Median Prediction Error Research Articles

Closed-Loop Feedback Computer-Controlled Phenylephrine for Maintenance of Blood Pressure During Spinal Anesthesia for Cesarean Delivery: A Randomized Trial Comparing Automated Boluses Versus Infusion.

We previously described the use of closed-loop feedback computer-controlled infusion of phenylephrine for maintaining blood pressure (BP) during spinal anesthesia for cesarean delivery. In this study, we report a modified system in which phenylephrine is delivered by intermittent boluses rather than infusion. We hypothesized that the use of computer-controlled boluses would result in more precise control of BP compared with infusions. Two hundred fourteen healthy patients having spinal anesthesia for elective cesarean delivery were randomized to have their systolic BP maintained by phenylephrine administered by computer-controlled continuous infusion or computer-controlled intermittent boluses. From induction of anesthesia until the time of uterine incision, a noninvasive BP monitor was set to cycle at 1-minute intervals. In the infusion group, the infusion rate was automatically adjusted after each BP measurement using a previously described algorithm. In the bolus group, the algorithm was modified so that the mass of drug that would have been delivered over 1 minute was instead injected as a rapid intravenous bolus after each BP measurement. The precision of BP control was assessed using performance error calculations and compared between groups, with the primary outcome defined as median absolute performance error, and the latter being a measure of inaccuracy showing an average of the magnitudes of the differences of measured BP values above or below the target values. The precision of BP control was greater, as shown by smaller values for median absolute performance error, in the bolus group (median 4.38 [quartiles 3.22, 6.25] %) versus the infusion group (5.39 [4.12, 7.04] %, P = .008). In the bolus group, phenylephrine consumption was smaller; this was associated with smaller values for median performance error compared with the continuous infusion group (P < .001), which indicates that values for systolic BP, averaged over time, were slightly lower in the bolus group. There were no differences in cardiac output, nausea or vomiting, or neonatal outcome between groups. We confirmed the hypothesis that BP control was more precise when computer-controlled phenylephrine was delivered using intermittent boluses rather than continuous infusion. However, the difference between groups was small and was not associated with any difference in clinical outcomes. In the infusion group, greater doses of phenylephrine were delivered, which was related to the time taken for the noninvasive BP monitor to complete measurements. The use of intermittent boluses may be a useful alternative in the design of closed-loop vasopressor systems.

Desflurane usage during anesthesia with and without N2O using FLOW-i Automatic Gas Control with three different wash-in speeds.

AGC® (Automatic Gas Control) is the FLOW-i's automated low flow tool (Maquet, Solna, Sweden) that target controls the inspired O2 (FIO2) and end-expired desflurane concentration (FAdes) while (by design) exponentially decreasing fresh gas flow (FGF) during wash-in to a maintenance default FGF of 300mL min-1. It also offers a choice of wash-in speeds for the inhaled agents. We examined AGC performance and hypothesized that the use of lower wash-in speeds and N2O both reduce desflurane usage (Vdes). After obtaining IRB approval and patient consent, 78 ASA I-II patients undergoing abdominal surgery were randomly assigned to 1 of 6 groups (n = 13 each), depending on carrier gas (O2/air or O2/N2O) and wash-in speed (AGC speed 2, 4, or 6) of desflurane, resulting in groups air/2, air/4, air/6, N2O/2, N2O/4, and N2O/6. The target for FIO2 was set at 35%, while the FAdes target was selected so that the AGC displayed 1.3 MAC (corrected for the additive affect of N2O if used). AGC was activated upon starting mechanical ventilation. Varvel's criteria were used to describe performance of achieving the targets. Patient demographics, end-expired N2O concentration, MAC, FGF, and Vdes were compared using ANOVA. Data are presented as mean ± standard deviation, except for Varvel's criteria (median ± quartiles). Patient demographics did not differ among the groups. Median performance error was -2-0% for FIO2 and -3-1% for FAdes; median absolute performance error was 1-2% for FIO2 and 0-3% for FAdes. MAC increased faster in N2O groups, but total MAC decreased 0.1-0.25 MAC below that in the O2/air groups after 60min. The effect of wash-in speed on Vdes faded over time. N2O decreased Vdes by 62%. AGC performance for O2 and desflurane targeting is excellent. After 1h, the wash-in speeds tested are unlikely to affect desflurane usage. N2O usage decreases Vdes proportionally with its reduction in FAtdes.

