Quantifying the associations between antibiotic exposure and resistance - a step towards personalised antibiograms.

Empirical antibiotic treatment is selected to target causative bacteria with antibiotics to which they are not resistant. We analysed the increase in bacterial resistance among individual patients associated with antibiotic exposure in the month prior to infection onset, compared to unexposed patients. From a series of prospective cohort studies in the period 2002-2011 at Beilinson Hospital, Israel, 4232 consecutive patients suspected of infection were included. We analysed resistance to antibiotics in bacterial isolates from patients with clinically significant and microbiologically documented infections, starting antibiotics after obtaining cultures (n = 775). In Gram-negative bacteria, significantly higher rates of resistance was associated with exposure to antibiotics, while no significant associations were found for Gram-positive bacteria. Significant odds ratios (ORs) for increased resistance to classes of antibiotics ranged from 2.1 to 3.3 in Gram-negative bacteria from patients exposed to any antibiotic(s), with quinolones having the highest OR, followed by aminoglycosides, penicillins with ß-lactamase inhibitor and cephalosporins. The majority of significant associations also had significant ORs after exposure to another class of antibiotics, indicating a substantial effect of cross-resistance. In conclusion, increased resistance was observed following exposure to antibiotics, both from the same class and from other classes. The results indicate a reason to adjust the expected coverage of empirical antibiotic treatments for patients recently exposed to antibiotics, with some antibiotics being more affected than others.

Evaluation of a method for converting venous values of acid-base and oxygenation status to arterial values.

OBJECTIVE: This paper evaluates a method in which arterial values of pH, carbon dioxide tension (Pco(2)) and oxygen tension (Po(2)) calculated from venous values and pulse oximetry are compared with simultaneously measured arterial values. METHODS: 103 adult patients from three departments (pulmonary medicine, thoracic intensive care and multidisciplinary intensive care) were studied. The patients belonged to three groups: (1) 31 haemodynamically stable patients with a diagnosis of chronic obstructive lung disease (COLD); (2) 49 haemodynamically stable patients without COLD; and (3) 23 haemodynamically unstable patients without COLD. Arterial and venous (peripheral and, where possible, central and mixed) blood samples were taken simultaneously and anaerobically. Peripheral arterial oxygen saturation was measured with a pulse oximeter. The principle of the method is to simulate the transport of venous blood back through the tissues using the respiratory quotient (adding oxygen and removing carbon dioxide) until simulated arterial oxygenation matches that measured by pulse oximetry. RESULTS: Calculated values of arterial pH and Pco(2) had very small bias and standard deviations regardless of the venous sampling site. In all cases these errors were within those considered acceptable for the performance of laboratory equipment, and well within the limits of error acceptable in clinical practice. In addition, the standard deviation (SD) of calculated values of pH and Pco(2) was similar to the variability between consecutive arterial samples. For peripheral oxygen saturation values < or =96%, the method can calculate Po(2) with an SD of 0.93, which may be useful in clinical practice. Calculations made from peripheral venous blood were significantly more accurate than those from central venous blood. CONCLUSION: Arterial pH and Pco(2) can be calculated precisely from peripheral venous blood in a broad patient population. The method has potential for use as a screening tool in emergency medical departments and in medical and surgical wards to assess a patient's acid-base and oxygenation status prior to sampling arterial blood or to help in the decision to refer the patient to the ICU. In departments where arterial blood gas values are used to monitor patients (eg, pulmonary medicine), the method might reduce the number of arterial samples taken by replacing them with peripheral venous blood samples, thus reducing the need for painful arterial punctures.

Prediction of hemodynamic changes towards PEEP titrations at different volemic levels using a minimal cardiovascular model.

A cardiovascular system model and parameter identification method have previously been validated for porcine experiments of induced pulmonary embolism and positive end-expiratory pressure (PEEP) titrations, accurately tracking all the main hemodynamic trends. In this research, the model and parameter identification process are further validated by predicting the effect of intervention. An overall population-specific rule linking specific model parameters to increases in PEEP is formulated to predict the hemodynamic effects on arterial pressure, pulmonary artery pressure and stroke volume. Hemodynamic changes are predicted for an increase from 0 to 10 cm H(2)O with median absolute percentage errors less than 7% (systolic pressures) and 13% (stroke volume). For an increase from 10 to 20 cm H(2)O median absolute percentage errors are less than 11% (systolic pressures) and 17% (stroke volume). These results validate the general applicability of such a rule, which is not pig-specific, but holds over for all analyzed pigs. This rule enables physiological simulation and prediction of patient response. Overall, the prediction accuracy achieved represents a further clinical validation of these models, methods and overall approach to cardiovascular diagnosis and therapy guidance.

