As demonstrated in this study, SDH patients represent a highly heterogenous group. Despite possessing a range of presentation types and varying degrees of neurologic injury, SDH outcomes are often reported almost exclusively by the age of the blood products at the time of presentation. Although this method has its merits, this nomenclature has had unintended, adverse effects in the literature. Insufficient stratification of these subpopulation groups has contributed to conflicting conclusions across studies, making it is difficult to develop a consensus of opinion or treatment guidelines. This is particularly relevant for the advanced geriatric population with acute SDHs where a call for guidelines regarding the role for surgery and possible rationing of care has been made [
9,
13]. With an anticipated surge in both the age and number of patients with SDHs requiring evacuation, the need for more accurate and reliable outcome reporting becomes greater each year [
16].
Through incorporating a schematic that stratified groups by combining HVI and acute blood products, we were able to delineate upon two distinct patient populations that would have traditionally been shrouded under a single, acute SDH designation. For instance, those with aSDH
HVI arrived at the hospital more often by ambulance (96% vs. 45%,
p < 0.001), had a significantly longer hospital stay (31 vs. 18 days,
p = 0.004), over double the rate of poor outcomes (50% vs. 23%,
p = 0.001), and over triple the mortality (25% vs. 8%,
p = 0.004) compared to the aSDH
wo group. Outcomes of the aSDH
wo also remained distinct from with mSDH group, demonstrating a difference in each metric listed above as well (Fig.
3). Among the acute SDH population, HVI alone carried over a 500% increased risk for poor outcomes (OR 6.85,
p < 0.001) controlling for patient gender, midline shift, and anticoagulation use. The association of female gender, midline shift, and anticoagulation use corroborate prior studies that also have reported these as risk factors for poor outcomes [
17‐
19].
Heterogeneity of Subdural Hematoma Outcome reporting
Until now, no prior studies have proposed a method to address the heterogeneity of SDH outcomes reported in the literature, to the best of our knowledge. A minority have attempted to stratify SDH outcomes based on traumatic injury type, presenting neurological condition, or GCS score [
20‐
23]. While some SDH scoring systems exist, they have been aimed primarily at prognosticating and guiding treatment decision-making [
24‐
26]. Consequently, there exists a need for a more specific and standardized categorization technique for SDHs beyond that of blood-age chronicity [
27].
Surgical intervention of SDH has become controversial due to concerns for efficacy in the face of alarmingly high poor outcomes and mortality in the advanced geriatric population [
13,
28]. Benedetto et al. (2017) reported a 6-month mortality rate of 67.2% in a series of 67 patients over 70 years with acute SDHs surgically evacuated [
9]. Petridis et al. (2009) reported an inpatient mortality of 53.8% in a series of patients 65 or older [
29]. For chronic SDHs, Whitehouse et al. (2016) reported 15 times the rate of poor outcomes of mortality for inpatient death for those 75 or older [
30]. However, other studies reporting nearly half or less the rate of poor outcomes also exist around the same time period [
15]. Won et al. (2017) found an inpatient mortality of only 28% in patients over 80 with aSDHs [
10]. Younsi et al. (2021) reported an inpatient mortality of 33% for aSDHs [
14]. Younger age groups have not been immune to these wide variations either. Lavrador et al. (2018) reported poor functional outcomes in 58% (GOS 1–3, 40/69) of those with aSDH while a series by Ryan et al. (2012) found 88% (184/210) of patients had good outcomes with a GCS score 13–15 at the time of discharge in the aSDH evacuated group [
12,
31].
Causes of Subdural Hematoma Outcome Heterogeneity
Though it is difficult to say with absolute certainty, there are likely numerous factors contributing to these differences. A medical center’s surrounding population, location within a city, and the standards of practice can all be contributors. Relatively small sample sizes on this topic has also been cited as a potential cause [
12]. In a large systematic review, Manivanne et al. (2021) cited differences in functional outcome metrics (i.e., GOS, GCS, Markslwalder scale, mRS) and follow-up time as possible contributors for heterogeneity of the pooled data from the literature [
13].
