See Figure 1 for an illustration of the exclusion process. Study quality and risk of bias for each of the included studies, according to the NHLBI quality assessment tool for observational cohort and cross-sectional studies, is presented in Table 1. All eight studies employed an observational cohort study design and were assigned a “C” Level of Evidence. One of the studies included a prospective cohort study design while the remaining seven studies included a retrospective study designs. The prospective study was assessed as a good study as the investigators had control over the quality, accuracy and completeness of collected data. In the remaining seven studies, a retrospective approach was used where investigators had to rely on pre-existing data that could not be confirmed nor deemed reliable. This creates a susceptibility to recall bias and attrition bias. Though not highly esteemed as randomized controlled studies, observational cohort studies can be efficient in answering specific type of research questions. However, special attention must be given to the presence of potentially confounding factors. Only four of the eight studies included addressed confounding factors; rendering the remaining four studies a “fair” quality rating.Participants in each of the eight studies were selected based on the presence of a TBI, with some studies including TBI severity in their definition of TBI. Based on the study designs utilized by included studies, selection bias regarding sampling was anticipated as participants are not randomized, rather they are selected based on the outcome and exposure of interest; in such study designs, convenience sampling is most often utilized. Due to the nature of the studies included in this review, allocation concealment and blinding of outcomes assessors is not feasible. Because the exposure of substance abuse has not been allocated randomly,vertical farm tower a causal effect may not be possible as other variables may be found to influence the outcomes studied, rendering all eight studies at a major disadvantage with potential bias in outcomes.
Study methods employed by each of the eight included studies varied with some studies utilizing medical chart reviews, while others utilized validated surveys and questionnaires to gather their data. The studies by Andelic et al., Barker et al., and Bombardier et al. all utilized the participants’ medical charts for retrospective review for presence of substances. The studies by Andelic et al., Nguyen et al., and O’Phelan et al. used trauma registry databases to collect data on TBI patients and the presence of substance abuse. Pakula et al. collected data on the presence of substance abuse in post-mortem patients with traumatic cranial injuries by evaluating autopsy reports. Finally, the studies by Bombardier et al., Kolakowsky-Hayner et al., and Kreutzer, Witol and Marwitz utilized questionnaires to interview participants. The variance in study methods, ranging from retrospective review of charts to the use of self-report methodology subjects the included studies to recall bias and unreliable data. A factor negatively contributing to the quality of the included studies is the variance in defining a TBI. Three of the studies did not provide a definition for what constitutes a TBI, nor did they describe the severity of TBI. The study by Andelic et al. defined TBI using the TBI Modified Marshall Classification. The study by Barker et al. defined TBI using the TBI Model Systems Data Base definition. Nguyen et al. used the International Classification of Diseases-Ninth Revision codes and the Abbreviated Injury Severity codes to define TBI. These codes are widely used in trauma data registries for entering and recording the injury type and severity, for performance improvement and billing purposes. However, reliability can be an issue as coding may be subjective. The information is extracted from the chart by registrars who read and enter notes written by physicians. Often, coding depends on physician documentation, attention by trauma registrars to the various sources of documentation and communicating with physicians when necessary. If not subject to continuous data validation, a data gap may ensue. The study by Pakula et al. defined a central nervous system injury by the presence of any of the following written diagnosis as found in the autopsy reports: 1) TBI, 2) skull base fracture, 3) spinal cord injury, and 4) cervical spine injury. Only one study, the study by O’Phelan utilized a Glasgow Coma Score to define a severe TBI.
The majority of the articles were subject to selection bias in terms of their participant population and methods of data collection: See table 2 for specifics. The included studies varied in their definition of TBI. One study used the Modified Marshall Classification of TBI which is a Computed Tomography scan derived metric used to grade acute TBI on the basis of CT findings. Another study defined TBI using the TBI Model Systems National Database definition. The TBIMS-NDB has been funded by the National Institute on Disability and Rehabilitation Research in the U.S. Department of Education to study the course of recovery and outcomes following a TBI. They describe the TBIMS-NDB TBI as: Damage to brain tissue caused by an external mechanical force as evidence by medically documented loss of consciousness or post-traumatic amnesia , or by objective neurological findings on physical or mental examination that can be reasonably attributed to TBI. Three of the eight studies did not specify how TBI was defined. One study used the following International Classification of Disease, 9th Revision codes to define TBI: 800.1- 800.39 ; 800.6-800.89 ; 801.1-801.39 ; 801.6-801.89 ; 803.6-803.89 ; 804.6-804.79 ; 851 ; 852 and 853 . Another study used the International Classification of Disease, 10th Revision codes to define TBI: S02.0xx ; S02.1 ; S06.1 ; S06.2 ; S06.3 ; S06.31; S06.32 ; S06.33 and S09.x . Finally, the last of the eight studies used autopsy reports to evaluate individuals with severe central nervous system injuries. For purposes of that study, severe CNS injuries were defined as “any traumatic brain injury, skull base fracture, spinal cord injury, or cervical spine injury.” Although all eight studies investigated marijuana exposure in TBI patients, only one study specifically investigated the use of marijuana alone on outcomes in TBI. All other remaining studies investigated the presence of all possible substances and/or drugs, meaning investigators were not specifically examining marijuana exposure by itself. In Nguyen et al. all patients who had sustained a TBI and had a urine toxicology screen were included. The actual noted presence of marijuana was obtained from the urine toxicology screen and not through any other modes of measurement. The authors classified study patients according to marijuana screen results which they defined as greater than 50 ng/ml. Though marijuana was noted to have been detected across all eight studies, the actual numerical or absolute value measured was never reported by any of the studies. Additionally, it is important to note that excluding the study by Nguyen et al., the presence of marijuana was not reported in a quantifiable manner, making any potential statistical inference impossible.
