikum Charité Berlin; Prof Detlef Claus, Klinik für Neurologie, Klinikum Darmstadt; Dr Georg Gahn, Klinik und Poliklinik für Neurologie, Universität Dresden; Prof Rüdiger von Kummer, Neuroradiologische Klinik, Universität Dresden; Dr Hans-Bernd Huelsboemer, Marien-Hospital Düsseldorf, Neurologische Abteilung; Dr Ulrich Junghans, Neurologische Klinik, Heinrich-Heine-Universität Düsseldorf; Dr Rene Handschu, Neurologische Klinik, Universität Erlangen-Nürnberg; Dr Christiane Ehrenfeld, Neurologische Klinik, Neurologische Universitätsklinik, Essen; Dr Rolf Diehl, Alfried Krupp Krankenhaus Essen, Neurologische Klinik; Dr Rolf Adams, Neurologische Abteilung, St. Katharinen Hospital, Frechen; Dr Thomas Kucinski, Neurologische Universitätsklinik Hamburg; Dr Peter Schellinger, Department of Neurology, Heidelberg; Dr Amélie Fellgiebel, Klinik für Neurologie, Itzehoe; Dr Andreas Thie, Medizinisches Zentrum Itzehoe, Krankenhaus Itzehoe; Prof Andreas Ferbert, Neurologische Klinik, Kassel; Dr Timm Strotmann, Klinik und Poliklinik für Neurologie und Psychiatrie, Universität zu Köln; Dr Michael Huppertz, Neurologische Klinik, Klinikum Lippe Lemgo; Dr S. Mink, Neurologische Klinik, Neurologische Klinik Ludwigsburg; Dr Michael Wunderlich, Klinik für Neurologie, Otto-von-Guericke-Universität Magdeburg; Dr Christian Peters, Neurologische Klinik, Universitätsklinikum Mannheim; Dr Daniel Böhm, Neurologische Klinik, Neurologische Klinik Merzig; Dr Frank Hamilton, Neurological Department, Klinikum Minden; Prof Otto Busse, Neurologische Klinik, Klinikum Minden; Dr Andreas Bitsch, Ruppiner Kliniken Neuruppin; Dr Catrin Utt, Neurologische Klinik, Asklepios Klinik Nidda-Bad Salzhausen; Dr Jürgen Haspel, Neurologische Klinik, Kreiskrankenhaus Nürtingen; Dr Alfons Meyer, Neurologische Abteilung, Krankenhaus Plau am See; Dr Anton Pencz, Neurologische Klinik, Klinikum Remscheid; Dr Juergen H. Faiss, Neurologische Klinik, Landesklinik Teupitz; Prof Roman Huber, Neurologische Klinik, Universität Ulm; Dr Gunther Landmann, Klinische Neurophysiologie, Universität Weimbar, Weimar; Dr Wolfgang Liesenhoff, Kreiskrankenhaus Ammerland, Neurologische Abteilung, Westerstede; Dr Joerg Wiesenfeldt, Verbundkrankenhaus Bernkastel-Wittlich, St Elisabeth Krankenhaus; Dr Achim Mallmann, Innere Klinik I, Stadtkrankenhaus Worms.
India: Dr Himanshu Roy, H M Patel Center of Medical Care and Education, New Delhi; Dr Krishnamoorthy Thamburaj, Sree Chitra Tirunal Institute for Medical Sciences and Technology, Trivandrum, Kerala; Dr Malay Patel, H M Patel Center of Medical Care and Education, Vallabh Vidyanagar.
