PROSAFE: a European endeavor to improve quality of critical care medicine in seven countries.

BACKGROUND: Long-lasting shared research databases are an important source of epidemiological information and can promote comparison between different healthcare services. Here we present PROSAFE, an advanced international research network in intensive care medicine, with the focus on assessing and improving the quality of care. The project involved 343 ICUs in seven countries. All patients admitted to the ICU were eligible for data collection. METHODS: The PROSAFE network collected data using the same electronic case report form translated into the corresponding languages. A complex, multidimensional validation system was implemented to ensure maximum data quality. Individual and aggregate reports by country, region, and ICU type were prepared annually. A web-based data-sharing system allowed participants to autonomously perform different analyses on both own data and the entire database. RESULTS: The final analysis was restricted to 262 general ICUs and 432,223 adult patients, mostly admitted to Italian units, where a research network had been active since 1991. Organization of critical care medicine in the seven countries was relatively similar, in terms of staffing, case mix and procedures, suggesting a common understanding of the role of critical care medicine. Conversely, ICU equipment differed, and patient outcomes showed wide variations among countries. CONCLUSIONS: PROSAFE is a permanent, stable, open access, multilingual database for clinical benchmarking, ICU self-evaluation and research within and across countries, which offers a unique opportunity to improve the quality of critical care. Its entry into routine clinical practice on a voluntary basis is testimony to the success and viability of the endeavor.

Mortality attributable to different Klebsiella susceptibility patterns and to the coverage of empirical antibiotic therapy: a cohort study on patients admitted to the ICU with infection.

PURPOSE: To evaluate the prognostic importance of different Klebsiella spp. sensitivity patterns: multi-susceptible Klebsiella (MS-K), extended-spectrum cephalosporin-resistant, but carbapenem-susceptible Klebsiella (ESCR-CS-K), and carbapenem-resistant Klebsiella (CR-K). METHODS: We developed a prognostic model to predict hospital mortality in patients with infection on admission to the intensive care units (ICUs), and assessed its calibration in the subgroups of interest: patients with infections due to MS-K, ESCR-CS-K, CR-K. We assessed the calibration of the model also in ESCR-CS-K treated empirically with carbapenems and with piperacillin-tazobactam. RESULTS: A total of 13,292 adults with an ongoing infection were admitted to 137 Italian ICUs in 2012-2013. Of 801 Klebsiella spp. infected patients, 451 had MS-K, 116 ESCR-CS-K, and 234 CR-K. The prognostic model calibrated well for the MS-K and ESCR-CS-K subgroups. In the CR-K subgroup there were more deaths than predicted (standardized mortality ratio 1.20; 95% CI 1.08-1.31), indicating a negative prognostic role of the infection, mainly in the medium and high risk-of-death patients. When infection was caused by ESCR-CS-K, treatment with piperacillin-tazobactam increased adjusted mortality among the most severe patients (similarly to CR-K), while treatment with carbapenems did not (similarly to MS-K). CONCLUSIONS: In low risk-of-death patients admitted to the ICU with a Klebsiella spp. infection, the appropriateness of empirical antibiotic therapy seemed uninfluential to eventual mortality, while it appeared to be crucial in high-risk ones. The use of piperacillin-tazobactam may be inappropriate in severe patients with ESCR-CS-K infection. CR-K is associated to a significant 20% increase of adjusted mortality, only for patients at higher risk of death.

Cirrhotic patients admitted to the ICU for medical reasons: Analysis of 5506 patients admitted to 286 ICUs in 8years.

