Construction and Validation of Risk Factors Prediction Model for Unplanned Readmission to the Surgical Intensive Care Unit: A Retrospective Study

In the high-stakes environment of a surgical intensive care unit, the moment a patient is deemed stable enough to leave is as critical as the surgery itself. These specialized wards are designed to support the body through its most vulnerable hours after an operation, monitoring vital signs and organ function with intense scrutiny. However, a patient's journey does not always end with a transfer to a regular hospital room. Sometimes, within days of leaving the intensive care unit, a patient's condition deteriorates, forcing an unplanned return. This event is not merely a logistical inconvenience; it is a serious warning sign that the patient's recovery was not as secure as hoped. Medical systems around the world track these returns closely because they are linked to higher risks of death, longer hospital stays, and a greater strain on medical resources. The challenge for doctors has long been to identify, before a patient walks out the door, who is likely to stumble and need to come back.

A team of researchers at a major hospital in Shanghai set out to solve this specific puzzle for patients who had undergone non-cardiac surgery. They looked back at the records of nearly 3,000 adults who had spent time in their surgical intensive care unit after operations on the liver, stomach, intestines, or other organs, but not the heart. By carefully reviewing the data from these patients, the researchers sought to find the specific clues that appeared in the medical records of those who were readmitted versus those who recovered smoothly. They focused on the period just before a patient was discharged, looking for patterns in blood tests, vital signs, and scoring systems that measure how sick a person is. The goal was to build a tool that could take these scattered pieces of information and combine them into a single, clear prediction of risk.

The study revealed that unplanned returns to the intensive care unit were relatively rare, occurring in about 2.4 percent of the patients they studied. When these returns did happen, the most common reason was respiratory failure, meaning the patients struggled to breathe effectively once they were in the regular ward. To understand why this happened, the researchers compared the medical data of the patients who returned with those who did not. They found that several specific factors were consistently different in the group that came back. These included the patient's overall severity score upon entering the unit, a quick assessment of their early warning signs, and specific measurements taken right before they left the intensive care unit.

The researchers identified seven key variables that acted as independent predictors of a return. These included the patient's score on a system that evaluates acute physiology and chronic health, a score that tracks early warning signs like breathing rate and consciousness, and the time it takes for their blood to clot. They also looked at the level of creatinine in the blood, which indicates how well the kidneys are working, the patient's heart rate, a measure of how efficiently oxygen moves from the lungs into the blood, and the level of hemoglobin, which carries oxygen throughout the body. The analysis showed that higher scores for illness severity and early warning signs, along with slower blood clotting, higher heart rates, and lower levels of oxygen efficiency or hemoglobin, all increased the likelihood of a patient needing to return to intensive care.

Using these seven factors, the team constructed a visual tool called a nomogram. This tool functions like a personalized calculator for doctors. To use it, a clinician would locate the patient's specific values for each of the seven factors on the chart, draw a line to a points scale, and add up the total points. The final sum points directly to a percentage that represents the patient's probability of being readmitted. The researchers tested this tool on the data they had, and it performed with high accuracy, correctly distinguishing between patients who would return and those who would not in the vast majority of cases. When they applied the tool to a separate group of patients to see if it held up, it continued to show strong reliability, correctly predicting outcomes in more than 85 percent of cases.

The study suggests that this tool can help medical teams make more informed decisions about when it is safe to discharge a patient. By quantifying the risk based on concrete data rather than just intuition, doctors can identify high-risk individuals before they leave the intensive care unit. For those patients, the tool might prompt a decision to keep them under observation for a little longer or to arrange for more intensive monitoring once they are in the regular ward. The researchers noted that their work was limited to a single hospital and focused only on non-heart surgeries, so the tool needs further testing in other settings to confirm its usefulness everywhere. However, the model provides a simple, individualized way to look at the complex mix of factors that determine a patient's recovery, offering a clearer path to preventing a dangerous and costly return to the intensive care unit.