An important factor at the crux of these differences is the proportion of patients with severe TBI and those with more isolated SDHs in the acute SDH population. As shown here, the proportion of patients with HVI aSDHs compared to those without has the potential to significantly affect outcomes. Albeit less often, SDHs can occur absent of a TBI event. Rapid acceleration-deacceleration movements with or without head strike can potentially still lead to shearing of bridging veins. Spontaneous SDHs can occur in patients on anticoagulation as well those with intracranial hypotension [
32‐
35].
Among those who experience a TBI of extreme severity, the SDH is often only one of many sources of neurological injury. These may include diffuse axonal injury (DAI), brain contusions, cerebral edema, blood brain barrier disruption, dysautoregulation of cerebral blood flow, changes in cellular metabolism, and surge of neurotransmitter release at neurotoxic levels [
36‐
40]. Prior efforts to stratify patients by neurological exam at the time of presentation or TBI severity but this is not well-standardized or commonly performed [
22,
41].
Devising an effective stratification system
In devising a stratification system that incorporated this information, we found designating those with a HVI injury in the acute SDH population (Table
1) effective. It provided sufficient stratification between patient populations to realize differences spanning several aspects of patients’ hospital course. Though HVI’s also occurred in those with subacute and chronic SDHs, the presence of older blood products in it of itself selected for patients with less severe or without TBIs. This is supported by HVI demonstrating no statistical association with patient functional outcomes (
p = 0.24) or mortality (
p = 1) in the mSDH group. Additionally, no differences in presentation types, GCS on arrival, or other outcome metrics in those with subacute and chronic SDHs were found. With the exception of those in the chronic group having greater mean age (74 vs. 65 years,
p = 0.01), these groups were very similar including the use of burr hole craniostomies versus craniotomies (20% chronic vs. 14% subacute,
p = 0.12). Consequently, the merging of these blood ages as a single category avoided redundancy as it pertained to general disease course for those being surgically treated. This coincides and supports articles in the literature that have categorized both subacute and chronic SDHs together as “chronic” [
27,
42,
43].
Another important feature in this design was keeping the stratification categories simple, with minimal group numbers. Defining a high-velocity mechanism of injury is still relatively intuitive to assign and easily retrievable in the history-of-present-illness. In maintaining a 3-tiered system that keeps blood age as a distinguishing feature, its adaptation is also a relatively minor transition from convention. Granted, there are likely other valuable categorizing features such as concussion grade, specific trauma event, or presenting GCS score. Yet these come at the expense of adding more categories and potentially requiring the retrieval of additional data points. Additional categories stand to lower the denominator across studies and could exacerbate the effect of exaggerated outcomes due to smaller sample sizes [
44]. Lastly, concomitant injuries are often sustained with motor-vehicle accidents, pedestrian versus motor vehicle, and other high kinetic energy mechanism of trauma [
45,
46]. By stratifying by HVI in the aSDH patient population, this likely better selects for those who sustained these other injuries rather than isolated GCS scores, which can also have a significant effect on outcomes [
47].
In the ad hoc analysis, aSDH
HVI demonstrated a greater odds ratio (OR 6.8 vs. 3.7) and predictive value for poor outcomes at discharge (AUROC 0.74 vs. 0.71) than poor GCS upon arrival. However, comparing these variables is challenging since they are not independent from one another, and interpretation may be limited. More commonly used statistical methods such as Chi2 or McNamer test cannot be used due to a failure to meet the required assumption for mutual independence [
48,
49]. Although demonstrating a correlation with poor outcomes is not the salient metric for evaluating the utility of a stratification system (as listed for the reasons above), these findings do suggest aSDH
HVI may provide a better representation of the underlying pathological changes seen in patients with different types of aSDHs. For instance, seizures, post-concussive symptoms, and SDH mass effect may have caused patients to present with a poor GCS score but still experience significant recovery following evacuation. Conversely, irreversible damage associated with DAI, brain contusions, and other sources of neuron injury that may have occurred after a HVI injury could have hindered as rapid or robust of a recovery.
Limitations
Study limitations include the retrospective study design. The generalizability of our institutional findings may be biased by the hospital location and referral pattern. Future investigation to further validate the utility of this proposed stratification system is needed. It would have been informative to compare the rates of DAI between the HVI and non-HVI patients with aSDHs, but this was not possible as MRIs were routinely performed in this patient population. Additionally, only surgical patients were included in this analysis. It is unclear the utility of this stratification system for non-surgical candidates and warrants further investigation.