arXiv subjects

Julie A. Simpson

Publications and source records attributed to Julie A. Simpson.

7 recordsLinked to original sources

arXiv2025

Handling multivariable missing data in causal mediation analysis estimating interventional effects

The interventional effects approach to causal mediation analysis is increasingly common in epidemiologic research, given its potential to address policy-relevant questions about hypothetical mediator interventions. Multiple imputation (MI) is widely used for handling missing data in epidemiologic studies. However, guidance is lacking on best practices for using MI when estimating interventional mediation effects, specifically regarding the role of the missingness mechanism in the method's performance, how to appropriately specify the MI model when g-computation is used for effect estimation, and suitable approaches to variance estimation. To address this gap, we conducted simulations based on the Victorian Adolescent Health Cohort Study. We considered seven missingness mechanisms involving varying assumptions about the influence of an intermediate confounder, a mediator, and/or the outcome on missingness in key variables. We compared the performance of complete-case analysis, six MI approaches using fully conditional specification (differing in how the imputation model was tailored), and a "substantive model compatible" multiple imputation-fully conditional specification approach. We evaluated MIBoot (MI, then bootstrap) and BootMI (bootstrap, then MI) approaches for variance estimation. All MI approaches, apart from those clearly diverging from best practice, yielded approximately unbiased estimates when none of the intermediate confounder, mediator, and outcome variables influenced missingness in any of these variables, and showed non-negligible bias otherwise. We observed the largest bias for interventional effects when each of the intermediate confounders, mediators, and outcomes influenced their own missingness. BootMI returned variance estimates with smaller bias than MIBoot.

stat.AP↗

arXiv2024

Handling missing data when estimating causal effects with Targeted Maximum Likelihood Estimation

Targeted Maximum Likelihood Estimation (TMLE) is increasingly used for doubly robust causal inference, but how missing data should be handled when using TMLE with data-adaptive approaches is unclear. Based on the Victorian Adolescent Health Cohort Study, we conducted a simulation study to evaluate eight missing data methods in this context: complete-case analysis, extended TMLE incorporating outcome-missingness model, missing covariate missing indicator method, five multiple imputation (MI) approaches using parametric or machine-learning models. Six scenarios were considered, varying in exposure/outcome generation models (presence of confounder-confounder interactions) and missingness mechanisms (whether outcome influenced missingness in other variables and presence of interaction/non-linear terms in missingness models). Complete-case analysis and extended TMLE had small biases when outcome did not influence missingness in other variables. Parametric MI without interactions had large bias when exposure/outcome generation models included interactions. Parametric MI including interactions performed best in bias and variance reduction across all settings, except when missingness models included a non-linear term. When choosing a method to handle missing data in the context of TMLE, researchers must consider the missingness mechanism and, for MI, compatibility with the analysis method. In many settings, a parametric MI approach that incorporates interactions and non-linearities is expected to perform well.

stat.ME↗

arXiv2022

A model for malaria treatment evaluation in the presence of multiple species

Plasmodium (P.) falciparum and P. vivax are the two most common causes of malaria. While the majority of deaths and severe morbidity are due to P. falciparum, P. vivax poses a greater challenge to eliminating malaria outside of Africa due to its ability to form latent liver stage parasites (hypnozoites), which can cause relapsing episodes within an individual patient. In areas where P. falciparum and P. vivax are co-endemic, individuals can carry parasites of both species simultaneously. These mixed infections complicate dynamics in several ways; treatment of mixed infections will simultaneously affect both species, P. falciparum can mask the detection of P. vivax, and it has been hypothesised that clearing P. falciparum may trigger a relapse of dormant P. vivax. When mixed infections are treated for only blood-stage parasites, patients are at risk of relapse infections due to P. vivax hypnozoites. We present a stochastic mathematical model that captures interactions between P. falciparum and P. vivax, and incorporates both standard schizontocidal treatment (which targets blood-stage parasites) and radical treatment (which additionally targets liver-stage parasites). We apply this model to assess the implications of different treatment coverage of radical cure for mixed and P. vivax infections and a so-called "unified radical cure" treatment strategy for P. falciparum, P. vivax and mixed infections. We find that a unified radical cure strategy, with G6PD screening, leads to a substantially lower incidence of malaria cases and deaths overall. We perform a one-way sensitivity analysis to highlight important model parameters.

