Clinical predictors of abnormal head computed tomography scan in patients who are conscious after head injury

Introduction

Patients often present with minor head injury in trauma and emergency departments. The brain computed tomography (CT) scan is a investigation of choice to diagnose intracranial lesions in such patients.[1] Fifteen percent of patients with mild traumatic brain injury (mTBI) have intracranial lesions in CT scan, among them <1% require any neurosurgical intervention.[2] However, the indications for a CT scan is debatable if the patient has Glasgow Coma Scale (GCS) 15 at the time of presentation. In various guidelines, the indication of CT scan varies on a number of clinical predictors.[345] In resource-constrained settings, indication for CT scan for patients with GCS 15 needs to be better defined.[67] Various clinical predictors have been used in earlier studies to determine the indications of CT scan in patients with mTBI, but most of them have not been validated in external settings.[8910111213] The aim of the present study was to determine presence of any clinical variable that serves as a risk factor for abnormal head CT in a patient with head injury with GCS 15, which may help in formulating guidelines for the management of TBI in our country and also help in the prioritization of resources in the community.

Materials and Methods

This study was conducted at a Level I neurotrauma center. Patients with head injury were evaluated by neurosurgery residents and data were entered in a structured “head injury proforma,” which consists of comprehensive clinical and head CT scan findings. The data were verified by a qualified neurosurgeon on duty. Data of all patients with head injury with GCS score 15, over a period of 2½ months were prospectively studied. CT scan findings were confirmed independently from images archived in the picture archival communication system of the institute by the senior author (DS). We used 15 point GCS for all patients.[14] To identify the predictors of our dependent variable “CT scan finding,” it was categorized as normal or abnormal. The abnormal CT scan was defined as any finding ascribed to head trauma including skull fracture, cerebral edema, epidural hematoma, subdural hematoma, cerebral contusion, intracerebral hematoma, intraventricular hemorrhage, and subarachnoid hemorrhage. The person reporting head CT scan was blinded to clinical findings. It was independently verified by blinding to GCS by one of the authors. The variables used for the analysis were age in years (<40, 41–60, >60), gender, mode of injury (road traffic accident (RTA), fall, assault, and other), duration since injury in hours (<6, 6–12, >12), and following symptoms: loss of consciousness (LOC), headache, vomiting, ear/nose bleed, and seizures.

Results

A total of 1,629 patients with TBI were evaluated during the study period of 2½ months, and 453 patients with GCS 15 at the time of presentation to casualty were identified. The mean age of the patients was 32.84 ± 17.46 years, and male to female ratio was 4:1. RTAs were the major cause of injury accounting for 53%, followed by assault (31%), falls (13.2%), and others (2.8%). Three hundred and twenty-seven (72.2%) patients had at least one of the following four symptoms: LOC, vomiting, ear/nose bleeding, or seizure following the head injury. Transient LOC lasting <30 min was most common symptom reported by 43% patients followed by vomiting (36%), ear/nose bleed (18.5%), and seizure (4.4%). The most common combination of symptoms was LOC and vomiting. The CT was abnormal in 43% of patients. The demographic and clinical details of patients with normal and abnormal CT scan are compared in Table 1. Details of abnormal CT scan findings are shown in Table 2. Twenty-two (5%) patients underwent surgery for the traumatic lesion. Significantly more number of patients with age <40 years had a normal CT scan. Significantly more number of patients with duration of <6 h after injury had a normal CT scan. The presence of any symptom had an odds ratio (OR) of 2.36 (95% confidence interval [CI] 1.52, 3.71), P < 0.001 for a positive CT scan, which was significant. Among individual symptoms, only vomiting was significantly associated with abnormal CT scan, OR 1.89 (95% CI 1.23, 2.80), P < 0.001. Logistic regression analysis applied on the test yielded the following equation:

Table 1 Comparison of variables between normal and abnormal computed tomography scan

Table 2 Abnormal computed tomography (CT) scan findings: n=195 (43.05%)

P = 1/(1 + e^{− logit(p)})

P = Probability of abnormal head CT scan

logit(p) = −1.258+ (0.247 × duration 6–12 h or 0.797 × duration > 12 h) +0.618 × any symptom + 0.321 × vomiting + (0.519 × age [41–60] or 0.731 × age (>60)).

Table 3 shows the accuracy of this equation to predict abnormal CT scan in patients with GCS 15 after head injury. The sensitivity is 63.72% (58.16–69.02%), specificity 58.82% (50.07–67.19%), positive predictive value 78.29% (72.76–83.17%), and negative predictive value 41.03% (34.05–48.28%) for prediction of abnormal head CT scan.

Table 3 Accuracy of prediction of abnormal CT scan

Discussion

Head injuries are commonly managed in the trauma and emergency departments, among them 70–80% are mild in nature.[15] Many head injured patients are conscious on arrival to casualty and do not have a neurological deficit. However, such a patient constitutes a potential problem for neurosurgeons because a patient with an apparent minor head injury can rarely develop an intracranial hematoma, deteriorate, and die. There is controversy regarding the policy for hospital admission and evaluation with CT scan for these patients. A CT scan is desirable for patients with minor head injury as it is useful for detection of a clinically significant intracranial lesion, prognostication, and decision for discharge. To obviate unnecessary CT scans, many guidelines are available for indication of CT scan for minor head injury in adults.[16] Among these, Canadian CT Head Rule is the most widely validated rule, with a sensitivity of 99–100% and a specificity of 48–77%.[161718] Other rules differ considerably in population, predictors, outcomes, methodologic quality, and performance. Many of them are not validated in a separate population, and their impact on practice has not been assessed. In our patient population, only four symptoms were present: vomiting, LOC, ear/nose bleeding, and seizures. We did not include scalp injury as a variable because it has little diagnostic value.[18] We do not perform coagulation profile for patients with minor head injury at our institute; hence, we did not include this variable. Posttraumatic amnesia (PTA) is also an important predictor of abnormal CT scan.[16] The timing of resolution of PTA is difficult to assess in the emergency department; hence, we did not include this variable as well. We included all CT scan findings attributed to trauma including skull fracture as inclusion criteria for an abnormal (positive) CT scan. We found that presence of any of the symptoms attributed to head injury such as LOC, vomiting, ear/nose bleeding, or seizures predicted an abnormal CT scan. Particularly vomiting after head injury was significantly associated with abnormal CT scan. Although the sensitivity of our prediction model was low, the specificity was higher than most of the available prediction rules. We had some limitations with our study as can be seen with the number of clinical predictors used in the study. Further, this study does not take children into a different group, very often children have different clinical predictor rules for CT scan. We did not define separate significant intracranial injury, which probably has more bearing on management and prognosis. However, the presence of any abnormality on CT scan requires neurological observation. Our study needs to be conducted in a larger population and needs to be validated in external settings. However, in spite of these limitations, our study answers one important question for neurosurgeons. What clinical predictors can be used to address the question that “should a patient be shifted to higher center for head CT scan, if the treating center does not have CT scan facility?” Yes, if the patient is older than 40 years of age and has any of following symptoms: vomiting, LOC, ear/nose bleeding, and seizures after head injury he/she should undergo a head CT though not all patients have an abnormal scan.

Conclusion

We attempted to develop a clinical prediction rule for abnormal CT scan in a patient who is conscious at the time of presentation to casualty after minor head injury. Our prediction rule had low sensitivity but better specificity than the available prediction rules. Our prediction rule can be used to develop guidelines for indication of CT scan in patients with head injury in India if it can be externally validated at various other centers on a larger scale.