1. Introduction
Minor head injury (MHI), in which affected patients have a Glasgow Coma Scale score of 14 or 15, is among the most common reasons why people go to the emergency department (ED). MHI accounts for approximately 88% of cases of head trauma. Despite the low risk for severe complications, MHI carries a potential risk of a poor outcome [
1]. For approximately 95% of patients with MHI, computed tomographic (CT) scans indicate no abnormality, and fewer than 1% are candidates for surgery. An increasing number of investigations are being conducted to identify the risk factors for intracranial hemorrhages [
2,
3,
4,
5,
6,
7,
8,
9,
10].
Although anticoagulant therapy is recognized as a risk factor for bleeding, ongoing use of antiplatelet drugs is still controversial. Reports suggest that people receiving antiplatelet therapy tend to bleed, and whether this represents an independent risk factor for hemorrhagic complications is unclear. The tendency to bleed is more pronounced with age [
11,
12,
13,
14,
15,
16,
17,
18,
19]. Since 2015, antiplatelets have increasingly been used as the primary method of preventing cardiovascular diseases [
20,
21,
22,
23].
The goal of our study was to assess the risk of bleeding in patients receiving antiplatelet therapy who visited our ED because of MHI. The secondary endpoints were hospitalization necessary for observation and monitoring intracranial hemorrhage (ICH), the need for cranial neurosurgical interventions for ICH, the severity of patients’ condition at discharge, and the rate of ED revisits for late bleeding.
2. Methods
2.1. Study Design
We conducted a retrospective single-center observational study involving all patients admitted to our ED in 2019 because of MHI. The primary endpoint was the diagnosis of post-traumatic ICH during observation and within 30 days. Secondary endpoints were the rate of hospitalization for observation and monitoring of ICH, the rate of ED revisits, the severity of patients’ condition at discharge, and the need for cranial neurosurgery during the hospital stay.
2.2. Inclusion and Exclusion Criteria
The study group involved patients with MHI, with Glasgow Coma Scale scores ≥14, aged >18 years, and receiving antiplatelet therapy (APT). The control group included patients with the same demographic data but who were not receiving APT or anticoagulant therapy and did not report any history of clotting disorders. The exclusion criteria were suspected or diagnosed skull fracture and spontaneous ICH. Of the patients whose records we reviewed, 129 were not eligible.
2.3. Study Population
The terms for our database search were “cranial trauma,” “intracranial hemorrhage,” and “head trauma.” Each patient’s data, including clinical reports, medical diary, and laboratory and radiological information, were examined with the use of the PIESSE digital platform.
We collected demographic data and information about the etiology and dynamic of injury, waiting time, process time, length of stay, blood pressure, means of arrival, triage code, exit code, and hematochemical examination results for all enrolled patients. In the triage guidelines of the American College of Surgeons, Sasser et al. described the dynamic criteria for major trauma [
24]: ejection from a vehicle, motorcyclist thrown from a vehicle, death in a vehicle, intrusion of an interior compartment of >30 cm, and falling from a height of >2 m (
Supplementary Table S1). When none of these criteria was present, the dynamics of trauma were considered to be minor.
Hospitalization for observation and monitoring of ICH, cranial surgery, death, and ED revisit within 30 days were documented. We viewed and evaluated all the performance data, and a neuroradiologist reviewed all CT scans. In patients receiving antiplatelet therapy, we reviewed each CT scan that was positive for ICH with regard to topography (intraparenchymal, subdural, epidural, and subarachnoid) and size. We also calculated the score on the Marshall classification of traumatic brain injury. Of the patients whose records we reviewed, 1926 consecutive patients with a diagnosis of MHI were enrolled; 483 were receiving APT and were assigned to the APT group, and the other 1443 were assigned to the control group (
Figure 1).
