Identifying the relationship between unstable vital signs and intensive care unit (ICU) readmissions: an analysis of 10-year of hospital ICU readmissions

Abstract

Advanced healthcare information systems capture extensive electronic healthcare records (EHR) to provide accurate healthcare delivery and make informed clinical and management decisions. EHR also include records of Intensive Care Unit (ICU) admissions, discharges, readmissions and more. This study aims to investigate trends and patterns from electronically recorded vital signs and identify the relationship between vital signs and ICU readmissions. After appropriate ethics approvals and permissions, we obtained access to the MIMIC-III dataset. A total of 150 patient data has been selected from the MIMIC-III dataset to evaluate the vital signs patterns, which included heart rate, respiratory rate, temperature for patients admitted twice in the ICU. Statistical analysis was conducted to identify the key patterns associated with the vital signs of the selected patient samples. The study findings indicate that the mean value of heart rate and respiratory rate was within the normal range for all incidents. However, there was a small difference between the readmission-day of the vital signs compared with the previous admission recordings. The finding also suggests that hospital length of stay (LOS) for patients was high for the second admission compared to the first admission. We identified a key relationship between the vital signs and ICU readmissions, it is evident that discharges with unstable vitals are directly linked to readmissions. Additionally, it is also observed that the LOS varies depending on the stability of the vital signs. In future, a larger sample with more variables are required, such as modifiable (medications, procedures, room temperature, discharge time) and non-modifiable (age, ethnicity, family history) which could explain the significance of longer LOS and its impact on patient-condition, readmission and mortality rate.