Introduction
The article by Chanas et al. examines the impact of a locally implemented best-practice advisory (BPA) on the speed of an antibiotic response in patients with the risk of septic shock. In general, this program evaluates four vital signs of patients in surgical intensive care units (ICU) – temperature, heart rate, respiratory rate, and white blood cells. Depending on the assessment, the BPA determines whether the patients are at immediate risk of septic shock and automatically notifies clinical professionals who administer antibiotics. The authors have found that this approach had a slight positive correlation with the speed of the response. However, the BPA assessment failed to achieve statistical significance in the sample that included both manual and automatic notifications. In summary, the implementation of the BPA had a positive impact on the speed of response; nevertheless, its uses were most effective as a complementary decision-making and evaluation tool instead of an absolutely accurate assessment of septic shock risk.
Discussion
The current article analyzes the effectiveness of a standardized approach to sepsis-related conditions and processes. Namely, the authors have implemented a customized protocol that proposes a streamlined action plan based on SIRS criteria (patients’ vital signs) and demographic parameters. Moreover, the outline concerns the whole organizational culture and involves the comprehensive cooperation of the emergency department (ED) and sepsis alert teams. Additionally, education of nurses relating to the dangers of sepsis and the necessary measures is critical in this context. The authors tested this methodology at a local hospital over a year period, revealing that the implemented protocol has positively impacted the speed of sepsis identification and antibiotic/fluid administration. Moreover, this approach has significantly decreased overall mortality, while the average length of stay (LOS) was relatively unaffected. As a result, the authors have successfully confirmed the high effectiveness of a complex approach that focuses on standardized protocols and in-house education and its positive impact on sepsis outcomes.
The article by Gatewood et al. is one of the earlier studies (2015) that examines the impact of a structured protocol on sepsis outcomes. The author’s methodology consists of three primary steps – manual screening, automatic ‘sepsis alert,’ and generated suggestions on treatment. The initial assessment, similar to other frameworks, evaluates the patients’ SIRS criteria to determine the risk of sepsis-related complications. Consequently, the program estimates the measurements and alerts the team of potential concerns for infection. Lastly, if the working clinician recognizes the alert, the system provides general recommendations on treatment. The proposed framework is a relatively simple but effective method to emphasize potential risks of sepsis-related complications and alert the team. As the authors have duly noted, this approach was underused (in 2015) and deserved more attention. Hence, the current article could be considered one of the initial studies that revealed the positive impact of early screenings and automated alert processes on sepsis outcomes among patients in ICUs.
The current article explains the benefits of automated systems on sepsis identification based on the developed Modified Early Warning System (MEWS). Namely, the framework estimates patients’ vital signs, assigning the measurements according to two thresholds. The first tier exposes the risks of sepsis, additionally alerting nurses for manual assessment. The second-tier threshold implies that the MEWS measurements are exceedingly high, and it is critical to notify the charge nurse and the rapid response team. The authors have found that this two-tiered approach was immensely effective in early identification, correctly determining eleven cases of sepsis out of twelve alerts. In summary, the implementation of automated systems, such as MEWS, can be helpful in preventing severe cases of sepsis and improving overall patient health outcomes.
The systematic review by Lee et al. has examined forty-four articles to determine the effectiveness of predictive analytics based on electronic health records (EHR) data on clinical outcomes. This framework investigates patients’ prior medical conditions to evaluate the possibility of further complications in the future. The authors have revealed that, in the case of sepsis (16% of articles), predictive analytics can be highly effective in estimating the possibility of sepsis-related issues in patients. However, Lee et al. have not specified study designs in the systematic reviews. In this sense, the predictive models include EHR analysis, early identification, machine learning algorithms, and other types of custom models. This approach could potentially alter the results; nevertheless, it is evident that predictive analytics have a positive impact on clinical outcomes, and hospitals should implement more EHR-based frameworks.
The current article emphasizes machine learning as an effective alternative to determine medical conditions and improve the overall accuracy of diagnoses. The authors have conducted fifteen interviews with working professionals (ED physicians and nurses) concerning the benefits of machine learning in the early identification of sepsis. All interviewees have attended the Sepsis Watch (machine learning program) quality improvement initiative and are knowledgeable in innovative approaches in healthcare. As a result, the authors have confirmed a mostly positive reaction to the Sepsis Watch in the early identification of sepsis.
Conclusion
Nevertheless, many clinicians were hesitant and demonstrated perceived uncertainty concerning the machine learning algorithms. The authors note that this drawback of the system is related to the insufficient exposure of nurses to innovative approaches. Ultimately, Sandhu et al. conclude that additional research is required to understand how to remove perceived barriers in the implementation of machine learning algorithms and other innovative programs.
References
Chanas, T., Volles, D., Sawyer, R., & Mallow-Corbett, S. (2019). Analysis of a new best-practice advisory on time to initiation of antibiotics in surgical intensive care unit patients with septic shock. Journal of the Intensive Care Society, 20(1), 34-39. Web.
Ferreira, T. B., Diaz, Y., Beg, R., Wawrzyniak, A., Brito, Y., Burik, Y.,… & Sneij, W. (2017). Before and after standardizing the controversial sepsis resuscitation bundle in a large hybrid academic center. Journal of Emergency and Critical Care Medicine, 1(19). Web.
Gatewood, M. O. K., Wemple, M., Greco, S., Kritek, P. A., & Durvasula, R. (2015). A quality improvement project to improve early sepsis care in the emergency department. BMJ Quality & Safety, 24(12), 787-795. Web.
Horton, D. J., & Graves, K. (2016). Building a sepsis alert system. Journal of Hospital Medicine, 11(1).
Lee, T. C., Shah, N. U., Haack, A., & Baxter, S. L. (2020). Clinical implementation of predictive models embedded within electronic health record systems: A systematic review. Informatics, 7(25). Web.
Sandhu, S., Lin, A. L., Brajer, N., Sperling, J., Ratliff, W., Bedoya, A. D.,… & Sendak, M. P. (2020). Integrating a machine learning system into clinical workflows: Qualitative study. Journal of Medical Internet Research, 22(11). Web.