Pharmacokinetics and pharmacodynamics of propofol: changes in patients with frontal brain tumours

BackgroundModels of propofol pharmacokinetics and pharmacodynamics developed in patients without brain pathology are widely used for target-controlled infusion (TCI) during brain tumour excision operations. The goal of this study was to determine if the presence of a frontal brain tumour influences propofol pharmacokinetics and pharmacodynamics and existing PK-PD model performance. MethodsTwenty patients with a frontal brain tumour and 20 control patients received a propofol infusion to achieve an induction–emergence–induction anaesthetic sequence. Propofol plasma concentration was measured every 4 min and at each transition of the conscious state. Bispectral index (BIS) values were continuously recorded. We used non-linear mixed-effects modelling to analyse the effects of the presence of a brain tumour on the pharmacokinetics and pharmacodynamics of propofol. Subsequently we calculated the predictive performance of Marsh, Schnider, and Eleveld models in terms of median prediction error (MdPE) and median absolute prediction error (MdAPE). ResultsPatients with brain tumours showed 40% higher propofol clearance than control patients. Performance of the Schnider model (MdPEpk −20.0%, MdAPEpk 23.4%) and Eleveld volunteer model (MdPEpk −8.58%, MdAPEpk 21.6%) were good. The Marsh model performed less well (MdPEpk −14.3%, MdAPEpk 41.4%), as did the Eleveld patient model (MdPEpk −30.8%, MdAPEpk 32.1%).The first-order rate constant (ke0; 0.108 min−1), the concentration in the effects compartment associated with 50% of the maximum effect (Ce50; 2.77 ml litre−1) and γ (1.49) did not significantly differ between groups. Lower baseline BIS values were found in patients with brain tumours (90 vs 95). ConclusionsBrain tumours might alter the pharmacokinetics of propofol. Caution should be exerted when using propofol TCI in patients with frontal brain tumours due to higher clearance. Trial registry numberNCT01060631.

Phenobarbital in intensive care unit pediatric population: predictive performances of population pharmacokinetic model.

An external evaluation of phenobarbital population pharmacokinetic model described by Marsot et al. was performed in pediatric intensive care unit. Model evaluation is an important issue for dose adjustment. This external evaluation should allow confirming the proposed dosage adaptation and extending these recommendations to the entire intensive care pediatric population. External evaluation of phenobarbital published population pharmacokinetic model of Marsot et al. was realized in a new retrospective dataset of 35 patients hospitalized in a pediatric intensive care unit. The published population pharmacokinetic model was implemented in nonmem 7.3. Predictive performance was assessed by quantifying bias and inaccuracy of model prediction. Normalized prediction distribution errors (NPDE) and visual predictive check (VPC) were also evaluated. A total of 35 infants were studied with a mean age of 33.5 weeks (range: 12 days-16 years) and a mean weight of 12.6 kg (range: 2.7-70.0 kg). The model predicted the observed phenobarbital concentrations with a reasonable bias and inaccuracy. The median prediction error was 3.03% (95% CI: -8.52 to 58.12%), and the median absolute prediction error was 26.20% (95% CI: 13.07-75.59%). No trends in NPDE and VPC were observed. The model previously proposed by Marsot et al. in neonates hospitalized in intensive care unit was externally validated for IV infusion administration. The model-based dosing regimen was extended in all pediatric intensive care unit to optimize treatment. Due to inter- and intravariability in pharmacokinetic model, this dosing regimen should be combined with therapeutic drug monitoring.