Lastly, six of the included studies investigated the presence of marijuana at the time of injury, while the remaining two studies measured the presence of marijuana use during the past year and post-mortem respectively. The study by O’Phelan et al. did not investigate any other time frame for which marijuana may have been used, rather, the authors only collected data on the presence of drugs at the time of injury. An important finding from the systematic literature review showed that marijuana was the most favored drug reported. However, only one study of the eight studies included explicitly searched for and found a connection between the presence of a positive toxicology screen for marijuana and mortality outcomes in TBI patients. Nguyen et al. three-year retrospective review of trauma registry data found that 18.4 percent of all cases meeting inclusion criteria had a positive marijuana screen and overall mortality was 9.9 percent . Nguyen et al. found that mortality in the marijuana positive group was significantly lower when compared to the marijuana negative group . Authors adjusted for the following differences between study participants: age, gender, ethnicity, alcohol,vertical farming greenhouse abbreviated injury scores, injury severity scores, and mechanism of injury. After adjusting for differences, Nguyen et al. found that a positive marijuana screen was an independent predictor of survival in TBI patients .This review sought to determine the use of marijuana and its role in TBI prevalence and outcomes. A key finding from this review is that there are few studies available that examine the specific role of marijuana exposure on TBI severity, leaving many questions unanswered. Furthermore, this review found that there is a significant variation in how substance abuse has been defined, conceptualized, and operationalized in TBI research. Another important finding was that the reviewed studies operationalized substance abuse inconsistently, often combining alcohol and drugs in one category titled ‘substance abuse,’ making it difficult to ascertain if there was an association between specific drugs, particularly marijuana, and TBI severity and outcomes. The difference in how substance abuse was operationalized in these reviewed studies has important implications for how findings are interpreted as well as provide recommendations for future research. Although there was no restriction made to the countries in which these studies were conducted, those meeting inclusion criteria were all studies conducted in the US except one from Norway. Therefore, the applicability of findings from that one non American study is limited. Additionally, it is difficult to draw valid and reliable conclusions when the studies reviewed utilized a wide variety of study objectives, sample size, study methods, and varying definitions for substance abuse classification.
The review showed a great variation existed across the studies in types of data collected and methods used, thus severely minimizing comparability. For example, the disparity in measurement of blood alcohol levels considerably reduce the reliability of data related to pre-injury intoxication. In the reviewed studies, information on alcohol and substance use was obtained from a range of different sources, including self-reports and patient records, as well as a variety of different measures rendering results unreliable across studies. This review set out to answer a specific question: what influence, if any, does marijuana exposure at time of injury have on TBI severity and outcomes? Only one study about marijuana’s effect on TBI outcomes was available. Nguyen et al. reported that a positive marijuana screen is an independent predictor of survival, suggesting a potential neuroprotective effect of cannabinoids in TBI. The rest of the studies yielded a variety of findings, with the most common finding being that marijuana and other drug use, including alcohol, are common before TBI. To clearly understand what marijuana’s influence on TBI is, potential confounding variables must be identified and controlled for. The literature review identified no consensus on relevant confounding variables aside from age and gender. The variability in all other demographic variables highlights the lack of certainty of the full range of relevant demographic variables. Another potentially important confounding variable is mechanism of injury. Historically, the most frequent cause of TBI related deaths in civilians was considered motor vehicle crashes. However, recent data show that falls are actually the leading cause of TBI related hospitalizations, with the second leading cause is being struck by another object. Importantly, only six of the studies included mechanism of injury as a variable in their analysis of findings. Five of the eight included studies did not address TBI severity as a variable. The remaining three studies each operationalized TBI severity utilizing different methods. Andelic et al. used the Marshall classification to classify neurological anatomical abnormalities as seen on CT scans. Nguyen et al. utilized the Abbreviated Injury Scale score for the head and neck region to classify TBI severity. The use of the AIS score is common in general research studies as often times the GCS score is not always recorded for each individual participant. Hence the only study showing a link between marijuana exposure and TBI severity did not use the gold standard of GCS to measure TBI specific severity. Finally, severity as a variable in the TBI population is an important characteristic and is a parameter of interest when answering the research question of whether or not marijuana influences TBI severity; available studies are not able to answer that question mostly because the majority of them did not measure severity in the first place.