The South Australian Branch of the Statistical Society would like to invite you to an in-person and online presentation by A/Prof Murthy Mittinty.
Date: 15th October (Wednesday) 2025
Time: 6:00 – 7:00 pm. Pre-meeting refreshments and networking from 5:30 pm.
Dinner: A dinner will be held after the meeting at Location Café Michael 2, 204 Rundle Street Adelaide.
Please RSVP to andrew.vincent@adelaide.edu.au for dinner by Monday 13th October, as we are usually unable to change the booking numbers at the last minute.
Venue: In-person and online:
Speaker: A/Prof Murthy Mittinty, Discipline of Biostatistics, College of Medicine and Public Health, Flinders University
Abstract
Randomised Controlled Trials (RCTs) are universally upheld as the gold standard for causal inference, while observational studies are routinely relegated to the realm of association. Yet, the rationale for this dichotomy is rarely interrogated. The central distinction between RCTs and observational studies, randomisation, is presumed to address both measured and unmeasured confounding by design. However, this belief often overshadows the reality that observational studies, when carefully structured and analysed, can yield causal insights, particularly when trials are infeasible, unethical, or prohibitively expensive.
The legacy of William G. Cochran’s seminal work on designing observational studies has been largely sidelined, despite its enduring relevance. Over the past three decades, methodological advancements, such as quantitative bias analysis, robust missing data techniques, enhanced computational capacities, and deeper understanding of confounding, selection bias, and measurement error, have significantly closed the gap between trial and non-trial inference. Simultaneously, the trial enterprise itself has evolved, with structured frameworks such as SPIRIT and CONSORT reinforcing principles of protocol clarity, documentation, and reproducibility.
Emulating a trial from observational data requires adopting these same principles: explicit protocol development, temporal alignment of exposure and eligibility, and principled handling of confounding and bias. This approach transforms observational research from exploratory analysis to structured causal inquiry. In this presentation, I advocate for a disciplined paradigm of “trial emulation” and illustrate, through practical examples, how such frameworks can yield results that closely mirror RCTs. This reconceptualization is essential for evidence generation in contexts where trials are neither practical nor ethical, yet decisions must be grounded in causality.
Biography
Associate Professor Mittinty is an accomplished academic and researcher in biostatistics, causal inference, missing data, and longitudinal data analysis. He serves as a Discipline Lead and developed the first Master’s in Clinical Epidemiology program at Flinders University. His work focuses on advancing statistical methodologies to strengthen healthcare decision-making and evidence-based policy. He has published widely, supervised research students across disciplines, and contributed to international collaborations. As an educator, he promotes critical thinking and data literacy, integrating innovation, integrity, and impact in health research.
@StatSocAus
