9 Are There Positive Deviant Providers in the Speed of Ordering Antibiotics for Sepsis?

Positive deviance (PD) is an application of complexity leadership theory that fosters organizational learning of difficult tasks. PD may also be able to avoid the increased burnout associated the top-down management tactics of traditional quality improvement (QI). We assessed whether PD may be an appropriate strategy to improve sepsis care by testing for the presence of heterogeneity in provider antibiotic ordering time (AOT) for severe sepsis or septic shock patients and identifying the positive deviant providers. We included all patients with severe sepsis or septic shock who presented to the University of Kansas Emergency Department over a 7-month period (September 2020 to March 2021), were cared for by categorical emergency medicine residents, and each member of the care team (resident physician, attending physician, and nurse) cared for at least three patients during the study period. First, we measured the percent heterogeneity (I2) of AOTs with separate random effects analyses for each personnel type (resident physician, attending physician, and nurse) on the patients’ health care teams. Then, the positive deviant personnel were identified as the individuals whose mean AOT and confidence interval were faster than the mean AOT for their personnel type. The results of the first analysis were displayed in forest plots. Second, to assess the contributions of each personnel type to the AOT, we performed regression analyses to measure the percent of variance (R2) of the AOTs attributable to each personnel type. The regression analyses were weighted by the inverse of the variance of the AOTs for each member of the patients’ care teams. The incremental contribution of each personnel type was measured by regression of their AOTs with the residual times of the regression of the dominant personnel. 371 patients were included. Random effects analyses showed statistically significant heterogeneity with “moderate” I2 values of 42%, 58%, and 32% for the residents, nurses, and attendings, respectively. There were 3, 6, and 2 positive deviants, respectively. Regression found that the residents, nurses, and attendings all made statistically significant contributions to the variation in AOT with percent R2s of 0.9%, 0.3%, and 0.3%, respectively. Regression of residuals showed significant, independent effects of the residents and nurses. The contributions to the variations are much higher when all ordering providers are studied (60.2% for ordering provider and 42.5% for nurse) We noted heterogeneity in mean AOT across both resident and attending physicians. In addition, the AOT for a given patient was associated with AOTs of other cases managed by the same care team. Within each personnel type, some providers on average demonstrated quicker AOTs compared to peers. Considering the practice environment is constant, the heterogeneity in mean AOT suggests differences in practice patterns. These results support the role of a positive deviance approach to QI with focus on the residents and nurses in identifying and disseminating best tactics within our emergency department.