Researchers used a computer algorithm to identify four types of sepsis patients, which could help providers offer more tailored treatments, according to a study published in JAMA and presented at the American Thoracic Society International Conference in Dallas.
For the study, researchers from the University of Pittsburgh examined EHR data for 20,189 patients at 12 hospitals affiliated with Pittsburgh-based UPMC. They analyzed 29 clinical variables found in the EHRs using computer algorithms. All patients analyzed were diagnosed with sepsis within six hours of arrival at the hospital between 2010 and 2012, according to Healio.
The algorithms identified four groups of sepsis patients, which each exhibited different tendencies toward certain medical conditions and different sepsis mortality rates:
- Alpha — This was the largest phenotype group, representing 33 percent of patients. The group had the lowest rates of abnormal laboratory tests, organ dysfunction and in-hospital mortality (2 percent).
- Beta — Twenty-seven percent of patients had this phenotype. This group included the most patients with chronic illnesses and renal dysfunction, as well as older patients.
- Gamma — Another 27 percent of patients had this phenotype. This group included patients with high inflammation and pulmonary dysfunction.
- Delta — This was the smallest group, representing 13 percent of patients who had higher rates of liver dysfunction and shock. Patients in this group also had the highest in-hospital mortality at 32 percent.
Using this information, clinicians could identify patients' sepsis types upon their arrival at emergency medical departments. This would help physicians treat sepsis patients earlier and allow for clinical trials tailored to specific sepsis types, the authors concluded.