Epidemiologic Methods最新文献

Objectives: In drug development, there is increasing interest in leveraging real-world data (RWD) to augment trial data and generate evidence about treatment efficacy. However, comparing patient outcomes across trial and routine clinical care settings can be susceptible to bias, namely due to differences in how and when disease assessments occur. This can introduce measurement error in RWD relative to trial standards and lead to bias when comparing endpoints. We develop a novel statistical method, survival regression calibration (SRC), to mitigate measurement error bias in time-to-event RWD outcomes and improve inferences when combining trials with RWD in oncology.

Methods: SRC extends upon existing regression calibration methods to address measurement error in time-to-event RWD outcomes. The method entails fitting separate Weibull regression models using trial-like ('true') and real-world-like ('mismeasured') outcome measures in a validation sample, and then calibrating parameter estimates in the full study according to the estimated bias in Weibull parameters. We evaluate performance of SRC under varying degrees of existing measurement error bias via simulation, and then illustrate how SRC can address measurement error when estimating median progression-free survival (mPFS) in newly diagnosed multiple myeloma RWD.

Results: When measurement error exists between trial and real-world mPFS, SRC can effectively account for its resulting bias. SRC yields greater reduction in measurement error bias than standard regression calibration methods, due to its suitability for time-to-event outcomes.

Conclusions: Outcome measurement error is important to address when combining trials and RWD, as it may lead to biased results. Our SRC method helps mitigate such bias, improving comparability between real-world and trial endpoints and strengthening evidence of treatment efficacy.

Objectives: The use of substances in adolescents is an increasing public health problem. Many high schools in Canada have implemented school-based programs to mitigate student substance use, but their utility is not conclusive. Polysubstance use data collected on students from multiple schools may be subject to informative cluster size (ICS). The objective of this study was to investigate whether a multivariate analysis approach that addresses ICS provides different conclusions from univariate analyses and methods that do not account for ICS.

Methods: We used data from the 2018/2019 cycle of the Cannabis, Obesity, Mental health, Physical activity, Alcohol, Smoking, and Sedentary Behaviour (COMPASS) study, an ongoing prospective cohort study that annually collects data from Canadian high schools and students. We compared results from four analytical approaches that estimate marginal associations between each school substance program and the four substance use behaviours (binge drinking, cannabis, e-cigarette, and cigarette): univariate generalized estimating equations (GEE), univariate cluster-weighted GEE (CWGEE), multivariate GEE, and multivariate CWGEE.

Results: We observed that the proportion of students who engage in each of the four behaviours was higher in small schools and lower in large schools. In general, the univariate and multivariate analyses produced comparable results. Some differences existed between multivariate CWGEE and GEE. CWGEE indicated that the school program on cannabis had an odds ratio (OR) and 95 % confidence interval (CI) of 0.83 (0.73, 0.95) on all substance use, but GEE produced a null association with an OR (95 % CI) of 0.92 (0.79, 1.07).

Conclusions: When ICS is present in clustered school data, weighted and unweighted analyses may produce different results. Care is needed to investigate the relationship between cluster size and the outcome, and use appropriate methods for analysis. Certain substance programs may influence student behaviour in other substances, highlighting the need for a multivariate analytical approach when studying the use of substances by adolescents.

When data on multiple indicators of underlying psychosocial constructs are collected, they are often intended as closely related assessments of a relatively unified phenomenon, or alternatively as capturing distinct facets of the phenomenon. Establishing distinctions among construct phenomena, assessments, or indicators is sometimes described as establishing discriminant validity. In the philosophical literature, often extreme instances or limit cases, actual or hypothetical, are used to identify settings in which one phenomenon is present and the other is not, to establish distinctions. We put forward an empirical analogue of this philosophical principle applied to distinctions amongst survey item responses. The quantiles of extreme differences matrix characterizes, for each pair of indicators, how large differences are between indicators at relatively extreme quantiles of the distribution of those differences. We discuss potential uses and properties of this matrix and related matrices for identifying relevant distinctions among indicators or facets of underlying construct phenomena.

Objectives: EpiEstim is a popular statistical framework designed to produce real-time estimates of the time-varying reproductive number, ${ℛ}_t$ . However, the methods in EpiEstim have not been tested in small, non-randomly mixing populations to determine if the resulting ${\widehat{ℛ}}_t$ estimates are temporally biased. Thus, we evaluate the temporal performance of EpiEstim ${\widehat{ℛ}}_t$ estimates when population structure is present, and then demonstrate how to recover temporal accuracy using an approximation with EpiEstim.

Methods: Following a real-world example of a COVID-19 outbreak in a small university town, we generate simulated case report data from a two-population mechanistic model with an explicit generation interval distribution and expression to compute true ${ℛ}_t$ . To quantify the temporal bias, we compare the time points when true ${ℛ}_t$ and estimated ${\widehat{ℛ}}_t$ from EpiEstim fall below the critical threshold of 1.

Results: When population structure is present but not accounted for ${\widehat{ℛ}}_t$ estimates from EpiEstim prematurely fall below 1. When incidence data is aggregated over weeks the estimates from EpiEstim fall below the critical threshold at a later time point than estimates from daily data, however, population structure does not further affect timing differences between aggregated and daily data. Last, we show it is possible to recover the correct timing when by using the lagging subpopulation outbreak to estimate ${\widehat{ℛ}}_t$ for the total population with EpiEstim.

Conclusions: ${ℛ}_t$ is a key parameter used for epidemic response. Since population structure can bias ${ℛ}_t$ near the critical threshold of 1, EpiEstim should be prudently applied to incidence data from structured populations.

Objectives: The addition of two-way interactions is a classic problem in statistics, and comes with the challenge of quadratically increasing dimension. We aim to a) devise an estimation method that can handle this challenge and b) to aid interpretation of the resulting model by developing computational tools for quantifying variable importance.

Methods: Existing strategies typically overcome the dimensionality problem by only allowing interactions between relevant main effects. Building on this philosophy, and aiming for settings with moderate n to p ratio, we develop a local shrinkage model that links the shrinkage of interaction effects to the shrinkage of their corresponding main effects. In addition, we derive a new analytical formula for the Shapley value, which allows rapid assessment of individual-specific variable importance scores and their uncertainties.

Results: We empirically demonstrate that our approach provides accurate estimates of the model parameters and very competitive predictive accuracy. In our Bayesian framework, estimation inherently comes with inference, which facilitates variable selection. Comparisons with key competitors are provided. Large-scale cohort data are used to provide realistic illustrations and evaluations. The implementation of our method in RStan is relatively straightforward and flexible, allowing for adaptation to specific needs.

Conclusions: Our method is an attractive alternative for existing strategies to handle interactions in epidemiological and/or clinical studies, as its linked local shrinkage can improve parameter accuracy, prediction and variable selection. Moreover, it provides appropriate inference and interpretation, and may compete well with less interpretable machine learners in terms of prediction.