Limited sampling strategies for determining the area under the plasma concentration–time curve for isoniazid might be a valuable approach for optimizing treatment in adult patients with tuberculosis

This study aimed to develop clinically feasible models of limited sampling strategy (LSS) for estimation of the area under the concentration–time curve (AUC24h) for isoniazid, that could be applied easily in daily clinical practice for dosage adjustment in adult patients with tuberculosis. Isoniazid plasma concentrations (n = 1665) from 185 adult tuberculous patients were used for the development and validation of LSS models to estimate AUC24h following administration of the standard 5 mg/kg dose of isoniazid. Population pharmacokinetic analysis for appropriate estimation of isoniazid pharmacokinetic parameters was performed in a modelling group (n = 100). The Bayesian estimates of AUC24h (AUCref) obtained for each individual were used as the dependent variable in the regression analysis for the development of various LSS models. The LSS models were validated in a separate cohort (n = 85). Several three and four time point LSS models were built and tested. Model H (AUC24h = −1.88 + 1.05 × C1 + 0.78 × C2 + 9.44 × C5) and Model I (AUC24h = −0.65 + 1.00 × C1 + 1.94 × C2 + 15.45 × C9) had the best performances [adj-R2 = 0.93, median prediction error (MPE) = −0.20, root median squared prediction error (RMSE) = 4.65 for Model H; adj-R2 = 0.96, MPE = −0.05 RMSE = 3.56 for Model I]. The very high R2 values (≥0.94) of these regression equations in the validation cohort confirmed their high reliability. These LSS models could be applied in the context of therapeutic drug monitoring programmes aiming to personalize isoniazid dosing regimens for adult patients with tuberculosis.

Pharmacokinetic modeling of gentamicin in treatment of infective endocarditis: Model development and validation of existing models.

Gentamicin shows large variations in half-life and volume of distribution (Vd) within and between individuals. Thus, monitoring and accurately predicting serum levels are required to optimize effectiveness and minimize toxicity. Currently, two population pharmacokinetic models are applied for predicting gentamicin doses in adults. For endocarditis patients the optimal model is unknown. We aimed at: 1) creating an optimal model for endocarditis patients; and 2) assessing whether the endocarditis and existing models can accurately predict serum levels. We performed a retrospective observational two-cohort study: one cohort to parameterize the endocarditis model by iterative two-stage Bayesian analysis, and a second cohort to validate and compare all three models. The Akaike Information Criterion and the weighted sum of squares of the residuals divided by the degrees of freedom were used to select the endocarditis model. Median Prediction Error (MDPE) and Median Absolute Prediction Error (MDAPE) were used to test all models with the validation dataset. We built the endocarditis model based on data from the modeling cohort (65 patients) with a fixed 0.277 L/h/70kg metabolic clearance, 0.698 (±0.358) renal clearance as fraction of creatinine clearance, and Vd 0.312 (±0.076) L/kg corrected lean body mass. External validation with data from 14 validation cohort patients showed a similar predictive power of the endocarditis model (MDPE -1.77%, MDAPE 4.68%) as compared to the intensive-care (MDPE -1.33%, MDAPE 4.37%) and standard (MDPE -0.90%, MDAPE 4.82%) models. All models acceptably predicted pharmacokinetic parameters for gentamicin in endocarditis patients. However, these patients appear to have an increased Vd, similar to intensive care patients. Vd mainly determines the height of peak serum levels, which in turn correlate with bactericidal activity. In order to maintain simplicity, we advise to use the existing intensive-care model in clinical practice to avoid potential underdosing of gentamicin in endocarditis patients.

Population pharmacokinetics of intravenous acetaminophen in Japanese patients undergoing elective surgery.

Intravenous (i.v.) acetaminophen is administered during surgery for postoperative analgesia. However, little information is available on the pharmacokinetics of i.v. acetaminophen in Japanese patients undergoing surgery under general anesthesia. The study was approved by the Institutional Review Board and registered at UMIN-CTR (UMIN000013418). Patients scheduled to undergo elective surgery under general anesthesia were enrolled after obtaining written informed consent. During surgery, 1g of i.v. acetaminophen was administered over 15, 60, or 120min. Acetaminophen concentrations (15 or 16 samples per case) were measured at time points from 0-480min after the start of administration (liquid chromatography-mass spectrometry/tandem mass spectrometry; limit of quantitation 0.1μg/mL). The predictive performance of three published pharmacokinetic models was evaluated. Population pharmacokinetics were also analyzed using a nonlinear mixed-effect model based on the NONMEM program. Data from 12 patients who underwent endoscopic or lower limb procedures were analyzed (male/female=7/5, median age 55years, weight 63kg). Anesthesia was maintained with remifentanil and propofol or sevoflurane. The pharmacokinetic model of i.v. acetaminophen reported by Würthwein et al. worked well. Using 185 datapoints, the pharmacokinetics of i.v. acetaminophen were described by a two-compartment model with weight as a covariate but not age, sex, or creatinine clearance. The median prediction error and median absolute prediction error of the final model were -1 and 10%, respectively. A population pharmacokinetic model of i.v. acetaminophen in Japanese patients was constructed, with performance within acceptable ranges.

