The present study describes occurrence of TICH in TBI patients with ground-level falls. A large cohort of 1938 patients with 347 patients on APT and 483 patients on ACT was analyzed. The results showed that treatment with APT was independently associated with TICH. However, the same association was not found between TICH and ACT. The results indicate that the risk of TICH in patients treated with ACT could be as low as for patients not taking any ACT or APT. Furthermore, characteristics generally associated with elevated risk of TICH (e.g., vomiting and loss of consciousness, etc.) were not associated with TICH after ground-level falls in the present analysis.
The significantly higher rate of TICH in APT-patients compared to TICH-rate in ACT-patients was contradicted by the overlapping RR 95% CIs. This could be due to insufficient sample size. To explore if there was an actual association between TICH-incidence based on ACT or APT, a regression analysis was performed. It was prudent to assume that ACT and/or APT could be surrogate markers for other potential risk factors such as frailty and high age or that they were just confounders that covaried with important symptoms and signs such as loss of consciousness, amnesia, etc. However, regression analysis showed that treatment with APT entailed elevated odds for TICH, and that this odds was independent from age and comorbidity quantified by the aa-CCI. Furthermore, treatment with ACT had no association with outcome of intracranial hemorrhage. Therefore, it is feasible that there is an actual difference in TICH-incidence between patients treated with ACT and APT and that a larger sample size would render non-overlapping CIs for the RR. The higher risk for TICH in APT-patients is not reflected in all the current guidelines and could potentially mean that some TICHS could be missed [10,11,12, 21].
Signs, symptoms, and items from the patient history such as post-TBI vomiting, loss of consciousness, amnesia, and seizures are traditionally regarded as important when risk-stratifying patients with TBI [10, 12, 21]. However, only amnesia was associated with TICH in the final regression model. It is difficult to know from the present study how much the sample size affects this analysis, but it can be assumed that these factors are not as important when risk-stratifying patients with ground-level trauma. A possible explanation for this could be that the patients that sustain ground-level falls, because of high age and frailty, are more sensitive to this relatively small trauma energy and thus more prone to develop TICH. However, the trauma energy might not be high enough to cause a concussion and subsequent concussion symptoms such as vomiting and loss of consciousness. These symptoms can also appear because of a TICH, but in this scenario, they typically appear hours after the trauma as meningism arises and the intracranial pressure rises and might not be present when the patient presents to the ED. Consequently, a guideline recommendation based on these features might erroneously risk-stratify some patients with ground-level falls, because the features could be absent, even if the patient has developed a TICH.
The median age of patients in this study is higher than ED cohorts that include TBI patients with all trauma levels [8, 16, 31]. The reason for this could be that ground-level falls that lead to a concussion in younger individuals are less common and do not cause injuries severe enough to prompt an ED visit. Despite this age difference, the rate of TICH in the entire cohort was around the same level as most other ED cohort studies . Age of 60–65 years is an absolute CT-indication in most of the large guidelines such as CCHR, NOC and NICE, whereas the SNC guidelines recommend a CT to all patients ≥ 65 years with APT [10,11,12, 21]. The SNC recommendation is more in line with the occurrence of TICH after ground-level falls, at least based on the results of the present study. A mandatory CT of TBI patients over 60 years might not be necessary and study of risk stratification based on APT among other features could theoretically help reduce the number of CTs.
It is also possible that a study looking at sub-classes or combinations of APT/ACT would indicate an elevated risk of certain pharmaceuticals (e.g., dual APT or some types of DOACs). The distribution of TICH in the present study within the two drug classes and in combination of APT and ACT could indicate this, but the numbers were deemed too small for further statistical analysis. Fakhry et al.  found that the incidence of TICH was not increased because of ASA single therapy . Moreover, van den Brand et al.  (meta-analysis of 20, 247 patients) concluded the same in patients with mild TBI (GCS 14–15 upon index visit) . However, both studies found increased risk in patients on aspirin–clopidogrel combination therapy [23, 32]. The conclusions of these two studies differ from the present study, but high-level evidence is still lacking, and to remedy this, we recommend prospective studies.
The present study’s relatively low incidence of TICH in ACT-patients has been reported by several other studies in the past years [1, 15,16,17,18,19,20]. Even though one-third of the ACT-patients in the present study were on VKAs and not DOACs, the TICH-incidence did not differ significantly from the TICH-incidence in the cohort of patients with ground-level traumas not on ACT or APT. Given the state of current research where VKA has shown higher TICH-incidence compared to DOAC, it is possible that the TICH-incidence would be even lower in this subgroup in a population where VKAs were less frequent [1, 15, 16]. If the RR of 1.78 for APT-patients can be properly demonstrated in a study with sufficient sample size, there is a substantial TICH-hazard for APT-patients. Furthermore, the increasing evidence showing that DOAC is safer than VKA is an indication that guidelines handling these patient categories differently are needed. A recent meta-analysis concluded that a selective use of head-CT in TBI patients with DOAC instead of a mandatory CT might be feasible .
The most important limitation of the present study is the retrospective data collection. This can lead to information bias both when interpreting medical records and handling missing data. Irrespective of counter-measures, this prevents us from drawing more than cautious conclusions from this study. The effect of the measures we took to reduce interpretation bias (e.g., following guidelines for retrospective reviews in general and the pro forma document in particular) was evident in the Cohen’s kappa coefficient analysis. It showed that the concordance between two separate reviewers was acceptable. Hence, data were collected in a fairly reproducible manner and this can be construed as increased data validity in the retrospective setting.
The pragmatic approach to missing data was discussed in detail in the research group prior to information gathering and deemed the best solution. Because it is not based on statistical methods, it bares the drawback of not giving any reliability measurement (e.g., confidence interval). Because of this, it is difficult to know the direction of this bias.
The choice to interpret “Head-CT not performed” as absence of TICH can lead to missed intracranial hemorrhages. However, it is unlikely that any TICHs with serious consequences were missed since screening for new ED visits 6 months after the index-TBI was performed. These events might have been missed if they took place outside the Region Skåne catchment area, but that risk was deemed small.