Anna Maria Zanaboni

Stratifying Risk Factors for Multidrug-Resistant Pathogens in Hospitalized Patients Coming From the Community With Pneumonia

BACKGROUND  Not all risk factors for acquiring multidrug-resistant (MDR) organisms are equivalent in predicting pneumonia caused by resistant pathogens in the community. We evaluated risk factors for acquiring MDR bacteria in patients coming from the community who were hospitalized with pneumonia. Our evaluation was based on actual infection with a resistant pathogen and clinical outcome during hospitalization. METHODS  An observational, prospective study was conducted on consecutive patients coming from the community who were hospitalized with pneumonia. Data on admission and during hospitalization were collected. Logistic regression models were used to evaluate risk factors for acquiring MDR bacteria independently associated with the actual presence of a resistant pathogen and in-hospital mortality. RESULTS  Among the 935 patients enrolled in the study, 473 (51%) had at least 1 risk factor for acquiring MDR bacteria on admission. Of all risk factors, hospitalization in the preceding 90 days (odds ratio [OR], 4.87 95% confidence interval {CI}, 1.90-12.4]; P = .001) and residency in a nursing home (OR, 3.55 [95% CI, 1.12-11.24]; P = .031) were independent predictors for an actual infection with a resistant pathogen. A score able to predict pneumonia caused by a resistant pathogen was computed, including comorbidities and risk factors for MDR. Hospitalization in the preceding 90 days and residency in a nursing home were also independent predictors for in-hospital mortality. CONCLUSIONS  Risk factors for acquiring MDR bacteria should be weighted differently, and a probabilistic approach to identifying resistant pathogens among patients coming from the community with pneumonia should be embraced.

Multidrug-resistant pathogens in hospitalised patients coming from the community with pneumonia: a European perspective

Background Probabilistic scores have been recently suggested to identify pneumonia caused by multidrug-resistant (MDR) bacteria. The aim of the study was to validate both Aliberti and Shorr scores in predicting MDR pneumonia, comparing them with healthcare associated pneumonia (HCAP) classification. Methods Two independent European cohorts of consecutive patients hospitalised with pneumonia were prospectively evaluated in Barcelona, Spain (BC) and Edinburgh, UK (EC). Data on admission and during hospitalisation were collected. The predictive value of the three scores was explored for correctly indicating the presence of MDR pneumonia via a receiver-operating characteristic (ROC) curve. Results A total of 1591 patients in the BC and 1883 patients in the EC were enrolled. The prevalence of patients with MDR pathogen among those with isolated bacteria was 7.6% in the BC and 3.3% in the EC. The most common MDR pathogen found in both cohorts was MRSA, followed by MDR P aeruginosa. A significantly higher prevalence of MDR bacteria was found among patients in the intensive care unit (ICU). The two probabilistic scores, and particularly the Aliberti one, showed an area under the ROC curve higher than the HCAP classification in predicting MDR pneumonia, especially in the ICU. Conclusions Risk scores able to identify MDR pneumonia could help in developing strategies for antimicrobial stewardship.

Learning fuzzy rules with tabu search-an application to control

We present an approach for the automatic definition of the fuzzy rules for a fuzzy controller based on the use of the tabu search (TS) scheme. We show also how the application of the TS process to the learning of a fuzzy rule base can be improved using heuristic symbolic meta rules. The paper is divided in two parts. The first part presents an introduction to TS and different learning schemes which can be used to apply it for the determination of the fuzzy control rules. The second part illustrates the application of the proposed techniques to a specific control problem-the parking of a truck and trailer. In particular, Section V illustrates the definition of a rule base for a static fuzzy controller, while Section VI presents the construction of an adaptive parking controller.

