Molly Leecaster

Modeling the variations in pediatric respiratory syncytial virus seasonal epidemics

BackgroundSeasonal respiratory syncytial virus (RSV) epidemics occur annually in temperate climates and result in significant pediatric morbidity and increased health care costs. Although RSV epidemics generally occur between October and April, the size and timing vary across epidemic seasons and are difficult to predict accurately. Prediction of epidemic characteristics would support management of resources and treatment.MethodsThe goals of this research were to examine the empirical relationships among early exponential growth rate, total epidemic size, and timing, and the utility of specific parameters in compartmental models of transmission in accounting for variation among seasonal RSV epidemic curves. RSV testing data from Primary Childrens Medical Center were collected on children under two years of age (July 2001-June 2008). Simple linear regression was used explore the relationship between three epidemic characteristics (final epidemic size, days to peak, and epidemic length) and exponential growth calculated from four weeks of daily case data. A compartmental model of transmission was fit to the data and parameter estimated used to help describe the variation among seasonal RSV epidemic curves.ResultsThe regression results indicated that exponential growth was correlated to epidemic characteristics. The transmission modeling results indicated that start time for the epidemic and the transmission parameter co-varied with the epidemic season.ConclusionsThe conclusions were that exponential growth was somewhat empirically related to seasonal epidemic characteristics and that variation in epidemic start date as well as the transmission parameter over epidemic years could explain variation in seasonal epidemic size. These relationships are useful for public health, health care providers, and infectious disease researchers.

The role of heterogeneity in contact timing and duration in network models of influenza spread in schools

Influenza poses a significant health threat to children, and schools may play a critical role in community outbreaks. Mathematical outbreak models require assumptions about contact rates and patterns among students, but the level of temporal granularity required to produce reliable results is unclear. We collected objective contact data from students aged 5–14 at an elementary school and middle school in the state of Utah, USA, and paired those data with a novel, data-based model of influenza transmission in schools. Our simulations produced within-school transmission averages consistent with published estimates. We compared simulated outbreaks over the full resolution dynamic network with simulations on networks with averaged representations of contact timing and duration. For both schools, averaging the timing of contacts over one or two school days caused average outbreak sizes to increase by 1–8%. Averaging both contact timing and pairwise contact durations caused average outbreak sizes to increase by 10% at the middle school and 72% at the elementary school. Averaging contact durations separately across within-class and between-class contacts reduced the increase for the elementary school to 5%. Thus, the effect of ignoring details about contact timing and duration in school contact networks on outbreak size modelling can vary across different schools.

WRENMining: Large-Scale Data Collection for Human Contact Network Research

Wireless sensor networks (WSNs) have come a long way to reach their ubiquitous state known today through scalable cost, low-power optimizations, and data management. As WSNs scale in size, the necessity for system designs - from low-level hardware implementations to data collection and management procedures - to account for handling extensive amounts of data is crucial. Several prominent papers address these issues for limited deployments of less than 200 nodes, but there are little resources available for multiple consecutive deployments of over 500 nodes. We present the engineering perspective on sensor data collection, management, and processing while collaborating with epidemiologists for the Wireless Ranging Enabled Node (WREN) network system for human contact research. The WREN and all supporting systems (base stations, software, and data procedures) sustain multiple high density, mobile deployments with fast turnovers. The WRENs completed 13 deployments over a period of 8 months to mine over 35 million contact points. We present our design considerations, challenges/experiences, and solutions to account for and correct time synchronization issues, along with our methodology for collecting, managing, and processing data.

SaTScan on a Cloud: On-Demand Large Scale Spatial Analysis of Epidemics.

By using cloud computing it is possible to provide on- demand resources for epidemic analysis using computer intensive applications like SaTScan. Using 15 virtual machines (VM) on the Nimbus cloud we were able to reduce the total execution time for the same ensemble run from 8896 seconds in a single machine to 842 seconds in the cloud. Using the caBIG tools and our iterative software development methodology the time required to complete the implementation of the SaTScan cloud system took approximately 200 man-hours, which represents an effort that can be secured within the resources available at State Health Departments. The approach proposed here is technically advantageous and practically possible.

