Oncology remains one of the most challenging areas for payers since it is literally in the eye of a perfect storm of expensive medications, high-cost specialized therapies, changing reimbursement rules that disrupt care and increasing cancer incidence across the US population with over 2 million new cancer cases projected in 2026 alone. The tension between rapidly evolving innovative treatments and the need for financial sustainability has reached a critical point, with cancer care costs continuing to increase for four consecutive years.
The financial burden of the disease continues to rank among the top five causes of employer healthcare spending as cancer affects the working age population. Over 40% of those newly diagnosed are between the ages of 20 to 64, the same population covered by employer sponsored health plans. The threat of even one high-cost cancer claim has the potential to exceed a health plan’s stop-loss limits quickly, leaving employers without the opportunity to plan financially or evaluate options for a less expensive viable path.
One key challenge is that cancer care doesn’t follow a clear, linear path. It moves through different stages and treatments that can shift quickly, causing massive cost escalation. While the trajectory follows understandable, clinical progression, the system can only identify the changes after the fact, once they appear in claims.
The problem is not a sheer lack of data. Conversely, in oncology, the system relies upon fragmented medical and pharmacy claims instead of a single, clean source of relevant information that is used appropriately in real-time. As a result, high-cost patients are often identified only after their treatment path is underway and the financial impacts are already in play. For employers and plan sponsors, that delayed visibility comes at a heavy financial price, leaving little opportunity to anticipate or manage the risk in advance.
The challenge: How and when data is used
Currently, clinical systems track numerous “signals” related to the care of a patient, including data from laboratory results, imaging, treatment response and physician intent. On the other hand, claims systems document the specifics of service use by a provider and the associated cost of providing the service. Although these two types of data represent the same patient experience, they function independently, which makes it impractical and difficult to form a single view of what is actually happening.
For example, in most cases, financial analyses hinge upon claims-based data that is standardized and widely available. Claims data are retrospectively focused, reflecting what has happened as opposed to anticipating what will occur in the future.
A costly cancer-related claim can surface weeks or months after the actual clinical event. This is because the claim is submitted after care is rendered and then adjudicated. By the time the financial impact is registered, the underlying disease progression and treatment decisions have long been underway. The result is an inevitable delay in the ability to anticipate or alter outcomes.
Why the system often feels reactive: Understanding cost through patient trajectory
Interpreting clinical data is far more complicated than analyzing claims data, which accounts for the difficulty in integrating clinical data into evaluation processes that rely heavily on claims data. As a result, most decisions rely upon information that reflects what has already happened rather than what may happen next.
To observers, cancer care expenses appear erratic or unpredictable, particularly when few patients account for a disproportionate share of total spend. Those high-priced cases usually occur over several stages in the natural history of the disease and are the result of a sequence of clinically appropriate decisions. A patient may initially respond to treatment, later show signs of progression and ultimately transition to a more advanced and expensive therapy. Each step in that process is justified from a clinical standpoint, but the cumulative financial impact often becomes apparent only after the sequence is complete.
Therefore, it is more helpful to think about where a patient is in the course of their treatment trajectory rather than looking at a series of isolated claims, evaluating how one decision might influence what comes next and calculating the cost. This insight gives greater understanding to stakeholders responsible for managing both clinical outcomes and financial exposure.
Potential future costs, even if not yet reflected in claims, may be indicated by changes in disease status, between lines of therapy or in treatment response.
What earlier visibility actually changes
Improving visibility into these transitions does not require perfect prediction. Often, a directional signal that a patient’s condition is changing can be enough to alter how stakeholders respond. For employers and plan sponsors, this allows additional time to plan financially, engage in care management or better understand the potential range of outcomes associated with a given case.
The distinction is important because managing oncology spend is as much about reducing cost as it is about reducing uncertainty. When high-cost cases emerge without warning, they limit the ability to act and force organizations into a reactive position. Earlier visibility allows for a more measured and informed approach, improving both financial planning and operational response.
From explaining the past to anticipating what’s next
Healthcare does not lack the data needed to better understand oncology costs. What it does lack is a consistent way to connect clinical progression with financial impact in a timeframe that aligns with strategic planning. Claims will continue to provide a reliable record of what has occurred, while clinical data will continue to offer signals about what may happen next.
The challenge lies in bringing these perspectives together in a way that reflects the real-world progression of disease and treatment. Until that happens, oncology will continue to feel more unpredictable than it actually is – not because the underlying patterns are unknown but because they are recognized too late to be useful, leaving stakeholders in a position where they are reacting to cost rather than managing it.
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.
This post appears through the MedCity Influencers program. Anyone can publish their perspective on business and innovation in healthcare on MedCity News through MedCity Influencers. Click here to find out how.
