Dataset Information

Comparing Strategies for Modeling Competing Risks in Discrete-Event Simulations: A Simulation Study and Illustration in Colorectal Cancer.

ABSTRACT: BACKGROUND:Different strategies toward implementing competing risks in discrete-event simulation (DES) models are available. This study aims to provide recommendations regarding modeling approaches that can be defined based on these strategies by performing a quantitative comparison of alternative modeling approaches. METHODS:Four modeling approaches were defined: 1) event-specific distribution (ESD), 2) event-specific probability and distribution (ESPD), 3) unimodal joint distribution and regression model (UDR), and 4) multimodal joint distribution and regression model (MDR). Each modeling approach was applied to uncensored individual patient data in a simulation study and a case study in colorectal cancer. Their performance was assessed in terms of relative event incidence difference, relative absolute event incidence difference, and relative entropy of time-to-event distributions. Differences in health economic outcomes were also illustrated for the case study. RESULTS:In the simulation study, the ESPD and MDR approaches outperformed the ESD and UDR approaches, in terms of both event incidence differences and relative entropy. Disease pathway and data characteristics, such as the number of competing risks and overlap between competing time-to-event distributions, substantially affected the approaches' performance. Although no considerable differences in health economic outcomes were observed, the case study showed that the ESPD approach was most sensitive to low event rates, which negatively affected performance. CONCLUSIONS:Based on overall performance, the recommended modeling approach for implementing competing risks in DES models is the MDR approach, which is defined according to the general strategy of selecting the time-to-event first and the corresponding event second. The ESPD approach is a less complex and equally performing alternative if sufficient observations are available for each competing event (i.e., the internal validity shows appropriate data representation).

SUBMITTER: Degeling K

PROVIDER: S-EPMC6311678 | biostudies-literature | 2019 Jan

REPOSITORIES: biostudies-literature

ACCESS DATA

Similar Datasets

Project description:BackgroundHometime, the total number of days a person is living in the community (not in a healthcare institution) in a defined period of time after a hospitalization, is a patient-centred outcome metric increasingly used in healthcare research. Hometime exhibits several properties which make its statistical analysis difficult: it has a highly non-normal distribution, excess zeros, and is bounded by both a lower and upper limit. The optimal methodology for the analysis of hometime is currently unknown.MethodsUsing administrative data we identified adult patients diagnosed with stroke between April 1, 2010 and December 31, 2017 in Ontario, Canada. 90-day hometime and clinically relevant covariates were determined through administrative data linkage. Fifteen different statistical and machine learning models were fit to the data using a derivation sample. The models' predictive accuracy and bias were assessed using an independent validation sample.ResultsSeventy-five thousand four hundred seventy-five patients were identified (divided into a derivation set of 49,402 and a test set of 26,073). In general, the machine learning models had lower root mean square error and mean absolute error than the statistical models. However, some statistical models resulted in lower (or equal) bias than the machine learning models. Most of the machine learning models constrained predicted values between the minimum and maximum observable hometime values but this was not the case for the statistical models. The machine learning models also allowed for the display of complex non-linear interactions between covariates and hometime. No model captured the non-normal bucket shaped hometime distribution.ConclusionsOverall, no model clearly outperformed the others. However, it was evident that machine learning methods performed better than traditional statistical methods. Among the machine learning methods, generalized boosting machines using the Poisson distribution as well as random forests regression were the best performing. No model was able to capture the bucket shaped hometime distribution and future research on factors which are associated with extreme values of hometime that are not available in administrative data is warranted.

Project description:It is recognized that the study of the disaster medical response (DMR) is a relatively new field. To date, there is no evidence-based literature that clearly defines the best medical response principles, concepts, structures and processes in a disaster setting. Much of what is known about the DMR results from descriptive studies and expert opinion. No experimental studies regarding the effects of DMR interventions on the health outcomes of disaster survivors have been carried out. Traditional analytic methods cannot fully capture the flow of disaster victims through a complex disaster medical response system (DMRS). Computer modelling and simulation enable to study and test operational assumptions in a virtual but controlled experimental environment. The SIMEDIS (Simulation for the assessment and optimization of medical disaster management) simulation model consists of 3 interacting components: the victim creation model, the victim monitoring model where the health state of each victim is monitored and adapted to the evolving clinical conditions of the victims, and the medical response model, where the victims interact with the environment and the resources at the disposal of the healthcare responders. Since the main aim of the DMR is to minimize as much as possible the mortality and morbidity of the survivors, we designed a victim-centred model in which the casualties pass through the different components and processes of a DMRS. The specificity of the SIMEDIS simulation model is the fact that the victim entities evolve in parallel through both the victim monitoring model and the medical response model. The interaction between both models is ensured through a time or medical intervention trigger. At each service point, a triage is performed together with a decision on the disposition of the victims regarding treatment and/or evacuation based on a priority code assigned to the victim and on the availability of resources at the service point. The aim of the case study is to implement the SIMEDIS model to the DMRS of an international airport and to test the medical response plan to an airplane crash simulation at the airport. In order to identify good response options, the model then was used to study the effect of a number of interventional factors on the performance of the DMRS. Our study reflects the potential of SIMEDIS to model complex systems, to test different aspects of DMR, and to be used as a tool in experimental research that might make a substantial contribution to provide the evidence base for the effectiveness and efficiency of disaster medical management.

Project description:ObjectivesThis feasibility study aimed to model in silico the current healthcare system for patients triaged to a primary care disposition following a call to National Health Service (NHS) 111 and determine the effect of reconfiguring the healthcare system to ensure a timely primary care service contact.DesignDiscrete event simulation.SettingSingle English NHS 111 call centre in Yorkshire.ParticipantsCallers registered with a Bradford general practitioner who contacted the NHS 111 service in 2021 and were triaged to a primary care disposition.Primary and secondary outcome measuresFace validity of conceptual model. Comparison between real and simulated data for quarterly counts (and 95% CIs) for patient contact with emergency ambulance (999), 111, and primary and secondary care services. Mean difference and 95% CIs in healthcare system usage between simulations and difference in mean proportion of avoidable admissions for callers who presented to an emergency department (ED).ResultsThe simulation of the current system estimated that there would be 39 283 (95% CI 39 237 to 39 328) primary care contacts, 2042 (95% CI 2032 to 2051) 999 calls and 1120 (95% CI 1114 to 1127) avoidable ED attendances. Modifying the model to ensure a timely primary care response resulted in a mean percentage increase of 196.1% (95% CI 192.2% to 199.9%) in primary care contacts, and a mean percentage decrease of 78.0% (95% CI 69.8% to 86.2%) in 999 calls and 88.1% (95% CI 81.7% to 94.5%) in ED attendances. Avoidable ED attendances reduced by a mean of -26 (95% CI -35 to -17).ConclusionIn this simulated study, ensuring timely contact with a primary care service would lead to a significant reduction in 999 and 111 calls, and ED attendances (although not avoidable ED attendance). However, this is likely to be impractical given the need to almost double current primary care service provision. Further economic and qualitative research is needed to determine whether this intervention would be cost-effective and acceptable to both patients and primary care clinicians.

Dataset Information

Comparing Strategies for Modeling Competing Risks in Discrete-Event Simulations: A Simulation Study and Illustration in Colorectal Cancer.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets