Background: Sleep disruptions for routine care, such as vital sign assessments, are common during hospitalization and are associated with negative health outcomes and patient satisfaction. While higher risk patients may benefit from increased monitoring at night, the tradeoff is less obvious for lower risk patients. We hypothesized that assigning overnight vital sign assessment based on patient acuity was feasible and would result in improved sleep and increased patient satisfaction, without an increased risk of adverse events.
Methods: We conducted a randomized controlled trial of adult surgical inpatients to test this hypothesis. Patients were randomized by random number generator to intervention versus control. The intervention consisted of placing a sign on the patient’s hospital room door instructing the nursing staff to avoid nighttime vital signs in patients on nights they were deemed low risk by their 10 pm eCART score, an electronic early warning analytic that uses vital signs and laboratory data to determine the risk of cardiac arrest, ICU transfer or death in the next eight hours. The control group received standard care, which included vital signs every four hours regardless of acuity. A continuous, under the mattress vital sign monitoring device (EarlySense Patient Monitoring System, EarlySense Inc., Waltham, MA) was used as a safety precaution on all patients during the study. All patients were surveyed regarding their subjective sleep quality and satisfaction the following day regardless of randomization group or acuity the night before using the Karolinska Sleep Log, Sleep Disruptions and Noises Questionnaire and Patient Satisfaction Questionnaire, which have all been previously validated for the inpatient setting. Likert based responses were compared using a Wilcoxon Rank Sum, with two-sided p-value <0.05 deemed statistically significant.
Results: A total of 829 patients were screened for randomization, of which 166 were successfully randomized (see Figure). Fifteen patients were lost to follow-up after randomization due to withdrawal from the study (one in each group) or being discharged prior to survey administration (8 intervention; 7 control), leaving 74 in the intervention group and 75 in the control for analysis. Mean age was 48 (SD 18) years, 72% were white race and 52% were female sex, without any significant difference by group. Of those, 116 met low-risk criteria the first night after randomization (59 intervention; 57 control). Intervention patients that were low-risk were more likely to report vital signs not being disruptive the night before (p<0.001) than their low risk control group counterparts (see Table). There was a trend toward these same patients reporting sleeping more soundly (p=0.1), having sleep be less disturbed by noise (p=0.05), and increased satisfaction with nursing care over the preceding 24 hours (p=0.16). Only one patient had an adverse event during the study period (an ICU transfer within 24 hours of randomization). However, the patient had an elevated eCART score and therefore was ineligible for the intervention, despite being randomized to that group.
Conclusions: Acuity based vital sign assessment using an electronic early warning score is feasible and associated with decreased sleep interruptions for vital sign assessments on nights the patients are deemed low-risk. This practice does not appear to be associated with patient harm, and may result in improved sleep and satisfaction with nursing care. A larger study is needed to better quantify the benefits.