Understanding disease transmission in the workplace is essential for protecting workers. To model disease outbreaks, the small populations in many workplaces require that stochastic effects are considered, which results in higher uncertainty. The aim of this study was to quantify and interpret the uncertainty inherent in such circumstances. We assessed how uncertainty of an outbreak in workplaces depends on i) the infection dynamics in the community, ii) the workforce size, iii) spatial structure in the workplace, iv) heterogeneity in susceptibility of workers, and v) heterogeneity in infectiousness of workers. To address these questions, we developed a multiscale model: A deterministic model to predict community transmission, and a stochastic model to predict workplace transmission. We extended this basic workplace model to allow for spatial structure, and heterogeneity in susceptibility and infectiousness in workers. We found a non-monotonic relationship between the workplace transmission rate and the coefficient of variation (CV), which we use as a measure of uncertainty. Increasing community transmission, workforce size and heterogeneity in susceptibility decreased the CV. Conversely, increasing the level of spatial structure and heterogeneity in infectiousness increased the CV. However, when the model predicts bimodal distributions, for example when community transmission is low and workplace transmission is high, the CV fails to capture this uncertainty. Overall, our work informs modellers and policy makers on how model complexity impacts outbreak uncertainty. In particular: workforce size, community and workplace transmission, spatial structure and individual heterogeneity contribute in a specific and individual manner to the predicted workplace outbreak size distribution.

了解工作场所的疾病传播对于保护工人至关重要。为了模拟疾病爆发，许多工作场所的人口较少，需要考虑随机效应，这会导致更高的不确定性。本研究的目的是量化和解释这种情况固有的不确定性。我们评估了工作场所疫情爆发的不确定性如何取决于 i) 社区的感染动态，ii) 劳动力规模，iii) 工作场所的空间结构，iv) 工人易感性的异质性，以及 v) 工人传染性的异质性。为了解决这些问题，我们开发了一个多尺度模型：预测社区传播的确定性模型和预测工作场所传播的随机模型。我们扩展了这个基本的工作场所模型，以允许空间结构以及工人的易感性和传染性的异质性。我们发现工作场所传播率和变异系数 (CV) 之间存在非单调关系，我们将其用作不确定性的度量。社区传播、劳动力规模和易感性异质性的增加降低了CV。相反，提高空间结构水平和传染性异质性会增加 CV。然而，当模型预测双峰分布时，例如当社区传播率较低而工作场所传播率较高时，CV 无法捕捉到这种不确定性。总的来说，我们的工作让建模者和政策制定者了解模型复杂性如何影响疫情的不确定性。特别是：劳动力规模、社区和工作场所传播、空间结构和个体异质性以特定和个体的方式影响预测的工作场所疫情规模分布。