A triage code was assigned according to predefined grids by the triage nurse to the patient. The exit code is a five-point quantitative one-dimensional numerical scale for assessing the severity of the patient according to medical judgment, scale in force in Italy. It was assigned by the emergency doctor to the patient at the time of admission based on clinical judgment. Exit code 1 represents the greatest severity, with 5 being the least.
In accordance with the International Emergency Severity Index and Canadian Triage and Acuity Scale, we used a modified system of triage coding involving five codes; code 1 represented the most severe condition of a patient upon entry into the ED, and code 5 represented the least severe condition (
Table 1) [
25,
26,
27,
28,
29,
30,
31,
32,
33,
34,
35,
36,
37]. At discharge, according to the clinical judgment of the ED physician, severity codes, called exit codes, that were based on the same scale as the triage codes (
Table 1) were assigned to the patients. These were used to measure the severity of the patient’s condition at discharge.
2.4. Statistical Analysis
Continuous variables were calculated either as means ± standard deviations or as medians and interquartile ranges, depending on the normality of the distribution; categorical variables were calculated as counts and percentages.
Stata statistical software (Stata Corporation, College Station, TX, USA) was used. We used the t test to compare normally distributed data from the two groups; for other comparisons, we used the nonparametric Wilcoxon test. Associations between categorical variables were checked using Pearson’s χ2 test or Fisher’s exact test.
To study the association with the outcome variable (bleeding), we examined the potential covariates in the association between treatment and bleeding with univariable logistic models. After selecting the covariates (predictors for which a significant statistical association with the outcome variable bleeding had been reported), we tested logistic multivariable models with the treatment condition (APT vs. no APT) and each covariate as predictors.
To estimate the average treatment effect of APT versus no APT, we performed propensity score matching using a logistic model. The missing potential outcome for each patient receiving APT was determined from an average of the outcomes of similar patients who received APT, and that for each patient not receiving APT was determined according to similar patients not receiving APT. Similarity between patients was based on estimated treatment probabilities (propensity scores). To compute the treatment effect, we calculated the average of the differences between the observed and potential outcomes for each subject. Statistical significance was determined as p < 0.05, and all the tests were two-sided.
3. Results
Of the 1926 patients enrolled, 483 patients were receiving APT, and 1443 were not.
Table 1 lists demographic and clinical data. Patients receiving APT had slightly lower heart rates, slightly lower oxygen saturation, and higher blood pressure. Scores on the Glasgow Coma Scale were similar in the two groups (
p = 0.43). The APT group was, on average, older (approximately 81 vs. 54;
p < 0.0001), and the cause of head injury in the group was more often minor (98.53 vs. 93.79%;
p < 0.0001). The APT group had a greater number of high priority triage codes before the medical examination (
p < 0.0001), shorter waiting times (approximately 46 vs. 72 min;
p < 0.0001), and a longer length of stay in the ED (approximately 358 vs. 260 min;
p < 0.0001;
Table 1). Dynamic criteria for major trauma were present in only a small percentage of cases, for both the group on APT (1% (
n = 5)) and the control group (4% (
n = 58)). The causes of trauma are listed in
Supplementary Table S2.
Among the APT group, 85% were taking acetylsalicylic acid; 9%, clopidogrel; 5%, ticlopidine; and 1%, indobuphen. APT did not appear to be statistically significant as a risk factor for bleeding, both when tested as the only predictor and after adjustment for age in the multivariable analysis. Age did appear to be a risk factor for bleeding, which was also true in the multivariable analysis, after adjustment for therapy (
Table 2 and
Table 3).
All the potential covariables in the association between treatment (APT vs. no APT) and the outcome (bleeding) were selected according to logistic regression models, as reported for statistical methodology. Diastolic blood pressure, trauma dynamics, oxygen saturation, and age ≥75 years were associated with bleeding, and these associations remained significant even after adjustment for treatment (APT vs. no APT;
Table 4).
A logistic model with propensity score matching, based on the aforementioned covariables, was used to estimate the average treatment effect of APT versus no APT. As presented in
Table 5, we found no statistically significant difference between the two groups after propensity score matching.