Estimates of Social Contact in a Middle School Based on Self-Report and Wireless Sensor Data

Estimates of contact among children, used for infectious disease transmission models and understanding social patterns, historically rely on self-report logs. Recently, wireless sensor technology has enabled objective measurement of proximal contact and comparison of data from the two methods. These are mostly small-scale studies, and knowledge gaps remain in understanding contact and mixing patterns and also in the advantages and disadvantages of data collection methods. We collected contact data from a middle school, with 7th and 8th grades, for one day using self-report contact logs and wireless sensors. The data were linked for students with unique initials, gender, and grade within the school. This paper presents the results of a comparison of two approaches to characterize school contact networks, wireless proximity sensors and self-report logs. Accounting for incomplete capture and lack of participation, we estimate that “sensor-detectable”, proximal contacts longer than 20 seconds during lunch and class-time occurred at 2 fold higher frequency than “self-reportable” talk/touch contacts. Overall, 55% of estimated talk-touch contacts were also sensor-detectable whereas only 15% of estimated sensor-detectable contacts were also talk-touch. Contacts detected by sensors and also in self-report logs had longer mean duration than contacts detected only by sensors (6.3 vs 2.4 minutes). During both lunch and class-time, sensor-detectable contacts demonstrated substantially less gender and grade assortativity than talk-touch contacts. Hallway contacts, which were ascertainable only by proximity sensors, were characterized by extremely high degree and short duration. We conclude that the use of wireless sensors and self-report logs provide complementary insight on in-school mixing patterns and contact frequency.

Assessing the potential adoption and usefulness of concurrent, action-oriented, electronic adverse drug event triggers designed for the outpatient setting.

Background: Adverse drug event (ADE) detection is an important priority for patient safety research. Trigger tools have been developed to help identify ADEs. In previous work we developed seven concurrent, action-oriented, electronic trigger algorithms designed to prompt clinicians to address ADEs in outpatient care. Objectives: We assessed the potential adoption and usefulness of the seven triggers by testing the positive predictive validity and obtaining stakeholder input. Methods: We adapted ADE triggers, “bone marrow toxin—white blood cell count (BMT-WBC),” “bone marrow toxin - platelet (BMT-platelet),” “potassium raisers,” “potassium reducers,” “creatinine,” “warfarin,” and “sedative hypnotics,” with logic to suppress flagging events with evidence of clinical intervention and applied the triggers to 50,145 patients from three large health care systems. Four pharmacists assessed trigger positive predictive value (PPV) with respect to ADE detection (conservatively excluding ADEs occurring during clinically appropriate care) and clinical usefulness (i.e., whether the trigger alert could change care to prevent harm). We measured agreement between raters using the free kappa and assessed positive PPV for the trigger’s detection of harm, clinical usefulness, and both. Stakeholders from the participating health care systems rated the likelihood of trigger adoption and the perceived ease of implementation. Findings: Agreement between pharmacist raters was moderately high for each ADE trigger (kappa free > 0.60). Trigger PPVs for harm ranged from 0 (Creatinine, BMT-WBC) to 17 percent (potassium raisers), while PPV for care change ranged from 0 (WBC) to 60 percent (Creatinine). Fifteen stakeholders rated the triggers. Our assessment identified five of the seven triggers as good candidates for implementation: Creatinine, BMT-Platelet, Potassium Raisers, Potassium Reducers, and Warfarin. Conclusions: At least five outpatient ADE triggers performed well and merit further evaluation in outpatient clinical care. When used in real time, these triggers may promote care changes to ameliorate patient harm.