Italy: Dr Vincenzo Sidoti, AOM Mellini, Bergamo; Dr Gian Paolo Anzola, Service of Neurology, S. Orsola Hospital FBF, Brescia; Dr Alessandro Romorini, Neurologia, Ospedale San Carlo Borromeo, Castano Primo; Dr Rinaldo Colombo, Asl4 Chiavarese Divisione Neurologia, Chiavari; Dr Biagio Samuele Cotticelli, Policlinico San Donato, Cremona; Dr Riccardo Tamarozzi, Arcispedale S. Anna-Azienda Ospedaliera Universitaria di Ferrara; Dr Annalisa Vinattieri, Divisione Neurologia Ospedale Prato, Firenze; Dr Giuseppe Cascone, A O Umberto I, Floridia; Prof Carlo Gandolfo, Dipartimento Di Scienze Neurologiche, Clinica Neurologica, Universita di Genova; Dr Alfonso Ciccone, Ospedale Niguarda Ca’ Granda, Milan; Dr Federica Lucchelli, Neurologia, Ospedale San Carlo Borromeo, Milan; Dr Rachele Tortorella, Division Neurologia Ospedale di Cremona, Milan; Dr Antonella Gatti, Ospedale Niguarda Ca’ Granda, Milan; Dr Gianluigi Valenti, Neurologia, Ospedale San Carlo Borromeo, Milan; Dr Patrizia Lattuada, Neurologia, Ospedale San Carlo Borromeo, Milan; Dr Jessica Mandrioli, Clinica Neurologica, Policlinico di Modena, Università di Modena e Reggio Emilia; Dr Emanuela Botto, Ospedale Maggiore della Caritu, Ospedale Maggiore della Caritu, Novara; Dr Lorenzo Lorusso, AOM Mellini Pavia; Dr Francesco Corea, Perugia Stroke Team, University of Perugia; Dr Maurizio Paciaroni, Perugia Stroke Team, University of Perugia; Dr Nicola Tambasco, Department of Neuroscience, University of Perugia; Dr Andrea Zini, Clinica Neurologica, Policlinico di Modena, Università di Modena e Reggio Emilia; Dr Carmela Gerace, Azienda Osp. S.camillo-Forlanini, Rome; Dr Angelamaria Santoro, Ospedale San Giacomo, Rome; Dr Claudio Pozzessere, Ospedale San Camillo-Roma; Dr Antonio Basile, Internal Medicine Stroke Unit, San Vito al Tagliame; Dr Carlo Buffa; Dr Edoardo Boccardi; Dr Francesco Augusto Marras, Italy.
Lithuania: Dr Aleksandras Vilionskis, Department of Neurology, Vilnius Emergency University Hospital.
New Zealand: Dr William Taylor, Hutt Valley District Health Board, Hutt Hospital, Lower Hutt.
Philippines: Dr Maria Teresa Canete, Chong Hua Hospital, Cebu City
Poland: Dr Michal Arkuszewski, Departament of Neurology, Ageing, Degenerative and Cerebrovascular Diseases, Chorzow; Dr Adam Kobayashi, Institute of Psychiatry and Neurology, Warsaw; Dr Iwona Sarzynska Dlugosz, Institute of Psychiatry and Neurology, Warsaw; Dr Tadeusz Mendel, Institute of Psychiatry and Neurology, Warsaw.
Portugal: Dr Antonio Fernandes, Hospital de S. José, Lisbon; Dr Manuel Correia, Serviço de Neurologia, Hospital Geral de Santo António, Porto.
Singapore: Dr Chee Kheong Ooi, National Neuroscience Institute.
Spain: Dr Ana Rodriguez, Hospital del Mar, Barcelona; Dr Jaume Roquer, Hospital del Mar, Barcelona; Dr Francisco Lacruz, Servicio de Neurologia, Hospital de Navarra, Pamplona; Dr Pilar Piñero, Hospital Virgen del Rocio, Seville.
Sri Lanka: Dr Udaya Ranawaka, National Hospital of Sri Lanka, Kelaniya.
Sweden: Dr Ann Charlotte Laska, Karolinska Hospital, Stockholm; Dr Anders von Heijne, Department of Diagnostic Radiology, Danderyd Hospital, Stockholm; Dr H-G Hårdemark, Department of Neurology, Karolinska Hospital, Stockholm; Dr Judit Pap, Kärnsjukhuset, Röntgenkliniken, Sjukhuset i Lidköping.
The Netherlands: Dr Diederik W. Dippel, Department of Neurology, Erasmus MC, Rotterdam; Dr Jorrit Hoff, LUMC, Leiden; Dr Jaap Kappelle, Academisch Ziekenhuis Utrecht; Dr Jan Lodder, Department Neurology, University Hospital Maastricht; Dr Wil Hoefnagels, Ziekenhuis Zeeuws-Vlaanderen, Ziekenhuis Zeeuws-Vlaanderen, Oostburg.
Turkey: Dr Dilek Necioglu, Þiþli Etfal Education and Research Hospital, Istanbul; Dr Goksel Bakac, Bakirkoy State Hospital for Neulogical and Psychiatrical Diseases, Istanbul; Dr Nazire Afsar, Marmara University School of Medicine, Marmara University Hospital, Istanbul; Dr Sara Bahar, Istanbul Medical Faculty.
United Arab Emirates: Dr Miklos Szolics, Tawam Hospital, Al Ain.
UK: Dr Peter Keston, Aberdeen Royal Infirmary; Dr Margaret Wilkinson, Radiology Department, Wansbeck General Hospital, Ashington; Dr Michael Power, Elderly Care Department, Ulster Hospital, Belfast; Dr Stephen Todd, Department of Geriatric Medicine, Belfast City Hospital; Dr David Sandler, BHH&S NHS Trust, Birmingham; Dr Tom Heafield, QEH, Birmingham; Dr Shelley Renowden, Department of Neuroradiology, Frenchay Hospital, Bristol; Dr William Stevenson, X Ray Department, Burnley General Hospital; Prof Mark Wiles, Department of Neurology, University Hospital of Wales, Cardiff; Dr Nicholas Bradey, James Cook University Hospital, Darlington; Dr Sam Chakraverty, Ninewells Hospital and Medical School, Dundee; Dr Donna Edwards, Radiology Department, Hairmyres Hospital, East Kilbride, Lanarkshire; Dr Andrew Chancellor, Western General Hospital, Edinburgh; Dr Andrew Farrall, Department of Clinical Neurosciences, Western General Hospital, University of Edinburgh; Dr Donald Collie, Department of Clinical Neurosciences, Western General Hospital, Edinburgh; Dr Gillian Potter, Edinburgh Royal Infirmary; Dr John Murchison, Edinburgh Royal Infirmary; Prof Joanna Wardlaw, Department of Clinical Neurosciences, Western General Hospital, Edinburgh; Prof Martin Dennis, Department of Clinical Neurosciences, Western General Hospital, Edinburgh; Prof Peter Sandercock, Department of Clinical Neurosciences, Western General Hospital, Edinburgh; Dr Philip White, Department of Clinical Neurosciences, Western General Hospital, Edinburgh; Dr Richard Lindley, Department of Clinical Neurosciences, Western General Hospital, University of Edinburgh; Dr Simon Hart, Department of Clinical Neurosciences, Western General Hospital, Edinburgh; Dr Sarah Keir, Department of Clinical Neurosciences, Western General Hospital, Edinburgh; Dr Margaret-Ann Macleod, Falkirk & District Royal Infirmary, Falkirk; Dr Margaret Brown, Southport DGH, Formby; Dr Joseph Harbison, Department of Geriatric Medicine, Queen Elizabeth Hospital, Gateshead; Dr Kausik Chatterjee, Stroke Research Unit, University of Newcastle Upon Tyne, Gateshead; Dr Paul Mudd, Stroke Research Unit, Queen Elizabeth Hospital, Gateshead; Dr John Calder, Victoria Infirmary, Glasgow; Dr John Greene, Department Neurology, Southern General Hospital, Glasgow; Dr Sanjay Maroo, Royal Hospital For Sick Children, Glasgow; Dr Brian MacInnes, NHS Argyll & Clyde, Greenock; Dr Peter Wanklyn, Leeds General Infirmary; Dr Suzanne Dawson, Leicester General Hospital, University Hospitals of Leicester NHS Trust; Dr Sarah Jones, University Hospital Lewisham, Lewisham; Dr Richard White, Walton Centre for Neurology, Walton Centre for Neurology and Neurosurgery NHS Trust, Liverpool; Dr Scott Ramsay, St John’s Hospital, Livingston; Dr Alan Brooks, Homerton Hospital, London; Dr Andy Evans, King’s College Hospital, London; Dr Birns Jonathan, Department of Stroke Medicine, King’s College Hospital, London; Dr Dawn Saunders, Department of Neuroradiology, National Hospital, London; Dr Kalman Kafetz, Whipps Cross University Hospital, London; Mr Mitesh Davda, London; Dr Pankaj Sharma, Clinical Neurology Department, Hammersmith Hospitals, Imperial College, London; Dr Steven Connor, Department of Neuroradiology, Kings College Hospital, London; Dr Chandrasekaran Sivakumar, Hope Hospital, Manchester; Dr Alexander Dyker, University of Newcastle-upon-Tyne; Dr Eluzai Hakim, Department of Adult Medicine & Rehabilitation, St. Mary’s Hospital, Newport; Mrs Julie Deakin Jones, CT Scanning, Ormskirk District General Hospital, Southport & Ormskirk NHS Trust, Ormskirk; Dr Enrico Flossmann, Clinical Neurology, Radcliffe