In patients aged ≥ 65 years admitted to the ICU for TBI regardless of their initial GCS score, unfavourable functional outcomes assessed by a GOSE score < 5 were found in 67%, and only 7 patients (2%) fully recovered or had minor symptoms that did not affect daily life (GOSE score = 8). Factors associated with unfavourable outcomes were age, male sex, GCS score, ISS and use of osmotherapy during their course of disease. Although the proportion of surgical interventions other than neurosurgery was not collected in our study, the number of traumatic associated lesions and complications during ICU stay did not differ between patients with favourable vs. unfavourable outcomes, suggesting that the severity of TBI was the main cause of an unfavourable outcome.
It is well known that compared to younger patients, elderly patients who experience TBI have a poorer outcome [6, 19], but few studies have focused on older patients admitted to the ICU [4, 5, 20]. A meta-analysis reported 11 studies from 1986 to 2008 and outcomes in TBI patients ≥ 60 years old. Mortality in patients who had severe (GCS score ≤ 8; 10 studies), moderate (GCS score 9–12; 5 studies) and mild (GCS score 13–15; 6 studies) TBI was 79.3%, 42.5% and 10.7%, respectively, and unfavourable outcomes (GOS score 2–3) were 13.8%, 29.5% and 7.0%, respectively [21]. Nevertheless, this meta-analysis now included old studies (5 prior to 2000) with patients not always managed in the ICU, of various ages and with assessments of the outcome at various times. More specifically, in a cohort of 1366 patients from 3 neurosurgical ICUs in Italy, Stocchetti et al. showed that 6-month mortality and unfavourable outcomes (GOS score 1–3) increased with age, particularly for patients ≥ 60 years old [5]. In a prospective multicentre study, outcome at 1 year was studied in 100 TBI patients who had > 65 years and an initial GCS score of 3 or 4. ICU mortality was 76%, and at 6 months, only 9 patients had a favourable outcome (GOS score 1–3) [20]. Recently, a prospective registry-based cohort study performed in the state of Victoria, Australia, focused on outcomes at 6 months in patients ≥ 65 years admitted for TBI [4]. In the subgroup of 263 patients managed in the ICU, 12% were functional independent (GOSE score 5–8), 15% were dependent (GOSE score 2–4) and 73% had died at 6 months [4]. Our results are in agreement with these previous studies, emphasizing a high ICU and 3-month mortality and unfavourable functional outcome in TBI older patients. As reported by Maiden et al., we found that age was independently associated with unfavourable outcomes, and in our study, a threshold of ≥ 77 years indicated the worst outcome.
Cerebral injuries had a major impact on outcome, although the severity of associated injuries, assessed by the AIS, and the cumulative number of body injuries did not seem to affect outcome. Head AIS severity was previously associated with low odds of functional recovery in univariate analysis, and an ISS ≥ 25 was shown to be independently associated with an unfavourable outcome [4]. In our study, ISS was also an independent risk factor for an unfavourable outcome, and the best threshold was also ≥ 25. Interestingly, in our study, the contribution of the TBI to the ISS value was high, particularly in patients who had an unfavourable outcome, confirming the high value of the severity of head trauma on outcome. From this perspective, it is not surprising that we found that the initial GCS, reflecting the severity of TBI, was an independent risk factor for unfavourable outcomes, with a threshold ≤ 9 being the most relevant to predict unfavourable outcomes. Usually, a GCS score > 8 is considered moderate. Nevertheless, the impact of a TBI on outcome in elderly individuals may lead us to question the relevance of a higher initial GCS score threshold in this population. Thus, the presence of several comorbidities, psychiatric history and/or frailty may amplify the consequences of a TBI that would be considered less severe in a younger patient. In this context, it was shown that compared with that in a younger population, even a moderate head injury resulted in increased mortality in subjects over 64 years of age [9, 22]. Nevertheless, the initial GCS score alone did not allow good discrimination of the outcomes. Indeed, in our cohort, 47.5% of the patients who had a GCS score ≤ 8 also had a GOSE score ≥ 5 at 3 months.
