Stelios Tzellos | Why Physician Behaviour Is the Most Underrated Variable in Pharma Forecasting

Stelios Tzellos

Most pharmaceutical forecasts spend enormous time modelling markets and surprisingly little time modelling the people actually making prescribing decisions.

The result is an industry that often understands treatment algorithms better than it understands physicians.

Stelios Tzellos of the UK has spent years working in oncology forecasting, analytics, and commercial strategy, and one of the most consistent issues he has observed is the tendency to treat physician behaviour as predictable when it rarely is.

Doctors Don’t Prescribe Like Spreadsheets Assume

Commercial forecasts often assume that physicians respond rationally and immediately to new evidence. Better clinical data appears, treatment guidelines update, and prescribing shifts accordingly.

Real clinical practice is far more complicated.

Oncologists build habits over years of treating patients. They develop trust in certain regimens. They weigh side effect management, institutional protocols, patient preferences, and practical workflow considerations alongside efficacy data.

That means adoption curves are influenced by human behaviour as much as scientific evidence.

Experience Shapes Prescribing Faster Than Publications

At GlobalData, Tzellos worked on oncology competitive assessments where physician sentiment shifted rapidly after conference presentations or peer discussions, often before formal publications or guideline changes occurred.

Doctors frequently trust the experience of respected colleagues more than early commercial messaging.

This creates a challenge for pharmaceutical forecasting because behavioural momentum can accelerate or slow uptake independently of the clinical profile.

The Community Oncology Gap

Another issue is that many pharmaceutical forecasts over-index on academic medical centres.

Academic oncologists are often early adopters. Community oncologists move differently. They may face operational barriers, lower biomarker testing rates, or different patient populations.

At IQVIA, Stelios Tzellos worked on oncology disease insights projects where commercial performance depended heavily on understanding these practice-setting differences.

A launch that performs strongly in academic centres but struggles in community settings will produce very different revenue outcomes than the forecast originally projected.

Why Commercial Teams Miss This

Most pharmaceutical organizations are structured around data streams rather than behavioural understanding.

Forecasting teams model patient populations. Medical teams analyze trial data. Market research teams conduct surveys.

But physician behaviour sits between all three, and it rarely fits cleanly into a single function.

Building Better Forecasts

At AstraZeneca, where Tzellos works across analytics, business insights, and oncology marketing, better forecasting increasingly requires integrating behavioural assumptions directly into strategic planning.

That means understanding not just what physicians say they will do, but what they actually do under real clinical conditions.

The Bigger Lesson

Stelios Tzellos built a career around connecting scientific evidence with commercial reality. His academic work at Imperial College London, consulting experience at GlobalData and IQVIA, and industry work at AstraZeneca all reinforce the same principle: pharmaceutical markets are ultimately shaped by human decisions.

The companies that understand physician behaviour most accurately will outperform competitors that rely on cleaner but less realistic assumptions.