Authors: Han Fu, David Hodgson, Emily Nightingale, James Azam

As a group involving ECRs working in various areas of infectious disease research, we found sex/gender is an important but under-addressed issue in our work. We decided to take the opportunity of our monthly meetings to discuss the issue together. During our discussion we looked at both sex and gender aspects ─ with sex referring to biological and physiological attributes, whilst gender refers to socially constructed characteristics.  In the one-hour hybrid session, we had fruitful reflections on the role that sex/gender plays in previous and current modelling research on infectious diseases. We identified gaps and challenges in data collection and model conceptualisation that hinder the incorporation of a sex/gender lens in infectious disease modelling.

Exploring social and behavioural aspects

Although information on sex/gender is routinely collected for most health-related research, it is not as commonly integrated into the modelling of infectious diseases in the same manner as other variables, such as age. Reflecting on our own experiences, we noted examples of research addressing sex-based (biological) differences, such as modelling the varying immune response between male and female populations to estimate the magnitude and distribution of disease burden. There is, however, little exploration of gender-based (social) differences in health behaviour and healthcare access, although this may also affect transmission dynamics, the impact of intervention measures, and disease outcome. The lack of engagement with gender-based factors is likely to stem from limitations in data availability, as well as a training gap for modellers that would enable them to consider disease dynamics from their social and behavioural aspects.

Challenges in incorporating a sex/gender lens

Incorporating a sex/gender lens into infectious disease modelling brings with it several challenges. As modellers we have to rely on appropriate data to provide inputs for models and to validate our assumptions. In many situations, disaggregated information by sex/gender and other characteristics that inform the risk level of disease is not available. Potential bias in representation could occur when contextual factors, such as household structure and labour migration schedule, are not fully considered in data collection. Another key consideration is whether there is a causal mechanism of sex/gender in infectious disease susceptibility and transmission. Furthermore, the interests of stakeholders and policy-makers could influence how this data is treated in mathematical models. For example, many countries planned their COVID-19 vaccination strategies based on age and job categories of target populations but not sex/gender. Without policy relevance, as researchers we tend to leave out the sex/gender heterogeneity in modelling to ensure models are kept as simple as possible. Lastly, software for simulating models often includes elements for flexibly specifying age profiles and probably sex, but not much flexibility is provided for specifying gender beyond the traditional binary categories. 

Next steps

Going forward we need to encourage modellers to consider sex/gender in their work. We must work to improve data collection approaches so that sex/gender variables can be easily and rigorously incorporated into analyses. It will also be important to engage decision-makers in incorporating a sex/gender lens into infectious disease policies and mitigation strategies. There is much to be learned from other disciplines, such as sociology, to sensitively explore the sex/gender role from social and cultural perspectives. Notably, the integration of sex/gender, especially in cases where these factors are not main study considerations, might come at the cost of model complexity, increasing the difficulties in conducting simulations and explaining outputs. Lastly, having software ready to deal with the complexity of sex/gender structures can reduce technical barriers and allow the role of sex/gender in infectious disease modelling research to be effectively addressed. Additionally, increased awareness and cultural/social acceptance of different gender identities has led to a greater range of non-binary responses being captured in health data. The development of modelling approaches within which this small but increasing group can be appropriately represented is an important challenge to be addressed going forward.

Through observation, reflection, and discussion in the session, we recognised that the consideration of sex/gender differences in infectious disease modelling is currently not sufficient and is often overlooked. Improvements in the study design framework, data collection, analytical methods and tools will tackle the challenges in integrating sex/gender into modelling research.

We would like to acknowledge active participation of attendees of the March CMMID ECR meeting and  support provided by the Women in CMMID group.