Background: Hospitalized patients are often disrupted at night for routine medical care, some of which is unnecessary. These sleep disruptions have implications for patient satisfaction, delirium, mobility, immune status, as well as hospital outcomes such as length of stay and readmissions. Interventions to improve inpatient sleep would benefit from the ability to objectively measure sleep, but existing methods to measure sleep are resource intensive and impose a burden on patients.
Purpose: We developed a new measure of inpatient sleep using data available in the electronic health record (EHR).
Description: Metric DevelopmentWe identified all adult patients discharged from the General Medicine Service of an 800-bed academic urban teaching hospital between April 1 and June 30, 2018, using patient-nights as a unit of analysis (n= 6,825). An EHR-derived variable—sleep opportunity (“SLOP”)—was defined as the longest span of uninterrupted time between the hours of 10pm and 6am, when we presumed most patients would be sleeping. SLOP was determined from data on timestamped interventions recorded in the EHR that correspond with a member of the healthcare team physically entering the patient room. These included: vital signs, scheduled medications, dose or bag changes for continuous intravenous infusions, blood draws, finger stick blood glucose, and nursing flowsheet data such as patient weight. We excluded remote vital sign monitoring by telemetry or pulse oximetry, as well as patient triggered activities including “as needed” medications as these interruptions were not within the control of the primary clinician.
Metric analysis
During the 3-month baseline period, the median SLOP was 4.43 hours, with an average number of 4.38 interruptions. Patients on vital sign checks that skip overnight measurements—”sleep promoting vitals”—had an average of 65.4 minutes more SLOP, suggesting metric plausibility.
Conclusions: We developed a completely EHR-derived metric to help approximate patient sleep based on overnight interruptions. As a next step, we are validating this measure against wrist actigraphy and patient reported outcomes. More research is needed, but we believe that the use of EHR-derived measures to track patient-centered outcomes holds promise for hospital-based improvement work by providing a platform for easily tracking outcomes over time.