Gilles Clermont

Intensive care unit safety culture and outcomes: a US multicenter study

OBJECTIVE Safety culture may influence patient outcomes, but evidence is limited. We sought to determine if intensive care unit (ICU) safety culture is independently associated with outcomes. DESIGN Cohort study combining safety culture survey data with the Project IMPACT Critical Care Medicine (PICCM) clinical database. SETTING Thirty ICUs participating in the PICCM database. PARTICIPANTS A total of 65 978 patients admitted January 2001-March 2005. INTERVENTIONS None. MAIN OUTCOME MEASURES Hospital mortality and length of stay (LOS). METHODS From December 2003 to April 2004, we surveyed study ICUs using the Safety Attitudes Questionnaire-ICU version, a validated instrument that assesses safety culture across six factors. We calculated factor mean and percent-positive scores (% respondents with mean score > or =75 on a 0-100 scale) for each ICU, and generated case-mix adjusted, patient-level, ICU-clustered regression analyses to determine the independent association of safety culture and outcome. RESULTS We achieved a 47.9% response (2103 of 4373 ICU personnel). Culture scores were mostly low to moderate and varied across ICUs (range: 13-88, percent-positive scores). After adjustment for patient, hospital and ICU characteristics, for every 10% decrease in ICU perceptions of management percent-positive score, the odds ratio for hospital mortality was 1.24 (95% CI: 1.07-1.44; P = 0.005). For every 10% decrease in ICU safety climate percent-positive score, LOS increased 15% (95% CI: 1-30%; P = 0.03). Sensitivity analyses for non-response bias consistently associated safety climate with outcome, but also yielded some counterintuitive results. CONCLUSION In a multicenter study conducted in the USA, perceptions of management and safety climate were moderately associated with outcomes. Future work should further develop methods of assessing safety culture and association with outcomes.

Classifying AKI by Urine Output versus Serum Creatinine Level

Severity of AKI is determined by the magnitude of increase in serum creatinine level or decrease in urine output. However, patients manifesting both oliguria and azotemia and those in which these impairments are persistent are more likely to have worse disease. Thus, we investigated the relationship of AKI severity and duration across creatinine and urine output domains with the risk for RRT and likelihood of renal recovery and survival using a large, academic medical center database of critically ill patients. We analyzed electronic records from 32,045 patients treated between 2000 and 2008, of which 23,866 (74.5%) developed AKI. We classified patients by levels of serum creatinine and/or urine output according to Kidney Disease Improving Global Outcomes staging criteria for AKI. In-hospital mortality and RRT rates increased from 4.3% and 0%, respectively, for no AKI to 51.1% and 55.3%, respectively, when serum creatinine level and urine output both indicated stage 3 AKI. Both short- and long-term outcomes were worse when patients had any stage of AKI defined by both criteria. Duration of AKI was also a significant predictor of long-term outcomes irrespective of severity. We conclude that short- and long-term risk of death or RRT is greatest when patients meet both the serum creatinine level and urine output criteria for AKI and when these abnormalities persist.

Agent-based model of inflammation and wound healing : insights into diabetic foot ulcer pathology and the role of transforming growth factor-β1

Inflammation and wound healing are inextricably linked and complex processes, and are deranged in the setting of chronic, nonhealing diabetic foot ulcers (DFU). An ideal therapy for DFU should both suppress excessive inflammation while enhancing healing. We reasoned that biological simulation would clarify mechanisms and help refine therapeutic approaches to DFU. We developed an agent‐based model (ABM) capable of reproducing qualitatively much of the literature data on skin wound healing, including changes in relevant cell populations (macrophages, neutrophils, fibroblasts) and their key effector cytokines (tumor necrosis factor‐α [TNF], interleukin [IL]‐1β, IL‐10, and transforming growth factor [TGF]‐β1). In this simulation, a normal healing response results in tissue damage that first increases (due to wound‐induced inflammation) and then decreases as the collagen levels increase. Studies by others suggest that diabetes and DFU are characterized by elevated TNF and reduced TGF‐β1, although which of these changes is a cause and which one is an effect is unclear. Accordingly, we simulated the genesis of DFU in two ways, either by (1) increasing the rate of TNF production fourfold or (2) by decreasing the rate of TGF‐β1 production 67% based on prior literature. Both manipulations resulted in increased inflammation (elevated neutrophils, TNF, and tissue damage) and delayed healing (reduced TGF‐β1 and collagen). Our ABM reproduced the therapeutic effect of platelet‐derived growth factor/platelet releasate treatment as well as DFU debridement. We next simulated the expected effect of administering (1) a neutralizing anti‐TNF antibody, (2) an agent that would increase the activation of endogenous latent TGF‐β1, or (3) latent TGF‐β1 (which has a longer half‐life than active TGF‐β1), and found that these therapies would have similar effects regardless of the initial assumption of the derangement that underlies DFU (elevated TNF vs. reduced TGF‐β1). In silico methods may elucidate mechanisms of and suggest therapies for aberrant skin healing.