PURPOSE: To describe characteristics and prognostic factors of cirrhotic patients admitted to a representative sample of Italian intensive care units (ICUs). MATERIALS AND METHODS: All patients admitted to 286 ICUs for medical reasons between 2002 and 2010 (excluding 2007) were considered. A logistic regression model was developed on cirrhotics to predict hospital mortality. The prediction was applied to different subgroups defined by both the level of unit expertise with cirrhotics and the overall unit performance, and compared to the actual mortality. RESULTS: 5506 cirrhotic patients (32.1% admitted to the ICU for non-cirrhotic-related reasons) were compared to 130,477 controls. Hospital mortality was higher in cirrhotics (57.2% vs. 35.0%, p<0.001). ICU volume of cirrhotic patients did not influence mortality, while the overall performance of the unit did. The standardized mortality ratio for overall lower-performing units was 1.09 (95%CI: 1.05-1.14), for the average-performing units it was 1.01 (95%CI: 0.98-1.04), for the higher-performing units it was 0.92 (95%CI: 0.89-0.96). CONCLUSIONS: The outcome of critically ill cirrhotic patients is quite poor, but not to limit their admission to the ICU. When cirrhosis accompanies other acute conditions, the general level of intensive care medicine is more important than the specific liver-oriented expertise in treating these patients.

[The register GiViTI about the use of the drug Xigris® in the Italian intensive care units]. / Il registro GiViTI sull'utilizzo del farmaco Xigris® nelle Terapie Intensive italiane.

INTRODUCTION: The Register was aimed at monitoring the use, possible side effects, and clinical effectiveness of Xigris® - drotrecogin alfa (activated) - for the treatment of severe sepsis in Italian intensive care units (ICUs). METHODS: Data collection was performed using an online web form or a specific electronic module of the software Margherita, available only for the ICUs adhering to the GiViTI. Drug purchase information available for each center was used to identify and stimulate collaboration of non-compliant centers. Several countermeasures were taken to have the largest participation. RESULTS: We analyzed data from 1001 patients treated in 161 ICUs between July 2003 and September 2007, corresponding to 70% of all the patients who received the drug in that period. The off-label use of the drug was frequent: 15.6% of cases before and 27.3% after the label change with the introduction of timing restrictions. Treatment was temporarily interrupted in 10%, and definitely stopped in 25% of cases, after the occurrence of adverse events, the most frequent being bleeding. Severe bleeding occurred in 3.8% of patients. Multivariable analysis, which allowed an adjusted comparison with a control group, showed that treatment increased mortality among elective-surgery patients (OR 2.79, 95%CI 1.31-5.97). DISCUSSION: The results of this study and other evidences led the European Medicines Agency (EMA) to require a confirmatory trial in 2007. In October 2011 Ely-Lilly, the producer of the drug, announced the worldwide withdrawal from the market of Xigris®, on the basis of the negative results of the confirmatory trial. CONCLUSIONS: The availability of purchase information is essential to carry out post-marketing drug surveillance studies, since it allows to identify and contact non-compliant centers. Actually, a representative sample of treated patients provides reliable information on the use, efficacy, and safety of the drug in daily clinical practice that could positively influence healthcare policies.

Is influenza A(H1N1) pneumonia more severe than other community-acquired pneumonias? Results of the GiViTI survey of 155 Italian ICUs.

PURPOSE: Uncertainty about the severity of the A(H1N1) pandemia persists. Information about disease severity can be obtained by investigating intensive care unit (ICU) admissions, especially when historical comparisons can be made with cases of community-acquired pneumonia (CAP). METHODS: This prospective observational study was conducted in 155 ICUs contributing to the GiViTI national database. To assess the impact on ICU workload, the occupancy rate during the epidemic phase was compared with influenza periods in previous years. A logistic regression model was developed to assess the prognostic importance of A(H1N1) influenza. RESULTS: The characteristics of the 319 A(H1N1) cases were similar to those reported in other studies, confirming the young age of patients (mean 43 years) and the higher prevalence among pregnant women and obese people. At the epidemic's peak (October-December 2009) the occupancy rate did not significantly differ from the same period of the previous year, and was significantly lower than the 2009 seasonal influenza outbreak (January-March 2009). Compared with CAP of other origin (3,678 patients), A(H1N1) pneumonia was associated with a lower risk of death. However, after adjusting for confounding this was no longer the case (OR 0.88; 95% CI 0.59-1.31; p = 0.52). CONCLUSION: This study confirmed the specific features of critically ill A(H1N1) patients (i.e., young age, pregnancy, obesity). The pandemic did not increase ICU workload compared with other periods. A(H1N1) pneumonia did not have a higher risk of death than CAP of different origin among patients admitted to the ICU.