This study demonstrated a strong, independent relationship between ONSD on the initial brain CT scan and the mortality rate among severe TBI patients admitted to a neurosurgical ICU.
Mean ONSD in this study was much higher than that previously reported under various measuring conditions, even in patients with severe TBI. However, these results are in accordance with previous studies measuring ONSD on CT. The first study measured ONSD in 100 healthy volunteers [
20] and found a mean ONSD of 4.4 mm with a 95% CI of 3.2 to 5.6. These values in healthy volunteers are higher than those proposed for the detection of elevated ICP by sonography. More recently, ONSD was evaluated on CT in normal-tension glaucoma patients compared to that of healthy volunteers [
21] and revealed a mean ONSD of 8.0 ± 1.0 in normal-tension glaucoma patients and 6.2 ± 0.9 in healthy volunteers. Experimentally, in an optic nerve model obtained from cadaveric donors without neurological injury, a clear relationship was demonstrated between increased intracranial pressure and increased ONSD, but with ONSD values at 6.7 mm with higher pressures [
22]. However, this study was purely experimental and was not conducted on injured patients under pathophysiological conditions. ONSD measurements have been more frequently reported with ultrasound. ONSD values greater than 7.0 mm have been reported in TBI patients with intracranial hypertension [
23]. However, in a retrospective study, ONSD ≥ 5.0 mm was proposed to detect elevated ICP [
24]. The inter-observer and intra-observer variations of ultrasound measurement of ONSD were recently calculated to be 0.5 mm and 0.2 mm, respectively [
25]. These results are very similar to those reported in the present study based on CT measurement of ONSD. MRI has also been proposed to measure ONSD with a lower range of values than those reported in this study [
18]. However, the limited accuracy of MRI with thicker brain slices may underestimate the real value of ONSD. With MRI measurement, errors have also been reported to be 1.3 to 1.5 mm [
26]. One MRI study on ONSD in TBI patients with invasive ICP has been published [
18]. The ONSD was significantly greater in TBI patients with raised ICP (> 20 mmHg; 6.31 +/- 0.50 mm) than in those with ICP of 20 mmHg or less (5.29 +/- 0.48 mm) or in healthy volunteers (5.08 +/- 0.52 mm). A study comparing MRI and B sonography in healthy subjects [
27] showed that sonography underestimated ONSD: 5.72 mm (5.51 to 5.93) for MRI vs. 4.08 (3.4 to 4.3) for sonography. A recent evaluation of ONSD measured by MRI in astronauts exposed to microgravity reported ONSD values of 6.5 ± 1.0 mm [
28]. The ONSD values measured by CT and MRI are higher than those reported with sonography. The ONSD measurements reported in the present study appear to be accurate in view of the low intra- and inter-observer variability determined prior to the study. A good correlation has been reported between CT scan and ultrasonography in healthy subjects [
29]. However, the previous study overestimated ONSD by more than 10% with CT compared to sonography [
29]. The mean ONSD in healthy subjects has been reported to be 5.5 mm [
30], very close to the limit for elevated ICP reported for sonography [
31], emphasising the differences in the measurements between these two methods.
Experimentally, Hansen
et al. showed a correlation between ONSD and ICP and ICP variations [
6,
9]. Other authors have used bedside transorbital sonography to assess ONSD in hydrocephalic patients [
32]. Many researchers have investigated the significance of this parameter as a means of detecting intracranial hypertension [
10,
17] in patients with TBI [
8,
33], thereby allowing rapid triage in the emergency unit [
34]. In the emergency department, an US evaluation of ONSD ≥ 5.0 mm was associated with CT findings of elevated ICP ([
35]. A link between MRI-determined ONSD and elevated ICP has also been reported [
18,
36]. However, to the best of our knowledge, no study has yet focused on brain CT scan that is routinely performed in severe TBI patients. ONSD is much easier to measure on CT scan than with sonography due to the good reproducibility of CT and the lack of a learning curve. Furthermore, with the development of telemedicine, tertiary reference centres now have access to the initial CT scan [
37]. ONSD measurement may facilitate transfer decisions. The relationship between intracranial hypertension and prognosis has been widely demonstrated [
38‐
41]. Furthermore, ICP monitoring
per se has never been found to be associated with improved outcomes in severe TBI patients [
42] compared to clinical and radiological evaluation. When combined with other parameters, the initial measurement of ONSD may be clinically useful in the early diagnosis setting and to rapidly initiate aggressive treatment of suspected elevated ICP.
Interpretation of our data is subject to several limitations. First, the sample size may be considered to be small. We nevertheless included all consecutive patients over the study period. Some patients with less severe forms of TBI may have been hospitalized in primary care hospital ICUs without referral to our university hospital after CT imaging. In contrast, high-severity TBI patients rapidly progressing to brain death would not have been considered for transfer due to local multi-organ donation procedures. Although this aspect clearly constitutes a limitation, the measurement technique is very easy to implement and depends more on the software tools than the physician's ability. Furthermore, all brain CT scans are now performed in millimetre slices, enabling easy measurement of ONSD. The inter- and intra-observer variability of ONSD measurement was in accordance with the results reported in other studies, with very good reliability. Finally, the rate of ICP monitoring was very low in these severe TBI patients. This is another limitation of the study despite recent studies showing similar outcomes with ICP management versus clinical and radiological evaluation [
42]. Brain CT scan ONSD measurements cannot be considered to be a monitoring tool for severe TBI patients. This study was not designed to evaluate this point, but only to consider available data before ICU admission. ONSD monitoring is much easier and less invasive by bedside sonography.