Waist-to-Height Ratio Is a Highly Sensitive Index for the Metabolic Syndrome in a Mediterranean Population

BACKGROUND The waist-to-height ratio (WHtR) is a potentially more reliable anthropometric index, particularly for populations of lower height. Performance of the WHtR versus body mass index (BMI) and enlarged waist circumference (WC) in the assessment of the metabolic syndrome was tested in nonobese males and females in a high-risk Italian population. METHODS WHtR, BMI, and WC were determined in 552 males and 552 females, together with the evaluation of associated metabolic syndrome variables (hypertension, hyperglycemia, hypertriglyceridemia, and low high-density lipoprotein cholesterol [HDL-C]). RESULTS WHtR > or = 0.5, the most frequently suggested threshold value, when added to any two nonanthropometric variables, gave a sensitivity for the identification of a metabolic syndrome of, respectively, 92.0% for males and 87.4% for females. Sensitivities for elevated WC (American Heart Association [AHA] criteria) and BMI > or = 25 proved lower. Areas under the receiver operating characteristic (ROC) curves for the different anthropometric indices confirmed that a WHtR > or = 0.5 provides a satisfactory balance between sensitivity and specificity. CONCLUSIONS WHtR > or = 0.5 may be the most effective anthropometric index for screening high-risk patients in the diagnosis of metabolic syndrome, with the advantage of the opportunity of direct comparisons with other populations.

Learning fuzzy decision trees

We present a recurrent neural network which learns to suggest the next move during the descent along the branches of a decision tree. More precisely, given a decision instance represented by a node in the decision tree, the network provides the degree of membership of each possible move to the fuzzy set z.Lt;good movez.Gt;. These fuzzy values constitute the core of the probability of selecting the move out of the set of the children of the current node.This results in a natural way for driving the sharp discrete-state process running along the decision tree by means of incremental methods on the continuous-valued parameters of the neural network. The bulk of the learning problem consists in stating useful links between the local decisions about the next move and the global decisions about the suitability of the final solution. The peculiarity of the learning task is that the network has to deal explicitly with the twofold charge of lighting up the best solution and generating the move sequence that leads to that solution. We tested various options for the learning procedure on the problem of disambiguating natural language sentences.

Low CURB-65 is of limited value in deciding discharge of patients with community-acquired pneumonia

BACKGROUND The relationship between clinical judgment and indications of the CURB-65 score in deciding the site-of-care for patients with community-acquired pneumonia (CAP) has not been fully investigated. The aim of this study was to evaluate reasons for hospitalization of CAP patients with CURB-65 score of 0 and 1. METHODS An observational, retrospective study of consecutive CAP patients was performed at the Fondazione Cà Granda, Milan, Italy, between January 2005 and December 2006. The medical records of hospitalized patients with CAP having a CURB-65 score of 0 and 1 were identified and reviewed to determine whether there existed a clinical basis to justify hospitalization. RESULTS Among the 580 patients included in the study, 218 were classified with a CURB-65 score of 0 or 1. Among those, 127 were hospitalized, and reasons that justified hospitalization were found in 104 (83%) patients. Main reasons for hospitalization included the presence of hypoxemia on admission (35%), failure of outpatient therapy (14%) and the presence of cardiovascular events on admission (9.7%). Used as the sole indicator for inappropriate hospitalization, the CURB-65 score had a poor positive predictive value of 52%. CONCLUSIONS Although the CURB-65 has been proposed as a tool to guide the site of care decision by international guidelines, this score is not ideal by itself, and should not be regarded as providing decision support information if a score of 0 and 1 is present. In CAP patients with CURB-65 scores of 0 or 1, further evaluations should be performed and completed by clinical judgment.

Criteria for clinical stability in hospitalised patients with community-acquired pneumonia

The American Thoracic Society (ATS) and Infectious Diseases Society of America (IDSA) suggested two sets of criteria in 2001 and 2007 to define clinical stability in community-acquired pneumonia (CAP). The present study aimed to evaluate the level of agreement between these two sets of criteria and how well they can predict clinical outcomes. A retrospective cohort study was carried out of 487 consecutive patients hospitalised with CAP. Level of agreement was tested using a survival curve analysis, while prediction of outcomes at 30-day follow-up was evaluated through receiver operating characteristic (ROC) analysis. A discrepancy between ATS 2001 and ATS/IDSA 2007 criteria in identifying clinical stability was detected in 62% of the patients. The median (interquartile range) time to clinical stability was 2 (1–4) days based on ATS 2001 and 3 (2–5) days based on ATS/IDSA 2007 criteria (p = 0.012). The daily distribution of patients who reached clinical stability evaluated with both sets was different (p = 0.002). The ROC analysis showed an area under the curve of 0.705 for the ATS 2001 criteria and 0.714 for ATS/IDSA 2007 criteria (p = 0.645). ATS 2001 and ATS/IDSA 2007 criteria for clinical stability in hospitalised patients with CAP are clinically equivalent and both can be used in clinical practice as well as in clinical research. ATS 2001 and ATS/IDSA 2007 criteria for clinical stability in hospitalised patients with CAP are clinically equivalent http://ow.ly/lAeKE