Model-based identification of PEEP titrations during different volemic levels.

A cardiovascular system (CVS) model has previously been validated in simulated cardiac and circulatory disease states. It has also been shown to accurately capture all main hemodynamic trends in a porcine model of pulmonary embolism. In this research, a slightly extended CVS model and parameter identification process are presented and validated in a porcine experiment of positive end-expiratory pressure (PEEP) titrations at different volemic levels. The model is extended to more physiologically represent the separation of venous and arterial circulation. Errors for the identified model are within 5% when re-simulated and compared to clinical data. All identified parameter trends match clinically expected changes. This work represents another clinical validation of the underlying fundamental CVS model, and the methods and approach of using them for cardiovascular diagnosis in critical care.

Relative tachycardia in patients with sepsis: an independent risk factor for mortality.

BACKGROUND: Excess activation of the sympathetic nervous system may be a risk factor for mortality in patients with the systemic inflammatory response syndrome (SIRS) or sepsis. AIM: To examine whether excessive tachycardia, relative to the degree of fever is an independent risk factor for death in patients with SIRS. DESIGN: Prospective observational study. SETTING: Departments of medicine in three university hospitals in Israel, Germany and Italy. METHODS: We collected data for 3382 patients with SIRS, whether community- or hospital-acquired, 91% with sepsis, as part of an ongoing trial. RESULTS: Overall 30-day mortality was 12% (408/3382). The pulse/temperature ratio was significantly higher in patients who died than in survivors: mean +/- SD 2.55 +/- 0.57 vs. 2.40 +/- 0.48 bpm/ degrees C (p < 0.0001). Excessive tachycardia was significantly associated with a mortality in a logistic model accounting for other strong predictors of mortality (OR 1.54, 95%CI 1.10-2.17). Patients with septic shock were the only group for whom this association did not hold. DISCUSSION: Our data are compatible with the hypothesis that some patients with sepsis experience an excess activation of the sympathetic nervous system, leading to a fatal outcome.

An automated method for measuring static pressure-volume curves of the respiratory system and its application in healthy lungs and after lung damage by oleic acid infusion.

Elastic pressure/volume (PV) curves of the respiratory system have attracted increasing interest, because they may be helpful to optimize ventilator settings in patients undergoing mechanical ventilation. Clinically applicable methods need to be fast, use routinely available equipment, draw the inspiratory and expiratory PV curve limbs, separate the resistive and viscoelastic properties of the respiratory system from the elastic properties, and provide reproducible measurements. This paper presents a computer-controlled method for rapid measurements of static PV curves using a long inflation-deflation with pauses, and its evaluation in six pigs before and after lung damage caused by oleic acid. The method is fast, i.e. 20.5 +/- 1.9 s (mean +/- SD) in healthy lungs and 17.7 +/- 4.1 s in diseased lungs, this including inspiratory and expiratory pauses of 1.1 s duration. In addition the only equipment used was a clinical ventilator and a PC. For healthy and damaged lungs expiratory PV curve limbs were very reproducible and were at higher volume than the inspiratory limbs, indicating hysteresis. For damaged lungs inspiratory PV limbs were reproducible. For healthy lungs the inspiratory limbs were reproducible but only after the first inflation-deflation. It is possible that during the first inflation alveoli are recruited which are not derecruited on deflation, shifting the inspiratory limb of the PV curve. The paused long inflation-deflation technique provides a quick, automated measurement of static PV curves on both inspiratory and expiratory limbs using routinely available equipment in the intensive care unit.

Reproduction of MIGET retention and excretion data using a simple mathematical model of gas exchange in lung damage caused by oleic acid infusion.

