Thank you, Dr. Kaiser, and thanks for the audience being here so late. We have very few people here, so. I'm Abdul. I'm a hospitalist at Mayo Clinic Health System, so I will be presenting our study. I want to thank you, our co-authors. So I don't have any discloyers. A couple of background slides before we dive into the study details. So we all are aware that when patients are admitted to the ICU, unfortunately, they face a lot of challenges. They suffer from sleep deprivation, lack of privacy, lack of independence, fear of dying. If we can prevent ICU admission, we may be able to limit psychological, physical, and financial burdens associated with those admissions. So keeping that in mind, so we also know that there has been a constant increase in the demand for critical care services, and we have been trying to meet that demand by increasing number of ICU beds and ICU staff. But still, there is significant shortage of resources, and that was very obvious during the COVID-19 pandemic. Sepsis still remains among the top five ICU admission diagnosis, even if patients don't have any respiratory needs for the ICU level of care. So over the last two to three decades, we have made significant improvement in sepsis-related management, and that improvement has been mainly because of our approach to resuscitation and physiologic support and better understanding the effects of our strategies. So similarly, if we have better strategies, we may be able to decrease burden on critical care services, and one such strategy could be to identify a group of patients who could be managed outside the ICU, and one such group could be low-risk septic shock patients whose only indication for admission to the ICU is low-dose short-term vasopressors, and they don't have any other indication. But unfortunately, we don't have that kind of a strategy to identify those patients early in their disease process. So the objective of our study was to identify a subset of septic shock patients who would require less than 24-hour vasopressor, and they don't need any other intervention requiring ICU level of care. We also aim to develop a prediction model to identify those patients and manage them via approach, a minimally invasive approach. What do we mean by minimally invasive approaches? To identify a subset of patients, as I mentioned in the previous slide, and prevent them from going to the ICU and manage them appropriately outside the ICU by fluid resuscitation, antimicrobials, and providing them vasopressors via peripheral venous access, avoiding invasive monitoring and monitoring them adequately via non-invasive measures. So we did a retrospective analysis of our institutional database of patients with sepsis. We focused on patients who were admitted to medical ICU from ER. We excluded patients who were already hospitalized, surgical patients, patients who were already on invasive or non-invasive mechanical ventilation before ICU admission. Candidates for minimally invasive approach were those who had a septic shock and had stayed in ICU for less than 48 hours. They did not require advanced respiratory support before their admission to the ICU. Also, during the whole hospitalization, they did not require invasive or non-invasive mechanical ventilation. They did not require continuous renal replacement therapy. And they were alive at the time of hospital discharge. So from our sepsis database, we had 1,795 MECU admissions, 184 were those with less than 48 hours of ICU stay, 109 were direct admits from ER to MECU, and 106, we had those met our criteria for minimally invasive approach. We had 97 patients for the comparative analysis. So none of the patients from both groups were on any advanced respiratory support at the time of admission to the ICU. Patients in the minimally invasive sepsis group did not require any advanced respiratory support during the whole hospitalization. But some of the patients in the comparative group did require invasive or non-invasive ventilation. At baseline, patients in the minimally invasive group were younger. They had lower lactate, creatinine, BUN, respiratory rate, and white blood cell counts. But they had a higher temperature. Both groups were similar otherwise. The labs and the vitals in this table were taken at the time of their ER disposition or ICU admission, which means that patient came to the ER, they got their initial treatment, and then they were ready to be admitted to the ICU. At that time, we had these labs and the vitals. To develop the prediction model, predictive variables were identified through logistic regression. And cutoff points for each variable were determined. Based on their clinical significance and statistical significance, each variable was assigned a point, which were translated into an eight-point scoring system. Model discrimination was assessed by a receiver operating characteristic curve, which yielded an AUC of 79%, giving reasonable accuracy to the model. Model fit and calibrations were also assessed. Model fit was assessed using a Hosmer-Lemsch-Agonis fit test. And it showed a P value of 0.94, which means that if there were any difference between the observed versus predicted risk, those were not statistically significant. Prediction plot was also reasonably well as shown in the graph for each point of the scoring system. At a cutoff value of 3, the model has a negative predictive value of 90%, which means a patient with a score of 3 or less, they could be considered for minimally invasive approach, which means they could avoid ICU admission and could be managed outside of the ICU via peripheral vasopressors and the rest of the measures that I mentioned. Our study is based on well-defined criteria. We clearly defined our target patient population. The model is based on the labs and vitals, which are readily available at the time of the ED disposition, which can aid into decision-making and clinical judgment. Study does have limitations because it's a single-center study. Results may not be generalizable. Model hasn't been validated yet. And similar to other models based on EHR database, our model also lacks the part of the clinical judgment at the time of the ED disposition. So in conclusion, a significant minority of sepsis patients who are at low risk of deterioration can be potentially managed outside the ICU. Once validated in an independent prospective sample, our prediction model can be used to identify candidates for such an approach, and they could avoid ICU admissions. Thank you.

septic shock patients

managing outside of ICU

vasopressor treatment

prediction model

low-risk septic shock patients