New alert system for deadly condition

Sixth-year medical student, Satyen Hargovan’s honours research project has generated a check-list of the top 10 factors that influence the survival of sepsis patients.

Sepsis is a common medical condition, where chemicals released by the immune system to fight an infection trigger the inflammation and eventual shut-down of multiple organs.

Some 15,700 patients with sepsis are admitted to Australian Intensive Care Units every year. About one in five die. The cost of treating each case is more than $39,000.

“The burden of sepsis is not only in relation to patient outcomes, but also the financial cost to the healthcare system,” Mr Hargovan said.

His project, known as the Cairns Sepsis Model – 4 (CSM- 4) is designed to speed up the assessment of patients within their first four hours in ICU.

The new four-hour mortality prediction model will enable clinicians to detect patients who are at a higher risk of dying, then help to guide their treatment decisions and discussions with the patient and their family.

“Essentially it means that if a patient has a five percent chance of dying, you might treat them less invasively,” he said. “For example, you won’t rush to place them in an induced coma, or insert lines into large arteries or veins.

“For a patient with a 50 percent chance of dying, you might use more invasive treatments. But if a patient has a 95 percent chance of dying, you might look at end-of-life measures that focus mainly on their comfort and quality of life.”

Mr Hargovan began his research by reviewing more than 50 mortality prediction studies, but soon realized that there was no model in the world specifically designed to predict mortality in adult sepsis patients within four hours of ICU admission.

“A good model needs to be highly accurate, sepsis-specific, and easy to use for clinicians at the bedside,” he said. “So we decided to create a model.”

A case history review of 500 adult sepsis patients admitted to the Cairns Hospital ICU over a four-year period identified 168 factors that could influence patient health outcomes, ranging from age, gender and socio-economic status to type of infection, degree of organ failure, vital signs, blood test results and treatment regimes.

Statistical analysis whittled the figure down to the 10 most significant factors, which include age, history of renal disease, level of consciousness, and abnormal levels of certain substances in the blood that represent varying degrees of organ dysfunction.

The mortality prediction model is inexpensive to use as it largely relies on information easily gleaned from patient history, physical examination, and blood tests that are routinely conducted on patients within their first four hours of admission to an ICU.

“One of its strengths, apart from accuracy, is its simplicity,” Mr Hargovan said. “Some mortality prediction models contain 20 or more variables. Ours contains just 10, which makes it easier for clinicians to utilise.”

The next step involves testing his model on a different and larger set of patients in order to validate his research, before it can be accepted for clinical use.

“Then we’d love to create an online calculator or an app that clinicians can use easily, at the bedside, and have a result within seconds,” he said.