Predictive performance of eleven pharmacokinetic models for propofol infusion in children for long-duration anaesthesia

BackgroundPredictive performance of eleven published propofol pharmacokinetic models was evaluated for long-duration propofol infusion in children. MethodsTwenty-one aged three–11 yr ASA I–II patients were included. Anaesthesia was induced with propofol or sevoflurane, and maintained with propofol, remifentanil, and fentanyl. Propofol was continuously infused at rates of 4–14 mg kg −1 h−1 after an initial bolus of 1.5–2.0 mg kg −1. Venous blood samples were obtained every 30–60 min for five h and then every 60–120 min after five h from the start of propofol administration, and immediately after the end of propofol administration. Model performance was assessed with prediction error (PE) derivatives including divergence PE, median PE (MDPE), and median absolute PE (MDAPE) as time-related PE shift, measures for bias, and inaccuracy, respectively. ResultsWe collected 85 samples over 270 (130) (88–545), mean (SD) (range), min. The Short model for children, and the Schüttler general-purpose model had acceptable performance (–20%≤MDPE ≤ 20%, MDAPE ≤ 30%, –4% h−1≤ divergence PE ≤ 4% h−1). The Short model showed the best performance with the maximum predictive performance metric. Two models developed only using bolus dosing (Shangguan and Saint-Maurice models) and the Paedfusor of the remaining nine models had significant negative divergence PE (≤–6.1% h−1). ConclusionsThe Short model performed well during continuous infusion up to 545 min. This model might be preferable for target-controlled infusion for long-duration anaesthesia in children.

Intraocular lens power calculations in eyes with previous hyperopic laser in situ keratomileusis or photorefractive keratectomy

To evaluate the accuracy of 7 intraocular lens (IOL) calculation formulas in patients with previous hyperopic laser in situ keratomileusis (LASIK) or excimer laser photorefractive keratectomy (PRK). Retrospective case series. Cullen Eye Institute, Baylor College of Medicine, Houston, Texas, and private practice, Mesa, Arizona, USA. The 7 formulas evaluated were the adjusted Atlas 0-3, Masket, Modified Masket, Haigis-L, Shammas-PL, Barrett True-K, and Barrett True-K No-History. The Masket and Modified Masket were calculated using the single-K version of Holladay 1 and Hoffer Q formulas; the adjusted Atlas 0-3 was calculated using the double-K version of Holladay 1 and Hoffer Q. The IOL power predicted by each formula was calculated by targeting the postoperative manifest refraction. The IOL prediction error was obtained by subtracting the predicted IOL power from the implanted IOL power. The mean IOL prediction error, median absolute refractive prediction error, and percentages of eyes within ±0.50 diopter (D) and ±1.00 D of the predicted refraction were calculated. Twenty-one eyes of 21 patients were evaluated. There were no significant differences in the median absolute refractive prediction error or percentages of eyes within ±0.50 D or ±1.00 D of the predicted refraction between formulas or methods. The IOL mean prediction errors were comparable between the Holladay 1 and Hoffer Q calculations for all formulas except for a greater error for the double-K version of the Hoffer Q of the adjusted Atlas 0-3. In eyes that had hyperopic LASIK or PRK, there were no significant differences in the accuracy between the 7 IOL calculation formulas.

