The Goldilocks Effect in the ICU-When the Data Speak, but Not the Truth

Timing and Spectrum of Antibiotic Treatment for Suspected Sepsis and Septic Shock: Why so Controversial?

Sepsis guidelines and mandates encourage increasingly aggressive time-to-antibiotic targets for broad-spectrum antimicrobials for suspected sepsis and septic shock. This has caused considerable controversy due to weaknesses in the underlying evidence and fear that overly strict antibiotic deadlines may harm patients by perpetuating or escalating overtreatment. Indeed, a third or more of patients currently treated for sepsis and septic shock have noninfectious or nonbacterial conditions. These patients risk all the potential harms of antibiotics without their possible benefits. Updated Surviving Sepsis Campaign guidelines now emphasize the importance of tailoring antibiotics to each patient's likelihood of infection, risk for drug-resistant pathogens, and severity-of-illness.

Oxygen administration during surgery and postoperative organ injury: observational cohort study

OBJECTIVE

To examine whether supraphysiological oxygen administration during surgery is associated with lower or higher postoperative kidney, heart, and lung injury.

DESIGN

Observational cohort study.

SETTING

42 medical centers across the United States participating in the Multicenter Perioperative Outcomes Group data registry.

PARTICIPANTS

Adult patients undergoing surgical procedures ≥120 minutes' duration with general anesthesia and endotracheal intubation who were admitted to hospital after surgery between January 2016 and November 2018.

INTERVENTION

Supraphysiological oxygen administration, defined as the area under the curve of the fraction of inspired oxygen above air (21%) during minutes when the hemoglobin oxygen saturation was greater than 92%.

MAIN OUTCOMES

Primary endpoints were acute kidney injury defined using Kidney Disease Improving Global Outcomes criteria, myocardial injury defined as serum troponin >0.04 ng/mL within 72 hours of surgery, and lung injury defined using international classification of diseases hospital discharge diagnosis codes.

RESULTS

The cohort comprised 350 647 patients with median age 59 years (interquartile range 46-69 years), 180 546 women (51.5%), and median duration of surgery 205 minutes (interquartile range 158-279 minutes). Acute kidney injury was diagnosed in 19 207 of 297 554 patients (6.5%), myocardial injury in 8972 of 320 527 (2.8%), and lung injury in 13 789 of 312 161 (4.4%). The median fraction of inspired oxygen was 54.0% (interquartile range 47.5%-60.0%), and the area under the curve of supraphysiological inspired oxygen was 7951% min (5870-11 107% min), equivalent to an 80% fraction of inspired oxygen throughout a 135 minute procedure, for example. After accounting for baseline covariates and other potential confounding variables, increased oxygen exposure was associated with a higher risk of acute kidney injury, myocardial injury, and lung injury. Patients at the 75th centile for the area under the curve of the fraction of inspired oxygen had 26% greater odds of acute kidney injury (95% confidence interval 22% to 30%), 12% greater odds of myocardial injury (7% to 17%), and 14% greater odds of lung injury (12% to 16%) compared with patients at the 25th centile. Sensitivity analyses evaluating alternative definitions of the exposure, restricting the cohort, and conducting an instrumental variable analysis confirmed these observations.

CONCLUSIONS

Increased supraphysiological oxygen administration during surgery was associated with a higher incidence of kidney, myocardial, and lung injury. Residual confounding of these associations cannot be excluded.

TRIAL REGISTRATION

Open Science Framework osf.io/cfd2m.

Surviving Sepsis Guideline–Directed Fluid Resuscitation: An Assessment of Practice Patterns and Impact on Patient Outcomes

IMPORTANCE:

Aggressive fluid resuscitation remains a cornerstone of the Surviving Sepsis Campaign (SSC) guidelines, but there is growing controversy regarding the recommended 30 mL/kg IV fluid dosage. It is contended that, in selected patients, this volume confers an increased risk of volume overload without either concomitant benefit or strong evidence in support of the recommended IV fluid dosage.

OBJECTIVES:

Assessment of practice patterns and their impact on patient outcomes following the surviving sepsis guidelines for fluid resuscitation.

DESIGN:

Large, multisite retrospective cohort study.

SETTING AND PARTICIPANTS:

The retrospective study included all adult patients who presented to the emergency department at one of 19 different Mayo Clinic sites throughout the Midwest, Southeast, and Southwest from August 2018 to November 2020 with suspected sepsis.

MAIN OUTCOMES AND MEASURES:

Eight-thousand four-hundred fourteen patients suspected to have sepsis were assessed regarding fluid resuscitation and outcomes among patients receiving 30 mL/kg IV fluid dosing compared with patients who did not. Patient demographics and clinical information were collected via electronic health records. Patients were divided into two cohorts: those who received 0–29.9 mL/kg of IV fluid and those who received 30.0+ mL/kg of IV fluid. Statistical analyses were performed to evaluate the impact of fluid dose on in-hospital death, 30-day mortality, ICU admission after diagnosis, dialysis initiation after diagnosis, ventilator use, vasopressor use, as well as ICU and hospital length of stay.

RESULTS:

We observed lower in-hospital mortality and 30-day mortality risk in the 30+ mL/kg dosing group. Increased fluid dosage did, however, carry a much greater chance of ICU admission. Most patients (72% after propensity score weighting) in our population received less than 30 mL/kg fluid (based on ideal body weight).

CONCLUSIONS AND RELEVANCE:

IV fluid dosing for sepsis resuscitation greater than 30 mL/kg was associated with decreased risk of in-hospital mortality, 30-day mortality, and reduced risk of requiring mechanical ventilation. Our data does ultimately seem to support the SSC recommendation.

Modelling hospital outcome: problems with endogeneity

Background

Mortality modelling in the critical care paradigm traditionally uses logistic regression, despite the availability of estimators commonly used in alternate disciplines. Little attention has been paid to covariate endogeneity and the status of non-randomized treatment assignment. Using a large registry database, various binary outcome modelling strategies and methods to account for covariate endogeneity were explored.

Methods

Patient mortality data was sourced from the Australian & New Zealand Intensive Society Adult Patient Database for 2016. Hospital mortality was modelled using logistic, probit and linear probability (LPM) models with intensive care (ICU) providers as fixed (FE) and random (RE) effects. Model comparison entailed indices of discrimination and calibration, information criteria (AIC and BIC) and binned residual analysis. Suspect covariate and ventilation treatment assignment endogeneity was identified by correlation between predictor variable and hospital mortality error terms, using the Stata™ “eprobit” estimator. Marginal effects were used to demonstrate effect estimate differences between probit and “eprobit” models.

Results

The cohort comprised 92,693 patients from 124 intensive care units (ICU) in calendar year 2016. Patients mean age was 61.8 (SD 17.5) years, 41.6% were female and APACHE III severity of illness score 54.5(25.6); 43.7% were ventilated. Of the models considered in predicting hospital mortality, logistic regression (with or without ICU FE) and RE logistic regression dominated, more so the latter using information criteria indices. The LPM suffered from many predictions outside the unit [0,1] interval and both poor discrimination and calibration. Error terms of hospital length of stay, an independent risk of death score and ventilation status were correlated with the mortality error term. Marked differences in the ventilation mortality marginal effect was demonstrated between the probit and the "eprobit" models which were scenario dependent. Endogeneity was not demonstrated for the APACHE III score.

Conclusions

Logistic regression accounting for provider effects was the preferred estimator for hospital mortality modelling. Endogeneity of covariates and treatment variables may be identified using appropriate modelling, but failure to do so yields problematic effect estimates.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12874-021-01251-8.