Methods
Study Design and Setting
This study is a retrospective cohort study and a part of the Tampere Traumatic Head and Brain Injury Study. The study was conducted at Tampere University Hospital between August 2010 and July 2012. Tampere University Hospital is a tertiary care hospital and the only neurosurgical referral hospital in the hospital district, providing health services and demanding specialized care for a total of approximately 900,000 residents from 52 municipalities, both urban and rural.
The study population included all consecutive patients assessed at the emergency department of Tampere University Hospital for head trauma and referred to head computed tomography (CT) scanning. Each admission for a new head trauma was recorded as its own case. The minimum criteria for TBI were based on the World Health Organization’s (WHO’s) Collaborating Centre for Neurotrauma Task Force on Mild Traumatic Brain Injury definition presented in 2004 [
22]. The Department of Veterans Affair/Department of Defense guideline’s definition of moderate and severe TBI was applied to classify the moderate and severe TBI cases [
23]. Additionally, the following exclusion criteria were used: (1) no traumatic intracranial lesions in primary CT, (2) age younger than 18 years at the date of trauma, (3) LOS in the university hospital less than 48 h for any reason, and (4) withdrawal of life-sustaining therapy after the first assessment in the emergency department.
In this study, anemia was defined by the WHO Department of Nutrition for Health and Development recommendations for the diagnosis of anemia [
24]. Accordingly, Hb concentration for anemia was set to < 130 g/L for men and < 120 g/L for women. The Hb concentration threshold for RBCTs among patients with acute TBI is traditionally considered to be 90–100 g/L [
1,
2]. An Hb concentration threshold of 100 g/L is also the threshold for a liberal transfusion strategy in the HEMOTION (HEMOglobin transfusion threshold in Traumatic brain Injury Optimization) Trial, a randomized prospective clinical trial evaluating the effects of RBCT thresholds on neurological functional outcome [
25]. Hence, this Hb concentration threshold was of special interest and was applied to our study population to evaluate the incidence and associated factors of severe anemia.
Main Variables and Data Sources
Data were obtained retrospectively from medical records and the Tampere Traumatic Head and Brain Injury Study database, with no direct contact to patients or their representatives. The main study variables included patient demographics, injury-related findings, and major comorbidities (e.g., chronic cardiovascular, pulmonary, kidney, or liver diseases; chronic bleeding disorders; and prior TBIs). Also, information on regular medications with effects on blood coagulation and clot formation was obtained, including the use of anticoagulants (warfarin and direct oral anticoagulants), antiplatelet medication (aspirin and adenosine diphosphate receptor inhibitors), SSRIs, SNRIs, and long-term oral corticosteroids. All the head CT scans were interpreted by a neuroradiologist, and the findings were coded in compliance with the Common Data Elements by the National Institute of Neurological Disorders and Stroke [
26]. From the in-hospital period, information on major neurosurgical operations (including craniotomy, hemicraniectomy, trepanation, or repairment of skull fractures), duration of ICU and ward admission, and hospital-acquired infections was obtained.
To investigate the incidence of anemia, each measured Hb value, with a corresponding time label, was collected until hospital discharge, interinstitutional transfer, or death or for a maximum of 30 days. Hb values from both blood count samples and arterial blood gas analysis samples were considered.
Statistical Methods
Variable distributions were defined with the Kolmogorov–Smirnov and Shapiro–Wilk test. For descriptive data, mean with standard deviation (SD) was used with normally distributed variables, and median with interquartile range was used with nonnormally distributed data. The independent-samples t-test for normally distributed variables and Mann–Whitney U-test for nonnormally distributed variables were used. The χ2 test was performed for categorial variables. Univariate and multivariable analyses were performed with a binary logistic regression model. Variables for the multivariable analysis were selected on the basis of clinical relevance and in an attempt to avoid overlapping of subgroups, prior to any analysis, and included age, sex, overall comorbidity, the use of any antithrombotic medication, chronic alcohol consumption, presence of extracranial injuries, and midline shift on primary head CT as a surrogate to severe space-occupying brain trauma.
All statistical analyses were performed with IBM SPSS Statistics (Armonk, NY) versions 25 and 27. For all analyses, a p value < 0.05 was considered statistically significant.
Study Ethics
Because this was a retrospective, medical-record-based study, written informed consent from the patients was not required. The primary Tampere Traumatic Head and Brain Injury Study was approved by the Ethical Committee of Pirkanmaa Hospital District, Tampere, Finland (ethics code: R10027). Institutional ethics and research board approval was also obtained.
Discussion
In our study, 45.5% of the patients developed severe anemia (Hb level less than 100 g/L) during hospitalization for acute moderate to severe TBI. According to WHO criteria, the incidence of anemia was almost 90%. The univariate analysis associated female sex, cardiovascular comorbidities, diabetes, and antithrombotic medication with severe anemia, as well as extracranial injuries, midline shift on admission CT, and major neurosurgical interventions. In the multivariable analysis, only midline shift and the presence of extracranial injuries resulted in a statistically significant association. Regardless of the subgroup, the greatest decline in Hb level was seen between 24 and 48 h after the trauma.