Multiple-Reflection Noise Attenuation Using Adaptive Randomized-Order Empirical Mode Decomposition

We propose a novel approach for removing noise from multiple reflections based on an adaptive randomized-order empirical mode decomposition (EMD) framework. We first flatten the primary reflections in common midpoint gather using the automatically picked normal moveout velocities that correspond to the primary reflections and then randomly permutate all the traces. Next, we remove the spatially distributed random spikes that correspond to the multiple reflections using the EMD-based smoothing approach that is implemented in the $f-x$ domain. The trace randomization approach can make the spatially coherent multiple reflections random along the space direction and can decrease the coherency of near-offset multiple reflections. The EMD-based smoothing method is superior to median filter and prediction error filter in that it can help preserve the flattened signals better, without the need of exact flattening, and can preserve the amplitude variation much better. In addition, EMD is a fully adaptive algorithm and the parameterization for EMD-based smoothing can be very convenient.

A Model for Predicting the Interindividual Variability of Drug-Drug Interactions.

Pharmacokinetic drug-drug interactions are frequently characterized and quantified by an AUC ratio (Rauc). The typical value of the AUC ratio in case of cytochrome-mediated interactions may be predicted by several approaches, based on in vitro or in vivo data. Prediction of the interindividual variability of Rauc would help to anticipate more completely the consequences of a drug-drug interaction. We propose and evaluate a simple approach for predicting the standard deviation (sd) of Ln(Rauc), a metric close to the interindividual coefficient of variation of Rauc. First, a model was derived to link sd(Ln Rauc) with the substrate fraction metabolized by each cytochrome and the potency of the interactors, in case of induction or inhibition. Second, the parameters involved in these equations were estimated by a Bayesian hierarchical model, using the data from 56 interaction studies retrieved from the literature. Third, the model was evaluated by several metrics based on the fold prediction error (PE) of sd(Ln Rauc). The median PE was 0.998 (the ideal value is 1) and the interquartile range was 0.96-1.03. The PE was in the acceptable interval (0.5 to 2) in 52 cases out of 56. Fourth, a surface plot of sd(Ln Rauc) as a function of the characteristics of the substrate and the interactor has been built. The minimal value of sd(Ln Rauc) was about 0.08 (obtained for Rauc = 1) while the maximal value, 0.7, was obtained for interactions involving highly metabolized substrates with strong interactors.

Dosimetric Implications of Residual Tracking Errors During Robotic SBRT of Liver Metastases

Although the metric precision of robotic stereotactic body radiation therapy in the presence of breathing motion is widely known, we investigated the dosimetric implications of breathing phase-related residual tracking errors. In 24 patients (28 liver metastases) treated with the CyberKnife, we recorded the residual correlation, prediction, and rotational tracking errors from 90 fractions and binned them into 10 breathing phases. The average breathing phase errors were used to shift and rotate the clinical tumor volume (CTV) and planning target volume (PTV) for each phase to calculate a pseudo 4-dimensional error dose distribution for comparison with the original planned dose distribution. The median systematic directional correlation, prediction, and absolute aggregate rotation errors were 0.3mm (range, 0.1-1.3mm), 0.01mm (range, 0.00-0.05mm), and 1.5° (range, 0.4°-2.7°), respectively. Dosimetrically, 44%, 81%, and 92% of all voxels differed by less than 1%, 3%, and 5% of the planned local dose, respectively. The median coverage reduction for the PTV was 1.1% (range in coverage difference, -7.8% to +0.8%), significantly depending on correlation (P=.026) and rotational (P=.005) error.With a 3-mm PTV margin, the median coverage change for the CTV was 0.0% (range, -1.0% to +5.4%), not significantly depending on any investigated parameter. In 42% of patients, the 3-mm margin did not fully compensate for the residual tracking errors, resulting in a CTV coverage reduction of 0.1% to 1.0%. For liver tumors treated with robotic stereotactic body radiation therapy, a safety margin of 3mm is not always sufficient to cover all residual tracking errors. Dosimetrically, this translates into only small CTV coverage reductions.