q-bio.PE↗

arXiv2022

Making SMART decisions in prophylaxis and treatment studies

The optimal prophylaxis, and treatment if the prophylaxis fails, for a disease may be best evaluated using a sequential multiple assignment randomised trial (SMART). A SMART is a multi-stage study that randomises a participant to an initial treatment, observes some response to that treatment and then, depending on their observed response, randomises the same participant to an alternative treatment. Response adaptive randomisation may, in some settings, improve the trial participants' outcomes and expedite trial conclusions, compared to fixed randomisation. But 'myopic' response adaptive randomisation strategies, blind to multistage dynamics, may also result in suboptimal treatment assignments. We propose a 'dynamic' response adaptive randomisation strategy based on Q-learning, an approximate dynamic programming algorithm. Q-learning uses stage-wise statistical models and backward induction to incorporate late-stage 'payoffs' (i.e. clinical outcomes) into early-stage 'actions' (i.e. treatments). Our real-world example consists of a COVID-19 prophylaxis and treatment SMART with qualitatively different binary endpoints at each stage. Standard Q-learning does not work with such data because it cannot be used for sequences of binary endpoints. Sequences of qualitatively distinct endpoints may also require different weightings to ensure that the design guides participants to regimens with the highest utility. We describe how a simple decision-theoretic extension to Q-learning can be used to handle sequential binary endpoints with distinct utilities. Using simulation we show that, under a set of binary utilities, the 'dynamic' approach increases expected participant utility compared to the fixed approach, sometimes markedly, for all model parameters, whereas the 'myopic' approach can actually decrease utility.

stat.ME↗

arXiv2018

Sequential infection experiments for quantifying innate and adaptive immunity during influenza infection

Laboratory models are often used to understand the interaction of related pathogens via host immunity. For example, recent experiments where ferrets were exposed to two influenza strains within a short period of time have shown how the effects of cross-immunity vary with the time between exposures and the specific strains used. On the other hand, studies of the workings of different arms of the immune response, and their relative importance, typically use experiments involving a single infection. However, inferring the relative importance of different immune components from this type of data is challenging. Using simulations and mathematical modelling, here we investigate whether the sequential infection experiment design can be used not only to determine immune components contributing to cross-protection, but also to gain insight into the immune response during a single infection. We show that virological data from sequential infection experiments can be used to accurately extract the timing and extent of cross-protection. Moreover, the broad immune components responsible for such cross-protection can be determined. Such data can also be used to infer the timing and strength of some immune components in controlling a primary infection, even in the absence of serological data. By contrast, single infection data cannot be used to reliably recover this information. Hence, sequential infection data enhances our understanding of the mechanisms underlying the control and resolution of infection, and generates new insight into how previous exposure influences the time course of a subsequent infection.

q-bio.PE↗

arXiv2017

A mechanistic model quantifies artemisinin-induced parasite growth retardation in blood-stage Plasmodium falciparum infection

Falciparum malaria is a major parasitic disease causing widespread morbidity and mortality globally. Artemisinin derivatives---the most effective and widely-used antimalarials that have helped reduce the burden of malaria by 60% in some areas over the past decade---have recently been found to induce growth retardation of blood-stage Plasmodium falciparum when applied at clinically relevant concentrations. To date, no model has been designed to quantify the growth retardation effect and to predict the influence of this property on in vivo parasite killing. Here we introduce a mechanistic model of parasite growth from the ring to trophozoite stage of the parasite's life cycle, and by modelling the level of staining with an RNA-binding dye, we demonstrate that the model is able to reproduce fluorescence distribution data from in vitro experiments using the laboratory 3D7 strain. We quantify the dependence of growth retardation on drug concentration and demonstrate the model's utility as a platform to propose experimentally-testable mechanisms of growth retardation. Furthermore we illustrate that a drug-induced delay in growth may significantly influence in vivo parasite dynamics, demonstrating the importance of considering growth retardation in the design of optimal artemisinin-based dosing regimens.

q-bio.PE↗

arXiv2016

A dynamic stress model explains the delayed drug effect in artemisinin treatment of Plasmodium falciparum

Artemisinin resistance constitutes a major threat to the continued success of control programs for malaria. With alternative antimalarial drugs not yet available, improving our understanding of how artemisinin-based drugs act and how resistance manifests is essential to enable optimisation of dosing regimens in order to prolong the lifespan of current first-line treatment options. Here, through introduction of a novel model of the dynamics of the parasites' response to drug, we explore how artemisinin-based therapies may be adjusted to maintain efficacy and how artemisinin resistance may manifest and be overcome. We introduce a dynamic mathematical model, extending on the traditional pharmacokinetic-pharmacodynamic framework, to capture the time-dependent development of a stress response in parasites. We fit the model to in vitro data and establish that the parasites' stress response explains the recently identified complex interplay between drug concentration, exposure time and parasite viability. Our model demonstrates that the previously reported hypersensitivity of early ring stage parasites of the 3D7 strain to dihydroartemisinin (DHA) is primarily due to the rapid development of stress, rather than any change in the maximum achievable killing rate. Of direct clinical relevance, we demonstrate that the complex temporal features of artemisinin action observed in vitro have a significant impact on predictions of in vivo parasite clearance using PK-PD models. Given the important role that such models play in the design and evaluation of clinical trials for alternative drug dosing regimens, our model contributes an enhanced predictive platform for the continued efforts to minimise the burden of malaria.

q-bio.PE↗