The incidence of severe anemia in our study is somewhat greater than that reported in the general ICU population [
27,
28] but remarkably lower than that reported among patients in the surgical ICU [
29]. The blood loss due to closed head trauma is relatively minute and is unlikely to cause a significant decrease in the Hb level alone, even when operated, which might explain the lower incidence of anemia compared with the general surgical ICU population. Instead, the development of anemia in patients with TBI is likely to be multifactorial, including possible hemodilution due to intravenous fluid resuscitation and bleeding (preoperative and perioperative) of extracranial traumas. Among patients with severe TBI and prolonged hospitalization, a late decrease in the Hb level is likely to resemble the anemia seen in the general critical care population during an extended LOS in the ICU, involving hemodilution, repeated phlebotomies, and inflammatory processes resulting in bone marrow suppression and impaired erythropoiesis [
30]. The incidence of severe anemia in our study is in line with the results from a large retrospective study by Salim and colleagues [
15], with a 46% incidence of severe anemia. However, there are some major methodological differences between the studies. The study population of Salim et al. [
15] included all patients with blunt TBI admitted to the surgical ICU, regardless of the severity of the TBI, thus including also mild TBIs. In addition, anemia was defined as an Hb level less than 90 g/L for three consecutive samples, a definition markedly stricter than that in our study.
Reported data of factors associated with the incidence of anemia in patients with TBI are scarce. Salim et al. [
15] did not investigate risk factors for anemia, but according to study population characteristics, anemic patients were more gravely injured, and the relative proportion of female patients was higher among anemic than nonanemic patients, findings in concordance with our study. Also, in line with our study, the large retrospective multicenter cohort study by Boutin et al. [
31] demonstrated the association between female sex, comorbidities, extracerebral traumas, and RBCT, possibly indirectly indicating their association with anemia. In addition, the meta-analysis of RBCTs in patients with TBI indicated the association between RBCT and the severity of TBI [
21]. The association between anemia and extracranial injuries seen in the multivariable analysis of our study is potentially related to bleeding from primary injuries and during surgery and fluid resuscitation for possible hypovolemia and hemodynamical instability. More excessive injuries might also lead to longer LOS in both the ICU and hospital, predisposing patients to anemia due to prolonged hospitalization as discussed earlier in this section. The role of midline shift in anemia is possibly associated with the same mechanisms: more severe trauma leads to longer hospitalization.
According to our univariate analysis, comorbidity, but not age, was associated with the development of anemia, suggesting that accumulation of chronic illnesses is more relevant than age-related fragility in development of acute anemia after TBI. This lack of association in the multivariable analysis might be at least partly related to overlapping of one or more subgroups: assumingly, patients with chronic (cardiovascular) diseases use antithrombotic medication more frequently than healthier patients, resulting in comparison of two groups with almost the same patients. We found no statistically significant association with the use of SSRI or SNRI medication and acute anemia in our study. The lack of association is likely related to the small proportion of the patients with continuous SSRI or SNRI medication and the relatively weak effect of the serotonin reuptake inhibitors on platelet function [
32].
To our knowledge, the pattern of Hb level trajectories during hospitalization has not been previously described in patients with acute TBI. In every studied subgroup, the major decline in Hb level was seen during the first 48 h from the injury (days 0–1). After the first 48 h, Hb levels tended to stabilize or slightly increase in most subgroups, a phenomenon possibly explained by early surgery of intracranial and extracranial traumas and hemodilution after fluid resuscitation. If the LOS in the hospital was prolonged, Hb levels decreased in patients with major comorbidities. In this subgroup of patients, frailty can be one factor associated with the lingering decrease in Hb levels. The decline in Hb levels seen with prolonged stay in the ICU correlates with findings of previous studies in the general ICU population [
28,
33]. Patients with a more severe TBI (based on midline shift on CT) had higher Hb levels subacutely (> 4 days post-TBI) compared with patients with a less severe injury on CT. This finding can possibly reflect more aggressive treatment of declined Hb levels in severely injured patients.
Our study has several limitations. Firstly, the retrospective single-center study setting might cause bias because of preferred treatment modalities and practices and reduce the generalizability to other hospitals and TBI populations. Secondly, excluding patients with an LOS in the study center of less than 48 h might lead to inclusion of patients with more severe traumas, and on the other hand, to exclusion of patients with fatal injuries, either intracranial or extracranial, and thus cause selection bias. Thirdly, the use of midline shift as a surrogate of more severe trauma can be controversary, possibly ignoring severe, bilateral intra-axial lesions. Fourthly, because of retrospective nature of the study, the timing and frequency of Hb measurements varied between patients, resulting in a less accurate and structured setting. Fifthly, the delay from the insult to the hospital admission varied between patients, causing possible confounding in the time-dependent development course of anemia. Sixthly, extracranial traumas were not classified because of type, location, or severity, leading to inaccuracy in susceptibility to bleeding. Lastly, the study did not include data of RBCTs, which are likely to affect the incidence and course of anemia.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.