Infirmary, Oxford; Dr Qais Arafat, X Ray Department, Peterborough Hospital; Dr Patrick Trend, Royal Surrey County Hospital, Petworth; Dr William Mukonoweshuro, Department of Radiology, Derriford Hospital, Plymouth; Dr John Herbetko, Department of Radiology, Poole Hospital; Dr Jonathan Jones, Poole Hospital; Dr Susan Walkden, Poole Hospital; Dr David Jarrett, Queen Alexandra Hospital, Portsmouth; Dr John Whalley, University Hospital Aintree, Preston; Dr Ian Holland, Medical Director’s Office, Scarborough Hospital; Dr Alan Fletcher, Northern General Hospital, Sheffield Teaching Hospitals NHS Trust; Dr Christine Emma Burness, Royal Hallamshire Hospital, Sheffield; Dr Chris McDermott, Department Neurology, Royal Hallamshire Hospital, Sheffield; Dr Graham Venables, Department of Geriatrics, Royal Hallamshire Hospital, Sheffield; Dr Thomas Powell, Royal Hallamshire Hospital, Sheffield; Dr Catherine Monaghan, North Tees University Hospital, Shildon; Dr Helen Watson, Radiology Department, Royal Shrewsbury Hospital; Dr Kevin Gill, Radiology Department, Royal Shrewsbury Hospital; Dr Joseph Kwan, Southampton General Hospital, Southampton University Hospitals Trust; Dr Mary Gawne-Cain, Southampton University Hospitals Trust, Southampton University Hospitals Trust; Dr Andrew Johnston, Queen Elizabeth Hospital, Stafford; Dr Peter McDermott, Stirling Royal Infirmary; Dr Navaratnam Bhuvanedran, Geriatric Medicine Department, Bucknall Hospital, North Staffordshire Combined Health Care Trust, Stoke on Trent; Dr Raja Reddy, North Staffordshire Combined Health Care Trust, Stoke on Trent; Dr Tim Cassidy, University of Sunderland, Sunderland Royal Hospital; Dr Adrian Parnell, Good Hope Hospital NHS Trust, Sutton Coldfield; Dr Sunil Bhandari, Hull Royal Infirmary, Swanland; Dr Mushtaq Wani, Morriston Hospital, Swansea NHS Trust; Dr Gerry Saldanha, Department of Neuroradiology, King’s College Hospital, Pembury Hospital, Tunbridge Wells; Mr Matthew Dunn, Department of Emergency Medicine, Warwick Hospital; Dr Brian Mucci, North Cumbria Acute NHS Trust, North Cumbria Acute NHS Trust, Whitehaven; Dr Ahmed Abdul-Hamid; Dr Keith Muir.
USA: Dr Christopher Hancock, University of Miami, Miami, Fla; Dr David Smith; Dr Lucas Elijovich, New York University, New York; Dr Mary Kalafut; Dr Salvador Cruz-Flores.
We are extremely grateful to the many interested physicians who participated in the ACCESS study.
The individuals who read all 6 batches of scans are listed in the Appendix. We are also grateful to the organizations that provided financial support for the study.
Sources of Funding
This study was supported by Chest Heart Stroke Scotland (RES000/7); University of Edinburgh; DesAcc, Inc, Chicago, Ill; the Department of Neurology, University of Heidelberg, Faculty of Medicine, Mannheim (to Dr Mielke); and the Health Foundation (2268/1282). The study was designed, conducted, and analyzed and is reported independently of any of the funding sources.
↵*Members of the ACCESS group are listed in the Appendix.
- Received October 4, 2006.
- Revision received October 25, 2006.
- Accepted October 29, 2006.
Wardlaw JM, Farrall AJ. Diagnosis of stroke on neuroimaging. BMJ. 2004; 328: 655–656.
von Kummer R, Holle R, Gizyska U, Hofmann E, Jansen O, Petersen D, Schumacher M, Sartor K. Interobserver agreement in assessing early CT signs of middle cerebral artery infarction. Am J Neuroradiol. 1996; 17: 1743–1748.
Wardlaw JM, Sellar RJ. A simple practical classification of cerebral infarcts on CT and its interobserver reliability. Am J Neuroradiol. 1994; 15: 1933–1939.
von Kummer R. Effect of training in reading CT scans on patient selection for ECASS II. Neurology. 1998; 51: S50–S52.