Efficient parameter estimation for models of healthcare-associated pathogen transmission in discrete and continuous time

We describe two novel Markov chain Monte Carlo approaches to computing estimates of parameters concerned with healthcare-associated infections. The first approach frames the discrete time, patient level, hospital transmission model as a Bayesian network, and exploits this framework to improve greatly on the computational efficiency of estimation compared with existing programs. The second approach is in continuous time and shares the same computational advantages. Both methods have been implemented in programs that are available from the authors. We use these programs to show that time discretization can lead to statistical bias in the underestimation of the rate of transmission of pathogens. We show that the continuous implementation has similar running time to the discrete implementation, has better Markov chain mixing properties, and eliminates the potential statistical bias. We, therefore, recommend its use when continuous-time data are available.

A dynamic transmission model to evaluate the effectiveness of infection control strategies

Abstract Background The advancement of knowledge about control of antibiotic resistance depends on the rigorous evaluation of alternative intervention strategies. The STAR*ICU trial examined the effects of active surveillance and expanded barrier precautions on acquisition of methicillin-resistant Staphylococcus aureus (MRSA) and vancomycin-resistant Enterococcus (VRE) in intensive care units. We report a reanalyses of the STAR*ICU trial using a Bayesian transmission modeling framework. Methods The data included admission and discharge times and surveillance test times and results. Markov chain Monte Carlo stochastic integration was used to estimate the transmission rate, importation, false negativity, and clearance separately for MRSA and VRE. The primary outcome was the intervention effect, which when less than (or greater than) zero, indicated a decreased (or increased) transmission rate attributable to the intervention. Results The transmission rate increased in both arms from pre- to postintervention (by 20% and 26% for MRSA and VRE). The estimated intervention effect was 0.00 (95% confidence interval [CI], −0.57 to 0.56) for MRSA and 0.05 (95% CI, −0.39 to 0.48) for VRE. Compared with MRSA, VRE had a higher transmission rate (preintervention, 0.0069 vs 0.0039; postintervention, 0.0087 vs 0.0046), higher importation probability (0.22 vs 0.17), and a lower clearance rate per colonized patient-day (0.016 vs 0.035). Conclusions Transmission rates in the 2 treatment arms were statistically indistinguishable from the pre- to postintervention phase, consistent with the original analysis of the STAR*ICU trial. Our statistical framework was able to disentangle transmission from importation and account for imperfect testing. Epidemiological differences between VRE and MRSA were revealed.

Improved hidden Markov model for nosocomial infections

We propose a novel hidden Markov model (HMM) for parameter estimation in hospital transmission models, and show that commonly made simplifying assumptions can lead to severe model misspecification and poor parameter estimates. A standard HMM that embodies two commonly made simplifying assumptions, namely a fixed patient count and binomially distributed detections is compared with a new alternative HMM that does not require these simplifying assumptions. Using simulated data, we demonstrate how each of the simplifying assumptions used by the standard model leads to model misspecification, whereas the alternative model results in accurate parameter estimates.

Extended models for nosocomial infection: parameter estimation and model selection

Abstract We consider extensions to previous models for patient level nosocomial infection in several ways, provide a specification of the likelihoods for these new models, specify new update steps required for stochastic integration, and provide programs that implement these methods to obtain parameter estimates and model choice statistics. Previous susceptible-infected models are extended to allow for a latent period between initial exposure to the pathogen and the patient becoming themselves infectious, and the possibility of decolonization. We allow for multiple facilities, such as acute care hospitals or long-term care facilities and nursing homes, and for multiple units or wards within a facility. Patient transfers between units and facilities are tracked and accounted for in the models so that direct importation of a colonized individual from one facility or unit to another might be inferred. We allow for constant transmission rates, rates that depend on the number of colonized individuals in a unit or facility, or rates that depend on the proportion of colonized individuals. Statistical analysis is done in a Bayesian framework using Markov chain Monte Carlo methods to obtain a sample of parameter values from their joint posterior distribution. Cross validation, deviance information criterion and widely applicable information criterion approaches to model choice fit very naturally into this framework and we have implemented all three. We illustrate our methods by considering model selection issues and parameter estimation for data on methicilin-resistant Staphylococcus aureus surveillance tests over 1 year at a Veterans Administration hospital comprising seven wards.