Prediction of Postoperative Intraocular Lens Position with Angle-to-Angle Depth Using Anterior Segment Optical Coherence Tomography

To evaluate the accuracy of a new formula for predicting postoperative anterior chamber depth (ACD) with preoperative angle-to-angle (ATA) depth using anterior segment (AS) optical coherence tomography (OCT) and to compare it with established methods. Retrospective consecutive case series. Three hundred four eyes (276 patients) implanted with acrylic intraocular lenses (IOLs) were divided randomly into a training set (152 eyes) and a validation set (152 eyes). Based on the training set data, the postoperative ACD measured 1 month after surgery was analyzed via multiple linear regression analysis with 5 preoperatively measured variables: ATA depth, ATA width, preoperative ACD measured with AS OCT, axial length (AL), and corneal power. A new regression formula for predicting postoperative ACD was developed using the results of the stepwise analysis. In the validation set data, the coefficients of determination (R2) between the measured postoperative ACD and the predicted postoperative ACD obtained using the new formula were compared with those obtained using the Sanders-Retzlaff-Kraff theoretic (SRK/T) and Haigis formulas. The absolute prediction errors were compared with each formula. Postoperative ACD, median absolute prediction error of postoperative ACD, and ocular biometric parameters. In the training set, ATA depth yielded the highest standard partial regression coefficient value, indicating that ATA depth is the most effective parameter for predicting postoperative ACD. The new regression formula was developed with 3 variables; ATA depth, preoperative ACD, and AL. In the validation set, the postoperative ACDs of the new formula, the SRK/T formula, and Haigis formula were predicted with R2 of 0.71, 0.36, and 0.55, respectively, and the medians of the absolute prediction errors were 0.10 mm, 0.65 mm, and 0.30 mm, respectively. The absolute prediction error with the new formula was significantly smaller than those obtained with the SRK/T and Haigis formulas (P < 0.0001). The new formula with 3 preoperative parameters-ATA depth, preoperative ACD, and AL-predicted postoperative ACD more accurately than the SRK/T and Haigis formulas. It may be possible to improve the accuracy of IOL power calculation using an improved postoperative ACD prediction with the ATA depth measured by AS OCT.

Comparison of Newer IOL Power Calculation Methods for Eyes With Previous Radial Keratotomy.

PurposeTo evaluate the accuracy of the optical coherence tomography–based (OCT formula) and Barrett True K (True K) intraocular lens (IOL) calculation formulas in eyes with previous radial keratotomy (RK).MethodsIn 95 eyes of 65 patients, using the actual refraction following cataract surgery as target refraction, the predicted IOL power for each method was calculated. The IOL prediction error (PE) was obtained by subtracting the predicted IOL power from the implanted IOL power. The arithmetic IOL PE and median refractive PE were calculated and compared.ResultsAll formulas except the True K produced hyperopic IOL PEs at 1 month, which decreased at ≥4 months (all P < 0.05). For the double-K Holladay 1, OCT formula, True K, and average of these three formulas (Average), the median absolute refractive PEs were, respectively, 0.78 diopters (D), 0.74 D, 0.60 D, and 0.59 D at 1 month; 0.69 D, 0.77 D, 0.77 D, and 0.61 D at 2 to 3 months; and 0.34 D, 0.65 D, 0.69 D, and 0.46 D at ≥4 months. The Average produced significantly smaller refractive PE than did the double-K Holladay 1 at 1 month (P < 0.05). There were no significant differences in refractive PEs among formulas at 4 months.ConclusionsThe OCT formula and True K were comparable to the double-K Holladay 1 method on the ASCRS (American Society of Cataract and Refractive Surgery) calculator. The Average IOL power on the ASCRS calculator may be considered when selecting the IOL power. Further improvements in the accuracy of IOL power calculation in RK eyes are desirable.