The multiple inert-gas elimination technique (MIGET) is a complex mathematical model and experimental technique for understanding pulmonary gas exchange. Simpler mathematical models have been proposed that have a limited view compared with MIGET but may be applicable for use in clinical practice. This study examined the use of a simple model of gas exchange to describe MIGET retention and excretion data in seven pigs before and following lung damage caused by oleic acid infusion and subsequently at different levels of positive end-expiratory pressure. The simple model was found to give, on average, a good description of MIGET data, as evaluated by a chi(2) test on the weighted residual sum of squares resulting from the model fit (P > 0.2). Values of the simple model's parameters (dead-space volume, shunt, and the fraction of alveolar ventilation going to compartment 2) compared well with the similar MIGET parameters (dead-space volume, shunt, log of the standard deviation of the perfusion, log of the standard deveation of the ventilation), giving values of bias and standard deviation on the differences between dead-space volume and shunt of 0.002 +/- 0.002 liter and 7.3 +/- 2.1% (% of shunt value), respectively. Values of the fraction of alveolar ventilation going to compartment 2 correlated well with log of the standard deviation of the perfusion (r(2) = 0.86) and log of the standard deviation of the ventilation (r(2) = 0.92). These results indicate that this simple model provides a good description of lung pathology following oleic acid infusion. It remains to be seen whether physiologically valid values of the simple model parameters can be obtained from clinical experiments varying inspired oxygen fraction. If so, this may indicate a role for simple models in the clinical interpretation of gas exchange.

Model predictive glycaemic regulation in critical illness using insulin and nutrition input: a pilot study.

Stress-induced hyperglycaemia is prevalent in intensive care, impairing the immune response. Nutritional support regimes with high glucose content further exacerbate the problem. Tight glucose control has been shown to reduce mortality by up to 43% if levels are kept below 6.1 mmol/L. This research develops a control algorithm with insulin and nutritional inputs for targeted glucose control in the critically ill. Ethics approval for this research was granted by the Canterbury Ethics Committee. Proof-of-concept clinical pilot trials were conducted on intubated, insulin-dependent Christchurch ICU patients (n=7) on constant nutritional support. A target 10-15% reduction in glucose level per hour for a desired glucose level of 4-6 mmol/L was set. 43% and 91% of glucose targets were achieved within +/-5 and +/-20%, respectively. The mean error was 8.9% (0.5 mmol/L), with an absolute range [0, 2.9] mmol/L. End glucose levels were 40% lower compared to initial values. All large target errors are attributable to sudden changes in patient physiology at low glucose values, rather than systemic deficiencies. Target errors are consistent with and explainable by published sensor error distributions. The results show that intensive model-based glucose management with nutrition control reduced absolute glucose levels progressively while reducing the severity of glycaemic fluctuation even with significant inter-patient variability and time-varying physiological condition. Trials spanning longer periods of time are in development to verify the short-term pilot studies performed and to test the adaptability of the controller. Clinically, these results indicate potential in clinical use to reduce ICU mortality as well as reduce risk of severe complications.

A method for calculation of arterial acid-base and blood gas status from measurements in the peripheral venous blood.

In non-emergency medical departments such as internal medicine sampling of arterial blood and analysis for acid-base status is not routinely performed. Peripheral venous blood is routinely taken but interpretation of its acid-base status is difficult. This paper presents a method for calculation of arterial acid-base and blood gas status from measurements in peripheral venous blood combined with a pulse oximeter measurement of arterial saturation. The use of the method has been illustrated using the data of three patients with different acid-base, haemodynamic, and metabolic conditions. The sensitivity of the method has been tested for measurement errors including venous blood acid-base and blood gas status and pulse oximetry; errors due to physiological assumptions including the values of RQ and strong acid production at the tissues; and errors due to air bubbles in the blood. Errors due to these effects are relatively insignificant except for errors in calculated arterial PO(2), particularly when SpO(2) is greater than 97%; and errors when the change in base excess across the sampling site due to strong acid production is greater that 1.3 mmol/l.

Mathematical models of oxygen and carbon dioxide storage and transport: the acid-base chemistry of blood.

This article describes a mathematical model of the acid-base chemistry of blood. The model is formulated from first principles by considering the "components" of blood and the reaction equations in the plasma and erythrocyte fractions. Equations are formulated to describe the total concentration of blood components, the physicochemical properties, and the equilibrium position of reactions. The model includes 28 equations and 12 parameters. All equations can be solved from six variables included in the model. The model uses simple mathematics, without introducing intermediate concepts or linear coefficients necessary for algebraic solution. Model equations are solved simultaneously using numerical methods. Model parameters are estimated and the model verified for plasma, fully oxygenated blood, and deoxygenated blood. Published data are used to estimate model parameters and normal conditions and to verify model simulations. The model reproduces experimental results, including addition or removal of CO2, or strong acid to plasma; CO2, strong acid or haemoglobin to blood; and the effects of deoxygenating blood. The model can also be used as the basis for models of whole body CO2 transport as illustrated in the accompanying article. As such, it is possible to simulate the effects on blood of physiological changes in ventilation or metabolism.

