All of Us Data Helps Better Predict Hospital Readmission for Patients with Sepsis

Last Reviewed: October 4, 2022

Key points

Researchers used All of Us data on social determinants of health to predict sepsis readmission in patients.
Social determinants of health that affect a person’s risk of hospital readmission include: delays in getting medical care, not having health insurance, unstable housing or employment, and lower education levels.
The diversity of the All of Us dataset was key in the success of the sepsis predictive models.

Sepsis is a serious and life-threatening medical emergency that happens when an infection spreads throughout the body and into the blood.

People who have been in the hospital with sepsis sometimes go back because their illness returns.

New insights from All of Us data found that researchers can better predict hospital readmission by looking at certain social and environmental factors, such as education, housing, employment, and access to care. These are known as social determinants of health. Older models for sepsis readmission included only demographics (such as age, gender, and race) and previous medical issues or health problems.

These new findings were published in the Journal of the American Medical Informatics Association.

Richer Data Yields Better Predictions

In this study, researchers created a new model to predict 30-day sepsis readmission. Within the Researcher Workbench, they examined data from nearly 9,000 participants with a history of sepsis. The researchers used electronic health record data (demographic and clinical information) from many different All of Us sites across the country. They also included participant responses to 88 survey questions about different social and environmental factors.

The researchers found several factors linked with higher chances of unplanned hospital readmission:

Delays in getting medical care due to lack of transportation
Lack of health insurance
Unstable housing or employment
Lower education and income levels

Men and Black or Asian patients were also more likely to need to return to the hospital than women or people of other races.

Informed Care for Sepsis Using Social Determinants of Health

This new model can aid health care staff in identifying patients who may be most at risk of readmission and finding ways to help. For example, some patients may benefit from additional follow-up after they leave the hospital. Case managers could arrange transportation to health check-ups for patients who need it, to prevent delays and improve access to medical care.

These results show the importance of diversity in All of Us. When research efforts include people from many different backgrounds, we can gather data points on a wider variety of disease outcomes. Researchers can then create models that apply to more people.

This study was based on early data from All of Us. Over time, more data will be available that can make more insights like this possible.