Table 6 lists results of the univariable logistic model for the association between treatment (APT vs. no APT) and the outcome variable (bleeding) and the results for treatment after adjustment for each covariable with multivariable models.
With regard to secondary outcomes, the APT group had higher severity codes at discharge (
p < 0.001), a higher rate of hospitalization (14.9% vs. 10.1%;
p = 0.002;
Table 1), more need of surgery (1.2% vs. 0.4%;
p < 0.0001;
Table 1), and a higher rate of ED revisits within 30 days of the injury for reasons related to head trauma, although this finding was not statistically significant (5.6% vs. 3.5%;
p = 0.0608;
Table 1). The various types of hemorrhage and the Marshall classifications of the patients are listed in
Supplementary Table S3. Rates of hospital mortality did not differ between the groups.
4. Discussion
Our study did not demonstrate that APT was a risk factor for ICH. The findings indicate a higher rate of hospitalization to monitor ICH, a greater need for cranial surgery, and a worse condition at discharge among patients receiving APT.
The two groups analyzed differed mainly in age. It is possible that age also affected the other characteristics in which the two groups differed. Systolic blood pressure was higher in patients receiving APT, presumably because they are at greater cardiovascular risk and are older [
38,
39,
40,
41,
42,
43]. In fact, arterial hypertension is very common in patients harboring other cardiovascular risk factors [
44,
45]. Moreover, in patients receiving APT, falling was the most frequent mechanism of head trauma. In view of their having been older, falling could probably be related to age, as described in the literature [
46,
47,
48].
The two groups were similarly hemodynamically stable. Glasgow Coma Scale scores were also similar. Inasmuch as MHI is defined on the basis of the Glasgow Coma Scale score, the similarity in scores highlights the comparability of the two groups.
The literature contains much debate as to whether APT predisposes to bleeding in the event of mild head trauma [
1,
12,
49,
50,
51,
52,
53,
54,
55,
56,
57,
58,
59,
60,
61,
62,
63,
64]. Our study indicates that APT is not related to a higher risk of bleeding, but it is related to a more severe clinical condition when bleeding does occur. In patients receiving APT, unfavorable conditions related to MHI, particularly the need for neurosurgery and hospitalization, are predictable. Patients receiving APT also need a longer stay in the ED, and longer lengths of stay are correlated with worse outcomes [
57,
65,
66]. These patients, therefore, have complex problems, and during triage, it seems appropriate that they are classified as patients at greater risk.
Various results of studies have led to different protocols for adult and pediatric patients with MHI in the emergency room [
11,
12,
13,
14,
15,
16,
17,
18,
19,
67]. In some hospitals, all these patients undergo CT examination; in others, CT scanning is performed only if patients meet an age criterion (typically over 65 years) [
2,
3,
4,
5,
6,
7,
8,
20,
21,
22]. To our knowledge, however, different durations of observation have not been proposed. Thus, for patients who report MHI and are taking antiaggregation therapy, an observation period of 4–8 h is recommended [
68]. Patients who are receiving APT, do not have focal deficits, and have stable lesions can be sent home after proper observation; patients receiving APT should be closely monitored, possibly for 24 h if they have bleeding.
Age remains instead an independent bleeding factor. Previous study results may have been skewed by the age variable. This possibility is supported by other studies [
59,
60,
61,
62,
63,
64,
66,
68,
69]. Further studies are necessary to better delineate the risk profile of dual APT in patients with MHI.
Limitations of the Study
This study had all the limitations of a retrospective and single-center study. The bleeding predictors analyzed were those that we deemed reliable in a retrospective study: dynamics of trauma, therapy in progress, Glasgow Coma Scale score, vital signs, sex, and age. We did not consider amnesia because it is not always clearly reported.
The rates of mortality in the two groups were not significantly different; however, we decided not to study it because the low incidence of death in cases of MHI would have necessitated a greater number of patients to be enrolled.