Mathematical models of oxygen and carbon dioxide storage and transport: interstitial fluid and tissue stores and whole-body transport.

This article describes a mathematical model of whole-body O2 and CO2 transport. The model includes representation of the acid-base chemistry of the blood, interstitial fluid, and tissues, plus transport of O2 and CO2 between compartments representing tissues, interstitial fluid, arterial and venous blood, and lungs. The model includes equations for calculation of all concentrations in the compartments, including equations describing the physicochemical properties and reaction equations of interstitial fluid and tissues. In addition, the model includes equations that describe the flow of substrate between the compartments and differential equations allowing calculation of the changes in state variables caused by the flow of substrates between the compartments. This model is designed to calculate the effects of metabolic and respiratory perturbations, such as variation in breathing pattern or production of strong acid at the tissues. The model reproduces the results of published experiments when used to simulate (1) normal conditions in the lungs, arterial and venous blood, interstitial fluid, and tissues during normal ventilation; (2) the characteristic two-exponential response to changes in minute ventilation; and (3) the relationship between arterial blood values of PCO2 and HCO3,p during inspiration of different fractions of CO2.

Quantitative analysis of individual motor unit potentials: a proposition for standardized terminology and criteria for measurement.

The physiology of the motor unit potential (MUP) is reviewed. The aim is to identify the electrophysiological events in the motor unit that generate the individual parts of the MUP. This is based on insight gained from new experimental techniques, such as single-fiber electromyography (EMG), scanning EMG, and simulation studies of the MUP. A terminology for the different parts of the MUP is also suggested, and nine parameters used to describe different features of the MUP are delineated: duration, spike duration, amplitude, area, spike area, phases, turns, satellites, and variability. Technical aspects, such as electrode type, filtering, and sampling rate of the computers, are discussed as well. In Appendix A, different manual and computer-aided methods for quantitative MUP analysis are described. Despite minor systematic differences between the methods, MUP durations measured by different methods correlate highly with each other (Appendix B). The manual and computer-aided methods have comparable variability between repeated measurements.

A method for diagnosing multiple diseases in MUNIN.

A new method for diagnosing multiple diseases in large medical decision support systems based on causal probabilistic networks is proposed. The method is based on characteristics of the diagnostic process that we believe to be present in many diagnostic tasks, both inside and outside medicine. The diagnosis must often be made under uncertainty, choosing between diagnoses that each have small prior probabilities, but not so small that the possibility of two or more simultaneous diseases can be ignored. Often a symptom can be caused by several diseases and the presence of several diseases tend to aggravate the symptoms. For diagnostic problems that share these characteristic, we have proposed a method that operates in a number of phases: in the first phase only single diseases are considered and this helps to focus the attention on a smaller number of plausible diseases. In the second phase, pairs of diseases are considered, which make it possible to narrow down the field of plausible diagnoses further. In the following phases, larger subsets of diseases are considered. The method was applied to the diagnosis of neuromuscular disorders, using previous experience with the so-called MUNIN system as a starting point. The results showed that the method gave large reductions in computation time without compromising the computational accuracy in any substantial way. It is concluded that the method enables practical inference in large medical expert systems based on causal probabilistic networks.

Dynamic updating in DIAS-NIDDM and DIAS causal probabilistic networks.

Diabetes advisory system (DIAS) is a decision support system, which has been developed to provide advice on the amount of insulin injected by subjects with insulin-dependent diabetes mellitus (IDDM). DIAS employs a temporal causal probabilistic network (CPN) to implement a stochastic model of carbohydrate metabolism. The CPN network has recently been extended to provide also advice to subjects with noninsulin-dependent diabetes mellitus (NIDDM). However, due to increased complexity and size of the extended CPN the calculations became unfeasible. The CPN network was, therefore, simplified and a novel approach employed to generate conditional probability tables. The principles of dynamic CPN's were adopted and, in combination with the method of conditioning, learning, and forecasting, were implemented in a time- and memory-efficient way. An evaluation using experimental data was carried out to compare the original and revised DIAS implementations employing data collected by patients with IDDM, and to assess the a posteriori identifiability of model parameters in patients with NIDDM.

Specification of models in large expert systems based on causal probabilistic networks.

Problems involved in the specification of large expert systems are discussed. In the specification of causal probabilistic networks conditional probability tables for all nodes have to be provided. These conditional probability tables can often be described by models that specify the nature of interaction between nodes. Various types of models are described and a program that handles such models is presented. Large causal probabilistic networks often contain several copies of identical tables or structures. A header facility that provides common definitions of such repeated elements is proposed. This facility makes specifications much shorter and easier to construct and maintain.