Predicting hospital mortality for patients in the intensive care unit: a comparison of artificial neural networks with logistic regression models.

ObjectiveLogistic regression (LR), commonly used for hospital mortality prediction, has limitations. Artificial neural networks (ANNs) have been proposed as an alternative. We compared the performance of these approaches by using stepwise reductions in sample size. DesignProspective cohort study. SettingSeven intensive care units (ICU) at one tertiary care center. PatientsPatients were 1,647 ICU admissions for whom first-day Acute Physiology and Chronic Health Evaluation III variables were collected. InterventionsNone. Measurements and Main Results We constructed LR and ANN models on a random set of 1,200 admissions (development set) and used the remaining 447 as the validation set. We repeated model construction on progressively smaller development sets (800, 400, and 200 admissions) and retested on the original validation set (n = 447). For each development set, we constructed models from two LR and two ANN architectures, organizing the independent variables differently. With the 1,200-admission development set, all models had good fit and discrimination on the validation set, where fit was assessed by the Hosmer-Lemeshow C statistic (range, 10.6–15.3;p ≥ .05) and standardized mortality ratio (SMR) (range, 0.93 [95% confidence interval, 0.79–1.15] to 1.09 [95% confidence interval, 0.89–1.38]), and discrimination was assessed by the area under the receiver operating characteristic curve (range, 0.80–0.84). As development set sample size decreased, model performance on the validation set deteriorated rapidly, although the ANNs retained marginally better fit at 800 (best C statistic was 26.3 [p = .0009] and 13.1 [p = .11] for the LR and ANN models). Below 800, fit was poor with both approaches, with high C statistics (ranging from 22.8 [p < .004] to 633 [p < .0001]) and highly biased SMRs (seven of the eight models below 800 had SMRs of <0.85, with an upper confidence interval of <1). Discrimination ranged from 0.74 to 0.84 below 800. ConclusionsWhen sample size is adequate, LR and ANN models have similar performance. However, development sets of ≤800 were generally inadequate. This is concerning, given typical sample sizes used for individual ICU mortality prediction.

Growth of intensive care unit resource use and its estimated cost in Medicare

Objective:The past 10–15 yrs brought significant changes in the United States healthcare system. Effects on Medicare intensive care unit use and costs are unknown. Intensive care unit costs are estimated using the Russell equation with a ratio of intensive care unit to floor cost per day, or “R value,” of 3, which may no longer be valid. We sought to determine contemporary Medicare intensive care unit resource use, costs, and R values; whether these vary by patient and hospital characteristics; and the impact of updated values on estimated intensive care unit costs. Design:Retrospective analysis of Medicare Inpatient Prospective Payment System hospitalizations from 1994 to 2004 using Medicare Provider Analysis and Review files. Setting:All nonfederal acute care US hospitals paid through the Inpatient Prospective Payment System. Subjects:Inpatient prospective payment system hospitalizations from 1994 to 2004 (n = 121,747, 260). Interventions:None. Measurements and Main Results:We examined resource use and costs (adjusted to y2004

A Patient-Specific in silico Model of Inflammation and Healing Tested in Acute Vocal Fold Injury

), calculating intensive care unit and floor costs directly and using these to generate year-specific R values. By 2004, 33% of Medicare hospitalizations had intensive care unit or coronary care unit care, with more than half of the increase in total hospitalizations because of additional intensive care unit hospitalizations. Adjusted intensive care unit cost per day remained stable (

IN SILICO MODELS OF ACUTE INFLAMMATION IN ANIMALS

2,616 vs.

Healthcare costs and long-term outcomes after acute respiratory distress syndrome: A phase III trial of inhaled nitric oxide*

2,575; 1994 vs 2004), yet adjusted floor cost per day rose substantially (

The role of initial trauma in the host's response to injury and hemorrhage: insights from a correlation of mathematical simulations and hepatic transcriptomic analysis.

1,027 vs.

Bridging the gap between systems biology and medicine

1,488) driven by decreased floor length of stay. Annual adjusted Medicare intensive care unit costs increased 36% to