Evidence synthesis on sex differences in mortality after traumatic brain injury: Systematic review integrating meta-analysis and explainable machine learning
This systematic review and meta-analysis of nearly one million adult traumatic brain injury patients reveals that females have lower mortality than males, while explainable machine learning identifies mean age, study sex composition, and study quality as key factors contributing to the heterogeneity in reported sex-specific mortality effects.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer
Every year, millions of people around the world suffer traumatic brain injuries, often from falls, vehicle crashes, or violence. While modern medicine has made great strides in saving lives, the risk of death after such an injury remains high, and it does not affect everyone equally. For decades, doctors and researchers have noticed that men and women sometimes survive these injuries at different rates, but the reasons behind this gap have been a source of confusion. Some studies suggest women are more likely to survive, while others show no difference or even the opposite. This inconsistency makes it difficult to build reliable tools that can predict a patient's outcome or to understand whether biology, social circumstances, or the way studies are conducted are driving these results. To clear up this uncertainty, a team of researchers set out to look at the entire body of existing evidence, not just to count who survived, but to understand why the numbers vary so wildly from one study to the next.
The researchers began by gathering every relevant study they could find, searching through eight major scientific databases for research published up to August 2025. They were looking for studies that focused on adults with brain injuries and that carefully accounted for two critical factors: how old the patient was and how severe the injury was. By filtering for these specific details, they ensured they were comparing apples to apples. From thousands of records, they selected twenty-one high-quality studies that together included nearly 900,000 patients and more than 98,000 deaths. When they combined the data from all these studies, a clear pattern emerged: women generally had lower odds of dying after a traumatic brain injury compared to men. Specifically, the analysis showed that women were less likely to die while in the hospital and less likely to die overall. However, this overall trend masked a much more complex reality. The difference in survival rates varied significantly depending on the specific study, the time period measured, and the group of people involved. In some cases, the difference was large; in others, it was small or non-existent.
To solve the puzzle of why these results differed so much, the team used a sophisticated approach that combined traditional statistical methods with a type of computer analysis known as explainable machine learning. Think of this process like a detective trying to figure out why a group of witnesses gives different accounts of the same event; instead of just accepting the conflicting stories, the detective looks for patterns in the witnesses' backgrounds, the time of day, and the location to see what might be influencing their reports. The researchers fed their computer models with details about each study, such as the average age of the participants, the proportion of men and women in the group, the quality of the research, and the country where the study took place. The computer then worked to identify which of these factors were most strongly linked to the differences in survival rates reported by the studies.
The analysis revealed that the variation in results was not random. The most important factors driving the differences were the average age of the people in the study, the mix of men and women in the group, and the overall quality of the research design. For instance, studies that included older populations or had a different balance of men and women tended to report different survival gaps than those with younger or more balanced groups. The quality of the study also mattered, with higher-quality research sometimes showing different patterns than lower-quality work. This suggests that the reason we see such conflicting results in the past is not necessarily because the biology of survival changes unpredictably, but because the groups of people being studied and the methods used to study them vary so widely. The researchers found that when you look at the raw numbers without considering these context clues, the picture is blurry. But when you account for who was in the study and how the study was done, the inconsistencies begin to make sense.
The findings have important implications for how we understand brain injuries and how we build tools to predict who will survive them. The researchers emphasize that the observed differences in survival between men and women are likely not just a simple matter of biology. Instead, they are shaped by a mix of biological factors, social circumstances, and the specific details of how the research was conducted. For example, the age of the patient matters deeply, as older adults face different risks and challenges than younger ones, and this age factor interacts with sex in complex ways. Furthermore, the study highlights that many past research efforts failed to report crucial details about their patients, such as their race, where they lived, or their social background. Without this information, it is impossible to know if the survival differences hold true for everyone or only for specific groups. The researchers argue that future studies must be more thorough in collecting and reporting these details. Only by understanding the full context of who is being studied can scientists develop accurate, fair, and reliable tools to help doctors make better decisions for every patient, regardless of their sex, age, or background.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.