Development of a methylation marker set for forensic age estimation using analysis of public methylation data and the Agena Bioscience EpiTYPER system

Individual age estimation has the potential to provide key information that could enhance and extend DNA intelligence tools. Following predictive tests for externally visible characteristics developed in recent years, prediction of age could guide police investigations and improve the assessment of age-related phenotype expression patterns such as hair colour changes and early onset of male pattern baldness. DNA methylation at CpG positions has emerged as the most promising DNA tests to ascertain the individual age of the donor of a biological contact trace. Although different methodologies are available to detect DNA methylation, EpiTYPER technology (Agena Bioscience, formerly Sequenom) provides useful characteristics that can be applied as a discovery tool in localized regions of the genome. In our study, a total of twenty-two candidate genomic regions, selected from the assessment of publically available data from the Illumina HumanMethylation 450 BeadChip, had a total of 177 CpG sites with informative methylation patterns that were subsequently investigated in detail. From the methylation analyses made, a novel age prediction model based on a multivariate quantile regression analysis was built using the seven highest age-correlated loci of ELOVL2, ASPA, PDE4C, FHL2, CCDC102B, C1orf132 and chr16:85395429. The detected methylation levels in these loci provide a median absolute age prediction error of ±3.07years and a percentage of prediction error relative to the age of 6.3%. We report the predictive performance of the developed model using cross validation of a carefully age-graded training set of 725 European individuals and a test set of 52 monozygotic twin pairs. The multivariate quantile regression age predictor, using the CpG sites selected in this study, has been placed in the open-access Snipper forensic classification website.

Development and Evaluation of a Gentamicin Pharmacokinetic Model That Facilitates Opportunistic Gentamicin Therapeutic Drug Monitoring in Neonates and Infants.

Trough gentamicin therapeutic drug monitoring (TDM) is time-consuming, disruptive to neonatal clinical care, and a patient safety issue. Bayesian models could allow TDM to be performed opportunistically at the time of routine blood tests. This study aimed to develop and prospectively evaluate a new gentamicin model and a novel Bayesian computer tool (neoGent) for TDM use in neonatal intensive care. We also evaluated model performance for predicting peak concentrations and the area under the concentration-time curve from time 0 h to time t h (AUC0- t). A pharmacokinetic meta-analysis was performed on pooled data from three studies (1,325 concentrations from 205 patients). A 3-compartment model was used with the following covariates: allometric weight scaling, postmenstrual and postnatal age, and serum creatinine concentration. Final parameter estimates (standard errors) were as follows: clearance, 6.2 (0.3) liters/h/70 kg of body weight; central volume (V), 26.5 (0.6) liters/70 kg; intercompartmental disposition (Q), 2.2 (0.3) liters/h/70 kg; peripheral volume V2, 21.2 (1.5) liters/70 kg; intercompartmental disposition (Q2), 0.3 (0.05) liters/h/70 kg; peripheral volume V3, 148 (52.0) liters/70 kg. The model's ability to predict trough concentrations from an opportunistic sample was evaluated in a prospective observational cohort study that included data from 163 patients and 483 concentrations collected in five hospitals. Unbiased trough predictions were obtained; the median (95% confidence interval [CI]) prediction error was 0.0004 (-1.07, 0.84) mg/liter. Results also showed that peaks and AUC0- t values could be predicted (from one randomly selected sample) with little bias but relative imprecision, with median (95% CI) prediction errors being 0.16 (-4.76, 5.01) mg/liter and 10.8 (-24.9, 62.2) mg · h/liter, respectively. neoGent was implemented in R/NONMEM and in the freely available TDMx software.

Performance of a closed-loop feedback computer-controlled infusion system for maintaining blood pressure during spinal anaesthesia for caesarean section: a randomized controlled comparison of norepinephrine versus phenylephrine

Closed-loop feedback computer-controlled vasopressor infusion has been previously described for maintaining blood pressure during spinal anaesthesia for caesarean section but there are limited data available comparing the relative performance of different vasopressors. The aim of this study was to compare the performance of norepinephrine versus phenylephrine in this system. Data from a randomized, two-arm parallel group, double-blinded controlled trial were reanalyzed. 104 patients scheduled for elective caesarean section under spinal anaesthesia were randomized to receive computer-controlled closed-loop infusion of either norepinephrine 5µgml-1 or phenylephrine 100µgml-1. This was started immediately after induction of spinal anaesthesia and used an algorithm designed to maintain systolic blood pressure near baseline until fetal delivery. Performance error calculations were used to compare the performance of the two vasopressors. The primary outcome was defined as the median absolute performance error. Median performance error, wobble and divergence were also compared. Median absolute performance error was smaller in the norepinephrine group (median 3.79 [interquartile range 2.82-5.17] %) versus the phenylephrine group (4.70 [3.23-6.57] %, P=0.028). In addition, median performance error was smaller (0.75 [-1.56-2.52] %) versus 2.61 [0.83-4.57] %, P=0.002) and wobble was smaller (2.85 [2.07-5.17] %) versus 3.39 [2.62-4.90] %, P=0.028) in the norepinephrine group versus the phenylephrine group. Divergence was similar between groups. The precision of the control of blood pressure was greater with norepinephrine compared with phenylephrine at the drug concentrations used.