We also found that male sex was independently associated with unfavourable outcomes. Head injuries are more frequent in men, but the proportion is balanced and/or reversed in older patients [1, 5]. However, the impact of sex on outcome provides various results [23]. In a review including 156 studies, the crude results reported a worse outcome in women, but analysis focused on moderate and severe TBIs (GCS score 3–12) showed a higher proportion of good outcomes in women compared to men [23]. Similarly, outcomes were better in women in large studies (> 10,000 patients) than in smaller studies [23]. The extent to which age may also affect the impact of sex on outcome also produces divergent results [24]. In analysing the sex-age interaction on outcome, comorbidities, degree of frailty, and severity of associated injuries need to be taken into account [24]. In our study the prevalence of unfavourable outcomes in men and women was 72.4% and 57.3% respectively while except for alcohol abuse comorbidities and severity of TBI did not differ.
Finally, the use of osmotherapy was found to be associated with unfavourable outcomes. The administration of osmotherapy probably reflects the detection of intracranial hypertension, but our study does not allow us to specify what motivated this prescription. Indeed, except for caricatured situations such as pupillary dilation, trans-cranial Doppler is a technique increasingly used in prehospital, emergency room and/or intensive care units to detect intracranial hypertension and can lead to the administration of osmotherapy without the use of an intracranial pressure sensor [25]. In our study the presence of intracranial hypertension measured by a pressure transducer did not differ between the two groups but we have too much missing data on the use of monitoring to make relevant conclusions.
Our study has some limitations that must be underlined. First, as with all prospective cohorts with retrospective analysis, some data were missing or incomplete. However, we excluded patients with missing data on the primary outcome and reported the number of missing data for all other variables. Second, all patients were managed in ICUs experienced in the care of severe trauma patients, including head trauma patients. Our results can only be applied to these patient populations and may be difficult to extrapolate to other care systems, although our results are in agreement with other studies performed in the same population. Third, the threshold of 65 years of age to define an elderly population may be debatable. However, given the unfavourable short- and long-term prognosis of severe trauma patients, particularly those suffering from head trauma, an age of 65 years is usually retained [4, 5]. Fourth, we collected data on the outcome at three months, and it cannot be excluded that an improvement can be observed beyond this date. Indeed, it was proposed that functional status should preferably be assessed from 6 months or beyond. It was also shown that the best improvement occurred between 2 weeks and 3 months, with a gain of approximately 1 point in the GOSE, and, to a lesser extent, between 6 and 12 months, with a gain of 0.2 points [26]. However in 263 TBI patients aged > 65 years the functional status did not significantly improve after 6 months [4]. Finally, in a recent multicentre study evaluating the effect of continuous hypertonic saline vs. standard care on outcome at 6 months, it appeared that the prevalence of a good neurologic outcome (GOSE score 6–8) did not significantly differ between 3 and 6 months (115/351 (32.8%) vs. 122/359 (34.0%)) [27]. Fourth, we did not evaluate frailty in our patients. Frailty is now recognized as an important determinant in outcome, notably in elderly ICU patients [28, 29]. A recent study evaluated a specific frailty score (CENTER-TBI frailty index) from a cohort of 2993 head injury patients. Among patients admitted to the ICU (1742 patients), the impact of frailty on adverse outcomes (GOSE score ≤ 4 at 6 months) was not as strong in the ICU as it was in a conventional hospital ward (cumulative OR [95% CI] = 1.02 [1.01–1.03] vs. 1.04 [1.03–1.06], p < 0.001), underlining the importance of the severity of head injury rather than frailty alone [29]. Fifth, the high frequency of withdrawal or withholding of life-sustaining therapies in ICU may have affected prognosis. In a prospective study performed in Europe and Israel, the majority of deaths in patients who experienced TBI was related to withdrawal of care (≈ 90% in western and ≈ 96% in northern Europe) [30]. In our study, it was difficult to draw clear explanations for these findings, as neither the reasons for withdrawal or withholding of life-sustaining therapies (WWLST) nor the timing of WWLST were collected, which may cause uncertainty when applied latter in the course of disease [30]. Nevertheless, the fact that few patients were in a vegetative state at 3 months may suggest that futility was the main reason for WWLST. Finally, according to the design of our study, we cannot exclude the possibility that other clinical and/or biological variables of potential interest were not collected.