Synthesis of fuzzy controllers through neural networks

Abstract Most of the current fuzzy logic control applications are designed using different heuristics for the controller synthesis, and then implemented using conventional programming languages on general purpose microcontrollers. We are proposing a methodology for the design of fuzzy controllers based on the cell-to-cell mapping approach for the fuzzy control law synthesis, and on neural networks for: (a) the discovery of the set of appropriate fuzzy rules that characterise a control law, and (b) the tuning of parameters that characterise membership functions (namely position and width). We suppose that training data are coming from sampling of the analytical control function. Two examples are shown, and a comparison with the results obtained by a generalized multilayer perceptron is discussed.

Learning rule representations from data

We discuss a procedure which extracts statistical and entropic information from data in order to discover Boolean rules underlying them. We work within a granular computing framework where logical implications between statistics on the observed sample and properties on the whole data population are stressed in terms of both probabilistic and possibilistic measures of the inferred rules. With the main constraint that the class of rules is not known in advance, we split the building of the hypotheses on them in various levels of increasing description complexity, balancing the feasibility of the learning procedure with the understandability and reliability of the formulas that are discovered. We appreciate the entire learning system in terms of truth tables, formula lengths, and computational resources through a set of case studies

Pneumonia in the Community Caused by Multidrug-Resistant Organisms: Keep Working on Probabilistic Scores

TO THE EDITOR—Two main concepts have gained importance during the past decade in both the literature and clinical practice: (1) Multidrug-resistant organisms (MDROs) causing pneumonia in the community represent a real and emerging problem, and (2) the classification of healthcare-associated pneumonia (HCAP) does not allow physicians to properly identify pneumonia caused by MDROs [1]. Risk factors of the HCAP classification should be weighted differently, and previous hospitalization and nursing home residency are major determinants for the acquisition of MDROs. As an alternative to the HCAP classification, some investigators have developed scoring systems to weight each risk factor for MDRO infection individually [2]. We read with interest the article by Shorr et al [3], who tested a score for predicting the presence of pneumonia caused by MDROs in a cohort of patients admitted to a single American hospital. Points were assigned based on the presence of the following risk factors: recent hospitalization, residence in long-term care facility, chronic hemodialysis, and admission to intensive care unit (ICU). The authors found that the clinical risk score performed moderately well at classifying patients regarding their risk for MDRO infection. The approach proposed by Shorr et al is innovative, even though the score has been derived from a retrospective cohort of patients with severe pneumonia. In fact, more than one-half of the patients were directly admitted to the ICU and the main pathogen isolated was methicillinresistant Staphylococcus aureus, followed by Pseudomonas aeruginosa and Streptococcus pneumoniae. In order to evaluate this score in a different population, we used a prospective cohort of 935 patients with pneumonia coming from the community who were admitted to an Italian university tertiary care hospital from April 2008 through April 2010. Study methodology and baseline characteristics of the study population have been published previously [4]. Eight patients from the study population were directly admitted to the ICU. Among the 170 patients who had a bacterium isolated, 33 had an MDRO. The score proposed by Shorr et al [3] was evaluated in comparison with the HCAP definition with regard to both the actual infection with an MDRO and the in-hospital mortality. The receiver operating characteristic (ROC) curves of both scores are depicted in Figure 1. With regard to the actual infection with an MDRO, the area under the ROC curve was 0.704 (95% confidence interval [CI], .592–.815) and 0.709 (95% CI, .604–.813) for the score and HCAP classification, respectively (Figure 1). With regard to the in-hospital mortality, the area under the ROC curve was 0.624 (95% CI, .573–.676) and 0.637 (95% CI, .587–.687) for the score and HCAP classification, respectively (Figure 1). In view of the important weight of ICU admission in the scoring system proposed by Shorr et al [3], this score could be fully appreciated in a selected population of patients with severe pneumonia. We strongly believe that the probabilistic approach should be pursued in evaluating the presence of MDROs causing pneumonia in the community. Different scoring systems should be tested in populations of patients hospitalized for moderate to severe pneumonia in a multicenter, prospective trial [4].