Evaluation and optimisation of propofol pharmacokinetic parameters in cats for target-controlled infusion

The aim of this study was to develop and evaluate a pharmacokinetic model-driven infusion of propofol in premedicated cats. In a first step, propofol (10 mg/kg) was administered intravenously over...

Accuracy of the Barrett True-K formula for intraocular lens power prediction after laser in situ keratomileusis or photorefractive keratectomy for myopia

To compare the accuracy of the Barrett True-K formula with other methods available on the American Society of Cataract and Refractive Surgery (ASCRS) post-refractive surgery intraocular lens (IOL) power calculator for the prediction of IOL power after previous myopic laser in situ keratomileusis (LASIK) or photorefractive keratectomy (PRK). Cullen Eye Institute, Baylor College of Medicine, Houston, Texas, and private practice, Mesa, Arizona, USA. Retrospective case series. The accuracy of the Barrett True-K formula was compared with the Adjusted Atlas (4.0 mm zone), Masket, modified-Masket, Wang-Koch-Maloney, Shammas, and Haigis-L methods to calculate IOL power. A separate analysis of 2 no-history methods (Shammas and Haigis-L) was performed and compared with the Barrett True-K no-history option. Eighty-eight eyes were available for analysis. The Barrett True-K formula had a significantly smaller median absolute refraction prediction error than all other formulas except the Masket, smaller variances compared with the Wang-Koch-Maloney, Shammas, and Haigis-L, and a greater percentage of eyes within ±0.50 diopter (D) of predicted error in refraction compared with the Adjusted Atlas, Masket, and modified Masket methods (all P < .05). In eyes with no historical data, the Barrett True-K no-history formula had a significantly smaller median absolute refraction prediction error and a greater percentage of eyes within ±0.50 D of the predicted error in refraction than the Shammas and the Haigis-L formulas (both P < .05). The Barrett True-K formula was either equal to or better than alternative methods available on the ASCRS online calculator for predicting IOL power in eyes with previous myopic LASIK or PRK.

The value of surrogate endpoints for predicting real-world survival across five cancer types

Objective It is unclear how well different outcome measures in randomized controlled trials (RCTs) perform in predicting real-world cancer survival. We assess the ability of RCT overall survival (OS) and surrogate endpoints – progression-free survival (PFS) and time to progression (TTP) – to predict real-world OS across five cancers.Methods We identified 20 treatments and 31 indications for breast, colorectal, lung, ovarian, and pancreatic cancer that had a phase III RCT reporting median OS and median PFS or TTP. Median real-world OS was determined using a Kaplan–Meier estimator applied to patients in the Surveillance and Epidemiology End Results (SEER)-Medicare database (1991–2010). Performance of RCT OS and PFS/TTP in predicting real-world OS was measured using t-tests, median absolute prediction error, and R2 from linear regressions.Results Among 72,600 SEER-Medicare patients similar to RCT participants, median survival was 5.9 months for trial surrogates, 14.1 months for trial OS, and 13.4 months for real-world OS. For this sample, regression models using clinical trial OS and trial surrogates as independent variables predicted real-world OS significantly better than models using surrogates alone (P = 0.026). Among all real-world patients using sample treatments (N = 309,182), however, adding trial OS did not improve predictive power over predictions based on surrogates alone (P = 0.194). Results were qualitatively similar using median absolute prediction error and R2 metrics.Conclusions Among the five tumor types investigated, trial OS and surrogates were each independently valuable in predicting real-world OS outcomes for patients similar to trial participants. In broader real-world populations, however, trial OS added little incremental value over surrogates alone.