Using probabilistic and decision-theoretic methods in treatment and prognosis modeling.

Causal probabilistic networks, also called Bayesian networks, allow both qualitative knowledge about the structure of a problem and quantitative knowledge, derived from case databases, expert opinion and literature to be exploited in the construction of decision support systems for diagnosis, treatment and prognosis. This mixing of qualitative and quantitative knowledge will be illustrated, using the selection of antibiotics for a subset of patients with severe infections. The subset consists of patients where bacteria or fungi have been found in the blood. A simple pathophysiological model of infection is used to calculate a prognosis, dependent on the choice of antibiotics. A decision-theoretic approach is used to balance the therapeutic benefit of antibiotic treatment against the cost of antibiotics in the form of direct monetary cost, side effects and ecological cost. A retrospective trial on patients with bacteria or fungi in the blood stemming from the urinary tract indicates that with this approach, it may be possible to suggest balanced choices of antibiotics that not only achieve greater therapeutic benefit, but also reduce the cost of therapy.

EMG-torque dynamics at different contraction levels in human ankle muscles.

The electrically elicited muscle twitch has been used to identify mechanical muscle properties in relaxed muscles. We attempted to characterize the mechanical muscle properties in an active muscle. Each subject was seated and his/her left foot was strapped to a platform. The ankle torque and electromyogram (EMG) of the ankle extensors and flexors were measured while the subject was asked to match the ankle torque to a pseudo randomized rectangular tracking signal. A system identification technique was used to determine the impulse response from EMG to torque at various contraction levels. The amplitude of the impulse response decreased markedly with the contraction level when the amplitude of the tracking signal was constant, whereas the amplitude of the impulse response increased with the amplitude of the tracking signal. An explanation for these findings could be seen in the results from the properties of individual motor units. Our results suggest that the rate modulation that occurs during rapid changes in the force in an already isometric contracted muscle is very efficient in generating force in the newly recruited motor units, but inefficient in motor units approaching tetanus.

Model-based biosignal interpretation.

Two relatively new approaches to model-based biosignal interpretation, qualitative simulation and modelling by causal probabilistic networks, are compared to modelling by differential equations. A major problem in applying a model to an individual patient is the estimation of the parameters. The available observations are unlikely to allow a proper estimation of the parameters, and even if they do, the task appears to have exponential computational complexity if the model is non-linear. Causal probabilistic networks have both differential equation models and qualitative simulation as special cases, and they can provide both Bayesian and maximum-likelihood parameter estimates, in most cases in much less than exponential time. In addition, they can calculate the probabilities required for a decision-theoretical approach to medical decision support. The practical applicability of causal probabilistic networks to real medical problems is illustrated by a model of glucose metabolism which is used to adjust insulin therapy in type I diabetic patients.

Causal probabilistic network and power spectral estimation used in sleep stage classification.

A new method for sleep-stage classification using a causal probabilistic network as automatic classifier has been implemented and validated. The system uses features from the primary sleep signals from the brain (EEG) and the eyes (AOG) as input. From the EEG, features are derived containing spectral information which is used to classify power in the classical spectral bands, sleep spindles and K-complexes. From AOG, information on rapid eye movements is derived. Features are extracted every 2 seconds. The CPN-based sleep classifier was implemented using the HUGIN system, an application tool to handle causal probabilistic networks. The results obtained using different training approaches show agreements ranging from 68.7 to 70.7% between the system and the two experts when a pooled agreement is computed over the six subjects. As a comparison, the interrater agreement between the two experts was found to be 71.4%, measured also over the six subjects.

The EMG diagnosis--an interpretation based on partial information.

There is a large difference between the prevalence of a given disease in the general population and in the population seen in the EMG lab. It can be argued that both prevalences are the correct choice as prior probabilities for the diseases. This paradox is resolved by recognizing that the EMG diagnosis is only based on the information provided by the EMG examination and thus only represents a partial view of the patient. We propose a solution summarizing the set of findings, signs and symptoms, lab results etc., that led to the referral of the patient for an EMG examination. This information is described by stochastic variables called FIDL factors (Found In Doctor's Lab). The approach is tested on the EMG expert system MUNIN with 30 previously evaluated cases. The results show that this solution improves the specificity of the diagnosis, without affecting the sensitivity.