Oncology remains one of the most challenging areas for payers, caught in a perfect storm of costly medications, high‑expense specialized therapies, shifting reimbursement rules, and a rising incidence of cancer. The clash between rapid innovation and financial sustainability has reached a critical juncture, with cancer‑care spending increasing for four consecutive years.
Cancer’s financial impact continues to rank among the top five drivers of employer health‑care costs because it affects the working‑age population. More than 40 % of new diagnoses occur in people aged 20‑64, the group covered by employer‑sponsored plans. A single high‑cost claim can quickly exceed stop‑loss limits, leaving employers with little time to plan or explore lower‑cost alternatives.
Oncology care follows a non‑linear trajectory, moving through stages and treatments that can change suddenly and drive cost spikes. Although the clinical progression is understandable, the system usually detects shifts only after they appear in claims, which hampers proactive management.
The problem is not a lack of data. Instead, oncology relies on fragmented medical and pharmacy claims rather than a unified, real‑time information source. Consequently, high‑cost patients are often identified only after treatment has begun and financial consequences are already unfolding, depriving employers and plan sponsors of the chance to anticipate or mitigate risk.
The challenge: How and when data is used
Clinical systems capture multiple “signals”—laboratory results, imaging, treatment response, and physician intent—while claims systems record service details and associated costs. Although both reflect the patient’s journey, they operate independently, making it difficult to create a single, coherent view of what is happening.
Financial analysis typically leans on claims‑based data, which is standardized but retrospective. A costly claim may surface weeks or months after the clinical event because claims are submitted after care is rendered and adjudicated, creating a lag between disease progression and financial recognition.
Why the system often feels reactive: Understanding cost through patient trajectory
Interpreting clinical data is more complex than analyzing claims, which explains why clinical information is often excluded from evaluation processes. As a result, decisions are based on what has already occurred rather than what may happen next, giving the impression that oncology costs are erratic.
A small number of patients can account for a disproportionate share of total spend. High‑price cases typically unfold over several disease stages, each step being clinically justified but cumulatively costly. The financial impact becomes clear only after the full sequence is complete.
Focusing on a patient’s treatment trajectory—considering how one decision influences the next—offers a clearer picture than evaluating isolated claims. This perspective helps stakeholders balance clinical outcomes with financial exposure.
Potential future costs may be signaled by changes in disease status, lines of therapy, or treatment response, even before they appear in claims.
What earlier visibility actually changes
Better insight into these transitions does not require perfect prediction; a directional signal that a patient’s condition is shifting can be enough to change stakeholder behavior. For employers and plan sponsors, earlier visibility provides time to plan financially, engage in care management, and understand the range of possible outcomes.
Managing oncology spend is about reducing both cost and uncertainty. Unheralded high‑cost cases force organizations into a reactive posture. Early warning enables a measured, informed approach, improving financial planning and operational response.
From explaining the past to anticipating what’s next
Health‑care already possesses the data needed to understand oncology costs. What is missing is a consistent method to connect clinical progression with financial impact on a timeline that matches strategic planning. Claims will continue to record what has happened, while clinical data will continue to signal what may happen next.
The challenge is to bring these perspectives together in a way that mirrors real‑world disease and treatment progression. Until that integration occurs, oncology will continue to feel unpredictable—not because the patterns are unknown, but because they are recognized too late to be actionable.
Photo: Seksan Mongkhonkhamsao, Getty Images
Arnav Saxena is a Machine Learning Engineer with a knack for problems where the data is messy, the requirements are fuzzy and the path forward isn’t obvious. He holds a Bachelor of Technology (B.Tech) in Applied Mathematics from Delhi Technological University, India and a Master of Science (M.S.) in Data Science from Columbia University, NY. Currently a Machine Learning (ML) Engineer at Evidium, his career spans five years across consulting and startups – two years at Bain & Company as an Analyst/Associate, followed by three years in data science roles at two early‑stage companies. Arnav works at the intersection of mathematical foundations and practical application, building frameworks that turn uncertain situations into useful insights and helping teams make better decisions with incomplete information.
