Can herbs cure sepsis? The EXIT-SEP trial was just published, and demonstrated a decrease in all cause mortality from xuebijing – a product manufactured by a pharmaceutical company from a combination of Carthamus tinctorius flowers (Honghua in Chinese), Paeonia lactiflora roots (Chishao), Ligusticum chuanxiong rhizomes (Chuanxiong), Angelica sinensis roots (Danggui), and Salvia miltiorrhiza roots (Danshen). What should we make of these somewhat surprising results?

The paper

Liu S, Yao C, Xie J, et al. Effect of an Herbal-Based Injection on 28-Day Mortality in Patients With Sepsis: The EXIT-SEP Randomized Clinical Trial. JAMA Intern Med. 2023 May 1:e230780. doi: 10.1001/jamainternmed.2023.0780. PMID: 37126332 NCT03238742

The Methods

The EXIT-SEP is a randomized, double-blind, placebo controlled, multicenter trial from 45 ICUs in China. 


Adult patients (between 18 and 75 years) with sepsis and a SOFA score between 2 and 13.

Key exclusions were: sepsis for more than 48 hours at the time of enrollment, severe primary disease (such as an unresectable tumor), severe liver or kidney dysfunction, and immunosuppression.


Xuebijing injection for 5 consecutive days.


Matching placebo


The primary outcome was 28 day all cause mortality.

The Results

Between 2017 and 2019, there were 4692 septic patients screened, and they ultimately enrolled 1817 patients. Presumably because of the exclusion criteria, the mean age is relatively young (56 years), with 65% of patients being male, and a relatively low rate of co-morbidities (30% HTN, 20% diabetes, and everything else in the low single digits). The mean APACHE II scores were also relatively low at about 12. The groups do appear to be evenly matched. 

Almost 25% of patients did not complete their course of the study medication.

Xuebijing resulted in a 7% absolute decrease in the primary outcome of 28 day all-cause mortality (26% vs 19%, p<0.001).

Secondary outcomes are also mostly positive (although I am a little more concerned about them, as they were changed from the original research protocol relatively late in the data gathering period).

My thoughts

Although there are always some minor points that can be discussed as part of a critical appraisal, this is a well done study, which followed the registered protocol, and demonstrated a large patient important benefit. Although I will mention a few critical appraisal issues, the key concepts when interpreting this study are broader, focused on pretest probability, and the general logic (or illogic) of herbal remedies.

Selection bias is always an issue to consider. The exclusion criteria of this study resulted in a much healthier population than the average ICU sepsis patient. There are also always concerns that a study completed entirely in China may not extrapolate to other areas of the world.

Given that the treatment course was only 5 days, it is a pretty big red flag that 25% of patients didn’t complete the treatment. However, it seems like this is mostly because the patients were transferred to the floor, and the rate seems similar between the groups. A somewhat bigger concern is that they don’t know the mortality status for 57 patients (which is pretty high, especially for an easily documented status of alive or dead), and the researchers just assumed the patients were alive.

It is also worth noting that a 7% decrease in absolute mortality is relatively unheard of, and so although it makes the results exciting, it also raises the ‘potentially too good to be true’ red flag.

The pretest probability for any study looking to decrease mortality in sepsis is going to be low. The pretest probability of any study looking at ‘alternative’ medicine is going to be low. Of the many forms of ‘alternative’ medicine, herbals probably have the highest pretest probability, because plant products contain thousands of biologically active substances, and we have formulated many of our drugs from those substances. (In contrast, the pretest probability of homeopathy is 0%, and so you can ignore the results of any trial, because it is simply scientifically impossible for homeopathy to provide benefit). When you start with a low pretest probability, a single study is never convincing in isolation. Even when the trial is high quality, there is always risk of both bias and chance findings. Replication is a core principle of science. Therefore, this trial should be seen as preliminary, and really needs replication before there is any clinical practice change. 

More importantly, herbal remedies will never be the best approach. A herbal remedy is a collection of thousands of chemicals, each with different biological actions. If a herbal remedy truly has an effect, it is highly probable that only one, or a small handful, of those chemicals is responsible. So the first big problem with herbals is that you are being exposed to thousands of additional pharmaceutical chemicals that are unnecessary and potentially harmful. 

A bigger issue is the lack of quality control in herbal remedies. The chemical composition of plants varies dramatically from crop to crop, depending on the time of year, weather conditions, and many other factors. (Just compare the flavor of an off season strawberry to one at the height of strawberry season, and the varying chemical composition of plants will become incredibly clear.) And that is just the variance of an idealized herbal remedy, without even accounting for the long history of adulterants found in these products, and variance in manufacturing processes. In essence, using a herbal remedy is using a random assortment of chemicals in random doses and praying for a good effect. That is a really bad plan.

Can herbals be useful? Of course, if we identify the important chemicals, purify them, and standardize the dose. But we just call that medicine.

Discussions of herbal remedies are frequently peppered with fallacies, and I won’t be able to cover them all here. There is nothing special about “natural”. Cyanide and sharks are both natural, but that doesn’t make them any less deadly. Plants do not evolve to help humans. Plants develop many of the chemicals they contain to act as poisons, specifically to prevent animals from ingesting them. You shouldn’t feel any better about these chemicals because they are “natural”.

Many people distrust the pharmaceutical industry (probably with good reason). However, they fail to recognize that the herbal supplement industry is also a multi-billion dollar industry, and the products are often sold by the exact same pharmaceutical companies, but just with less regulation. If you distrust “big pharma”, you should distrust “big herbal” even more, because there is just as much money to be made, but with far less regulation.

For that matter, the definition of herbal is extremely nebulous. Xuebijing is a product manufactured for intravenous injection by the Tianjin Chase Sun Pharmaceutical corporation. How is this different from a medication? If they had given this product an unpronounceable generic name, and marketed the hell out of the brand name “AntiSEP”, we would have just treated it like a drug. (Although, hopefully, we would have criticized the drug for its incredibly poor manufacturing standards).

At the end of the day, these are important results, especially if they are replicated. However, the appropriate clinical response is not to use a random mixture of chemicals in varying concentrations in a herbal product. The appropriate response is to begin work on isolating the individual chemicals and/or combinations that might actually be having a beneficial impact here, so that we can produce a safe and consistent medication for our patients.

Chemical components of xuebijing

Pay close attention to the high variability, represented in the +/- of the average content level. Would you accept a pharmaceutical agent whose active ingredients varied that much from batch to batch?

Bottom line

This is a promising study, but should be considered preliminary research, both because the results need replication, but also because we should be working towards being much more precise in the formulation, rather than exposing patients to this inconsistent brew of chemicals. 


Liu S, Yao C, Xie J, et al. Effect of an Herbal-Based Injection on 28-Day Mortality in Patients With Sepsis: The EXIT-SEP Randomized Clinical Trial. JAMA Intern Med. 2023 May 1:e230780. doi: 10.1001/jamainternmed.2023.0780. Epub ahead of print. PMID: 37126332