Development and external validation of a perioperative prediction model for high hidden blood loss after proximal femoral nail antirotation in older patients with intertrochanteric fractures
This study developed and externally validated a perioperative prediction model incorporating preoperative anticoagulant use, albumin levels, and operation time to effectively identify older patients at high risk for hidden blood loss following proximal femoral nail antirotation surgery.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
When an older person breaks the bone near the hip, the injury is often severe enough to require surgery to hold the pieces together while they heal. Surgeons frequently use a metal rod called a proximal femoral nail to fix these breaks, a procedure that is generally considered safe and minimally invasive. However, even when the surgery looks clean and the amount of blood seen on the floor or in the drain is small, many patients lose a significant amount of blood that cannot be seen. This hidden loss happens because blood seeps into the soft tissues around the fracture, fills the hollow space inside the bone, or is lost through the body's natural breakdown of red blood cells. This invisible bleeding can leave patients dangerously low on hemoglobin, the protein that carries oxygen, leading to fatigue, heart strain, and a slower recovery. For doctors, the challenge has been figuring out which patients are most likely to suffer from this hidden loss before it becomes a crisis, so they can prepare the right amount of blood for a transfusion or take extra steps to monitor the patient closely.
A team of researchers from Renmin Hospital in Hubei, China, set out to solve this problem by creating a simple tool to predict who would lose a large amount of hidden blood after this specific type of hip surgery. They looked back at the medical records of 186 older patients who had undergone the procedure at their hospital to build a model, and then tested that same model on a completely separate group of 84 patients from a different hospital to see if it would work in the real world. They defined "high" hidden blood loss as anything over 650 milliliters, a threshold they determined using statistical methods to represent a clinically significant amount. By analyzing dozens of potential factors, from the patient's age and fracture type to their blood work and how long the surgery took, the researchers narrowed down the list to just three key indicators that could reliably forecast the risk.
The study found that the most important clues were already available to the medical team either before or during the operation. The first clue was whether the patient was taking blood-thinning medication, such as anticoagulants, before the surgery. Patients on these drugs were much more likely to experience significant hidden blood loss, likely because their blood is less able to clot quickly at the site of the injury. The second clue was the patient's level of albumin, a protein in the blood that reflects their nutritional status and overall health. Patients with lower levels of albumin were at higher risk, suggesting that a body in a state of poor nutrition or inflammation is less resilient to the stress of surgery and bleeding. The third clue was the length of the operation itself. The longer the surgery took, the more blood was lost, which makes intuitive sense as longer procedures often involve more complex bone repair and greater tissue manipulation.
Using these three factors, the researchers built a prediction model that performed very well when tested on the second group of patients. In the initial group, the model achieved an AUC of 0.7, but when applied to the independent group of 84 patients, its AUC jumped to nearly 0.9. This high level of success suggests that the model is robust and can be trusted to distinguish between patients who will have a quiet recovery and those who will need close monitoring. The researchers also created a simpler version of the tool that only used information available before the surgery started, which could help doctors plan ahead, though this version was slightly less accurate than the one that included the length of the operation.
Despite the strong performance, the researchers were careful to note that the model tended to overestimate the exact amount of risk when applied to the second group of patients, predicting a higher chance of blood loss than actually occurred. This happened because the second group of patients was generally healthier and had fewer complications than the first group. To fix this, the team showed that a small mathematical adjustment could align the predictions with reality without changing the model's ability to rank patients by risk. This finding highlights that while the tool is excellent at spotting who is at risk, hospitals may need to tweak the final numbers to match their specific patient population before using it to make decisions about blood transfusions.
Ultimately, this work offers a practical way to move from guessing to knowing in the management of hip fracture patients. By focusing on just three straightforward factors—medication history, a simple blood test, and surgical duration—doctors can identify patients who are likely to lose hidden blood and prepare for it in advance. This approach does not replace the need for clinical judgment, but it provides a clear, evidence-based guide to help ensure that older patients receive the right care at the right time, potentially preventing complications and helping them return to their lives more quickly. The study concludes that while the tool is ready for use, it should be validated further in different settings to ensure it works perfectly for every patient it is meant to help.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.