Study setting and population
In early 2007, S100B was introduced into clinical practice within the existing SNC guidelines to create new local management routines (Figure 1). The addition of S100B was applied to a group of patients, typically considered as intermediate risk for intracranial complication, where CT is normally recommended. We set the time interval for S100B sampling at 3 hours post injury, reflecting the evidence available in 2007 [23]. Also, evidence for S100B use in children at this time was relatively weak and the new guidelines were therefore used only in adults.
After a 6 months adjustment period, we undertook a prospective cohort validation study in Halmstad Regional hospital, Sweden, from November 2007 to May 2011, to evaluate the adapted guidelines explained above. Our hospital is a level II trauma centre with 24-hour emergency care, anaesthesiology, radiology, surgery and intensive care.
We consecutively enrolled all adult patients with MHI and S100B sampling. Initial inclusion criteria were therefore analogous to the MHI group in the SNC guidelines; adult patients with acute trauma to the head with GCS 14–15 during examination and loss of consciousness < 5 minutes and/or amnesia, with the addition of the S100B sample. Patients with anti-platelet agents (such as aspirin or clopidogrel) were included. Exclusion criteria were age less than 18 years, non-Swedish citizens (difficult to follow up), neurological deficits, additional risk factors from the SNC guidelines (therapeutic anticoagulation or haemophilia, clinical signs of depressed skull fracture or skull base fracture, posttraumatic seizures, shunt-treated hydrocephalus and multiple injuries) and patients where serum sampling for S100B was done more than 3 hours post-injury.
Our goal was to include 500 patients in the study, based upon consensus in the study group when considering the aim of the study. A sample size calculation was not performed.
The study was conducted in accordance to the Helsinki Declaration and approved by the Lund regional ethical committee, Lund, Sweden (reference number 19/2007). Since the study did not involve any change in patient management and based upon clinical practice, informed consent was not necessary and the ethics committee concurred with this decision.
Blood sampling and biochemical analysis
A 5ml blood sample was drawn from each patient’s cubital vein in the ED. Samples were analysed with the fully automated Elecsys® S100 (Roche AB) at the Clinical Chemistry Department of Halmstad Regional hospital, Sweden. Roche AB report a range between 0.005 μg/L and 39 μg/L and a within-series coefficient of variance of <2.1%. Based on the available evidence at this time, we chose a cut-off level for normal levels of less than 0.10μg/L and a window of sampling of 3 hours from the time of the accident [19, 23]. Lab results were available to treating physicians within 1 hour after sampling.
CT examinations
CT scans were performed with a GE VCT Ligthspeed 64 multislice detector with a 0,625/0,625mm, 0,5 seconds rotation time and pitch of 0,531:1. 10mm thick slices were used as part of the standard CT protocol for these patients. CT scans are always analysed by a board certified radiologist and confirmed by a consultant radiologist. Since S100B was used clinically, radiologists were not blinded to S100B results. A CT scan was considered positive if any signs of cranial (skull fracture) or intracranial pathology (hematoma, air or contusion) were present.
Standardized assessment of patients
Supervised interns and surgical residents from the ED of the Halmstad Regional Hospital assessed patients. These physicians underwent several educational sessions on evaluating patients with MHI using the new guidelines. Physicians were instructed to follow the new guidelines for all non-severe head injury patients even though deferral from these due to clinical judgement was allowed.
Data registration and follow-up
Details of how patients were managed, including patient characteristics, type of injury, patient history, medications, clinical examination results, CT results, admission type and duration were documented in an Excel spreadsheet.
Patients were asked to answer a questionnaire sent by mail 3 months after the injury, which was repeated if no answer was received. For patients who did not return the questionnaire after these attempts, a blinded assessor conducted the questionnaire via telephone. Included in this questionnaire were questions that would identify a significant intracranial complication [7]. In cases where patients could not be reached by mail or telephone, medical records and national mortality databases were consulted for evidence of complications and/or death. Considering the rigid and transparent organisation of the health care system in Sweden, these methods would identify all patients with significant (enough to result in new neuroimaging, neurosurgery or death) intracranial complications.
Our outcome endpoint for the study was significant intracranial complication, which was defined as either a traumatic complication on emergency CT or, via follow-up, new neuroimaging showing traumatic intracranial complication or neurosurgery and/or death due to an intracranial complication.
Sensitivity, specificity, positive and negative predictive values were estimated from cross tabulation between S100B and significant intracranial complications and reported with corresponding 95% confidence intervals. Values are reported to two significant figures.