No abstract

Background Patients with hematologic malignancies who are admitted to hospital are at increased risk of deterioration and death. Rapid response systems (RRSs) respond to hospitalized patients who clinically deteriorate. We sought to describe the characteristics and outcomes of hematologic oncology inpatients requiring rapid response system (RRS) activation, and to determine the prognostic accuracy of the SIRS and qSOFA criteria for in-hospital mortality of hematologic oncology patients with suspected infection. Methods We used registry data from two hospitals within The Ottawa Hospital network, between 2012 and 2016. Consecutive hematologic oncology inpatients who experienced activation of the RRS were included in the study. Data was gathered at the time of RRS activation and assessment. The primary outcome was in-hospital mortality. Logistical regression was used to evaluate for predictors of in-hospital mortality. Results We included 401 patients during the study period. In-hospital mortality for all included patients was 41.9% (168 patients), and 145 patients (45%) were admitted to ICU following RRS activation. Among patients with suspected infection at the time of RRS activation, Systemic Inflammatory Response Syndrome (SIRS) criteria had a sensitivity of 86.9% (95% CI 80.9–91.6) and a specificity of 38.2% (95% CI 31.9–44.8) for predicting in-hospital mortality, while Quick Sequential Organ Failure Assessment (qSOFA) criteria had a sensitivity of 61.9% (95% CI 54.1–69.3) and a specificity of 91.4% (95% CI 87.1–94.7). Factors associated with increased in-hospital mortality included transfer to ICU after RRS activation (adjusted odds ratio [OR] 3.56, 95% CI 2.12–5.97) and a higher number of RRS activations (OR 2.45, 95% CI 1.63–3.69). Factors associated with improved survival included active malignancy treatment at the time of RRS activation (OR 0.54, 95% CI 0.34–0.86) and longer hospital length of stay (OR 0.78, 95% CI 0.70–0.87). Conclusions Hematologic oncology inpatients requiring RRS activation have high rates of subsequent ICU admission and mortality. ICU admission and higher number of RRS activations are associated with increased risk of death, while active cancer treatment and longer hospital stay are associated with lower risk of mortality. Clinicians should consider these factors in risk-stratifying these patients during RRS assessment. Electronic supplementary material The online version of this article (10.1186/s13054-019-2568-5) contains supplementary material, which is available to authorized users.

Background Diabetic ketoacidosis (DKA) and hyperglycemic hyperosmolar state (HHS) are life-threatening complications of diabetes mellitus which require prompt treatment with large volume crystalloid fluid administration. A variety of crystalloid fluids is currently available for use and differs in their composition and ion concentrations. While there are potential pros and cons for different crystalloid fluids, it remains unknown if any particular fluid confers a clinical outcome benefit over others in the treatment of hyperglycemic emergencies. Methods A systematic search of MEDLINE, Embase, and the Cochrane Library of Systematic Reviews will be conducted to identify eligible studies, which will include observational and interventional studies involving adult and pediatric patients admitted to the hospital with either DKA or HHS. The interventions will include intravenous treatment with 0.9% saline versus other buffered (Ringer’s lactate, Hartmann’s, etc.), and non-buffered (0.45% saline) crystalloid fluids. The primary outcome is mortality at the latest follow-up time point. Secondary outcomes will include mortality at specific time points, length of hospital stay, development of acute kidney injury, requirement for renal replacement therapy, altered level of consciousness, and the time to normalization of several serum biochemical parameters. Where appropriate, meta-analyses will be performed for the outcomes and conducted separately for adult and pediatric patient populations. Discussion DKA and HHS are dangerous complications of diabetes mellitus and account for significant morbidity and mortality. Given the importance of crystalloid fluid administration in the management of these conditions, a systematic synthesis of the existing evidence base will identify potential evidence gaps and may help guide future clinical practice.

Objectives: Machine learning models have been used to predict mortality among patients requiring rapid response team activation. The goal of our study was to assess the impact of adding laboratory values into the model. Design: A gradient boosted decision tree model was derived and internally validated to predict a primary outcome of in-hospital mortality. The base model was then augmented with laboratory values. Setting: Two tertiary care hospitals within The Ottawa Hospital network. Patients: Inpatients over the age of 18 years who experienced a rapid response team activation between January 1, 2015, and May 31, 2016. Interventions: None. Measurements and Main Results: A total of 2,061 rapid response team activations occurred during the study period. The in-hospital mortality rate was 29.4%. Patients who died were older (median age, 72 vs 68 yr; p < 0.001), had a longer length of stay (length of stay) prior to rapid response team activation (4 vs 2 d; p < 0.001), and more often had respiratory distress (31% vs 22%; p < 0.001). Our base model without laboratory values performed with an area under the receiver operating curve of 0.71 (95% CI, 0.71–0.72). When the base model was augmented with laboratory values, the area under the receiver operating curve improved to 0.77 (95% CI, 0.77–0.78). Important mortality predictors in the base model were age, estimated ratio of Pao 2 to Fio 2 (calculated using oxygen saturation and estimated Fio 2), length of stay prior to rapid response team activation, and systolic blood pressure. Conclusions: Machine learning models can identify rapid response team patients at a high risk of mortality and potentially supplement clinical decision making. Incorporating laboratory values into model development significantly improved predictive performance in this study.