Vancomycin versus β-Lactam Antibiotics and Acute Kidney Injury in ICU Patients with Osteomyelitis: A Causal Forest Analysis of Heterogeneous Treatment Effects
Using a Causal Forest analysis of MIMIC-IV data, this study reveals that while vancomycin does not significantly differ from β-lactam antibiotics in overall acute kidney injury risk for ICU osteomyelitis patients, substantial heterogeneous treatment effects exist, with specific subgroups characterized by lower anion gap, thrombocytopenia, lower neutrophil percentage, anemia, and absence of heart failure deriving greater renal benefit from vancomycin.
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
In the high-stakes environment of an intensive care unit, doctors often face a difficult choice when treating severe bone infections known as osteomyelitis. They must select an antibiotic that will kill the bacteria causing the infection without damaging the patient's kidneys, which are already under immense stress. Two common classes of drugs used for this purpose are vancomycin, a powerful medication often reserved for resistant bacteria, and beta-lactams, a group that includes familiar antibiotics like cefazolin. For years, medical debate has centered on whether vancomycin is inherently more dangerous to the kidneys than its counterparts. While some studies suggest it causes more kidney failure, others find the risk to be similar. The core of the problem is that patients are not all the same; a treatment that harms one person might be perfectly safe for another. Understanding exactly which patients fall into which category has been difficult because traditional methods often look only at the average outcome for a whole group, potentially hiding the specific risks and benefits for individuals.
A team of researchers set out to solve this puzzle by looking at a massive collection of medical records from a database called MIMIC-IV, which contains detailed information on thousands of patients admitted to an intensive care unit in Boston. They focused specifically on 884 adults who were admitted with osteomyelitis and treated with either vancomycin or a beta-lactam antibiotic within the first three days of their stay. Instead of simply counting how many people developed kidney injury in each group, the team used a sophisticated computer method called a causal forest. This approach acts like a highly detailed map, allowing the researchers to see how the effect of the drug changes from one patient to the next based on their unique medical history, lab results, and physical condition. They defined kidney injury as a specific, measurable rise in creatinine levels within a week of starting treatment, a standard way doctors monitor kidney health.
When the researchers looked at the entire group of patients together, they found that vancomycin did not show a statistically significant advantage or disadvantage over beta-lactam antibiotics regarding the risk of kidney injury. On a population level, the two drugs appeared to carry a similar risk profile. However, the story changed dramatically when they examined the individual patients. The computer analysis revealed that the effect of the drug was not uniform; for some patients, vancomycin was associated with a much lower risk of kidney injury, while for others, the risk was higher or unchanged. The researchers identified a distinct group of patients who derived a clear benefit from vancomycin. These individuals tended to be younger and had specific characteristics in their blood work: they had lower levels of anion gap, a measure of acid balance in the blood; lower platelet counts; lower percentages of neutrophils, a type of white blood cell; and higher levels of hemoglobin. They were also less likely to have a history of heart failure.
In contrast, patients who did not fit this profile, particularly those with higher anion gaps, normal platelet counts, and a history of heart failure, did not see the same protective benefit from vancomycin. In fact, for the group identified as high-benefit, the data showed a stark difference in outcomes. Among those patients who received vancomycin, the rate of kidney injury was 12 percent, whereas those who received beta-lactam antibiotics in the same group saw a rate of 35 percent. This gap suggests that for this specific subset of patients, choosing vancomycin could significantly reduce the likelihood of kidney damage. The researchers used these findings to build a simple scoring tool, or nomogram, that doctors could potentially use to estimate whether a specific patient is likely to benefit from vancomycin based on their lab results and medical history.
The study does not claim that vancomycin is universally safer or more dangerous than beta-lactams. Instead, it suggests that the answer depends entirely on the individual patient. The findings challenge the idea of a one-size-fits-all approach to antibiotic selection in critical care. By identifying that patients with lower anion gaps, lower platelet counts, lower neutrophil percentages, higher hemoglobin, and no heart failure are the ones who gain the most protection from kidney injury when treated with vancomycin, the study offers a path toward more personalized medicine. The researchers acknowledge that their work is based on retrospective data, meaning they looked back at records rather than conducting a new experiment, and that future studies are needed to confirm these patterns. Nevertheless, the analysis provides a clear framework for understanding why previous studies might have produced conflicting results and offers a concrete way to tailor treatment decisions to the specific needs of the patient in front of the doctor.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.