How Real-World Data and Advanced Models Can Solve Oncology’s Growing Drug Wastage Problem

Drug wastage is a significant and often under-recognized challenge in oncology, particularly in injectable or infusion-based treatments, where fixed vial sizes and patient-specific dosing requirements rarely match. Once opened, many of these high-cost drugs cannot be reused because of stability concerns and strict medical, legal, and regulatory constraints. Moreover, combination therapies add to complexity, and resizing vials remains a challenge due to high development costs and uncertain market returns.

Now, with the Centers for Medicare & Medicaid Services (CMS) penalizing for discarded drug units, wastage has become a visible financial burden. But by leveraging a data-driven approach, pharma manufacturers can avoid penalties. Here’s how.

Drug wastage: A quantifiable burden on the healthcare system

Wastage in oncology injectables is not a new problem but a growing one. Since 2017, CMS requires providers to report discarded quantities of single-use vial drugs. In a further step, beginning in 2023, CMS mandated that manufacturers must refund Medicare for discarded amounts exceeding a 10% threshold.

This is particularly relevant because, on average, the top 25 high wastage drugs consistently show discard rates above 10%. From 2017 to 2022, over $4.5 billion worth of drugs was discarded under Medicare Part B.

Top oncology drugs by discarded units %

In fact, drug wastage has been growing steadily. In 2022, the average wastage for the top 25 discarded drugs reached 12.28%, compared to 8.87% in 2017.

This represents not just a cost overrun but also a compliance risk, particularly as CMS penalties are now in effect. Even though commercial payers are not currently enforcing penalties, manufacturers are under pressure to reduce drug wastage.

Why proactive planning for CMS penalties matters for drug manufacturers

For drug manufacturers, especially those in the oncology, proactive financial planning is critical. A significant portion of their business often more than half is tied to Medicare, as many oncology patients are over the age of 65.

One major challenge here is the timing of CMS (Centers for Medicare & Medicaid Services) penalty notifications. CMS typically sends the final penalty bill at the end of the year, but this bill can arrive as late as the first or second quarter of the following year.

For instance, the bill for 2025 might not be issued until Q1 or Q2 of 2026. Despite this delay, manufacturers must plan for these costs well in advance. Ideally, they need to project and allocate potential penalty amounts during their 2024 planning cycle for 2025 and continue to refine these estimates on a quarterly basis throughout the year.

Without timely analysis and forecasting, drug manufacturers risk unexpected financial exposure.

Why measuring drug wastage is difficult in practice

While the need to measure and reduce wastage is clear, actually doing so is fraught with challenges. Accurately calculating wastage requires granular patient-level data: including weight, height, treatment regimen, dose administered, and vial size. However, most real-world data, whether open or closed, fall short in one or more of these areas.

Common challenges include:

Incomplete or inconsistent data

• Missing values such as dose administered, vial size, or drug strength may not always be recorded.
• Ambiguity in recorded units is very common if integration between systems is incomplete.

Incomplete or unavailability of height and weight

• Body surface area (BSA), which is commonly used to calculate drug dosage, is typically derived from a patient’s height and weight. In some cases, dosage can also be calculated based on weight alone.
• Either or both may not be captured in open claims data and are often incomplete even in closed claims data.

Measurement timing mismatch

The most recent height or weight may not reflect the values at the time of drug administration.

Gap in medication records

Incomplete administration history and gaps in data capture disrupt accurate Integrated patient insights.

Together, these challenges highlight why measuring drug wastage isn’t easy with the data currently available. It also underscores the need for a more structured approach; one that brings together an understanding of drugs administration, selection of the right data sources, and the use of advanced models.

A structured, data-drive approach to predict and reduce drug wastage

1. Understand drug administration cycles

2. Select and combine the right real-world data

Choose the right combination of RWD sources to ensure both breadth and depth of insights.

To overcome the limitations of each, use multi-source analysis and triangulation to generate reliable insights.

3. Map and validate patient journeys

Use the selected datasets to map patient journeys. Only include complete patient journeys to ensure accuracy. Validate these journeys against approved dosage cycles for the selected drug.

4. Calculate drug wastage

5. Forward looking: Use AI/ML models to bridge data gaps

Advanced models can be used to handle missing data and estimate missing vitals and dosage patterns. The process includes

Data collection

Use closed claims or EHR data to collect patient-level details such as demographics, diagnosis timing, comorbidities, and treatment stats.

Cleaning and preprocessing

Handle missing values, transform features, detect outliers, and split data into training and testing sets.

Model selection

Choose appropriate models like RandomForestRegressor, XGBoost, or MultiOutputRegressor based on the data structure.

Training and evaluation

Train models on prepared data and evaluate performance using MAE, RMSE, and R² metrics.

Identifying patient vitals

Leverage trained models to estimate missing patient vitals such as height and weight.

Monitoring and refinement

Continuously monitor model performance and retrain based on model drift or defined triggers.

Case insight: Alignment with CMS benchmarks

In a representative example involving a drug used in multiple myeloma (drug 1 as shown in the table), our approach consistently estimated wastage within 2–3% of CMS-reported values over a five-year span. Additionally, pattern remained consistent across data sources with the slight variation in the absolute value.

In our approach, BSA helped determine the dosage administered for these drugs, which come in standard single dosage vials.

This consistency confirms that the approach is not only directionally accurate but robust enough for use in real-time decision-making.

Business implications and strategic value of the case insight

For pharmaceutical manufacturers, the benefits of this approach go beyond regulatory compliance. The initial findings from the analytical approach showed strong alignment with CMS-reported drug wastage figures, validating the approach and providing a solid foundation to refine using predictive models.

Real-time wastage identification

Using real-world data, manufacturers can now identify drug wastage as it occurs, gain timely insights, and take corrective actions before wastage reaches reportable levels or results in penalties.

Packaging and distribution strategy

Insights from the study can inform packaging design such as the feasibility of smaller vial sizes or optimized distribution to reduce avoidable wastage and associated financial risks.

Precision in every dose, and every data point

Drug wastage in oncology is not simply a cost issue, it reflects inefficiencies across formulation, packaging, clinical practice, and data usage. As CMS and other global regulators tighten accountability, manufacturers must adapt with better tools, deeper data, and smarter models.

By leveraging a triangulated, AI-enabled approach, pharma companies can take control over reducing drug wastage. The payoff isn’t just in savings, but in better planning, smarter distribution, and more sustainable healthcare delivery.

References:

1. Centers for Medicare & Medicaid Services. Medicare Part B Discarded Drug Units [Internet]. Baltimore (MD): CMS; [cited 2025 Mar]. Available from:
https://data.cms.gov/summary-statistics-on-use-and-payments/medicare-medicaid-spending-by-drug/medicare-part-b-discarded-drug-units

2. Evaluation of Five Formulae for Estimating Body Surface Area of Patients Including Just a Child [Internet]. PubMed Central; [cited 2025 Jul 23]. Available from:
https://pmc.ncbi.nlm.nih.gov/articles/PMC4250987/

3. Fryar CD, Carroll MD, Gu Q, Afful J, Ogden CL. Anthropometric reference data for children and adults: United States, 2015–2018. Vital Health Stat 3(46) [Internet]. Hyattsville (MD): National Center for Health Statistics, CDC; January 2021 [cited 2025 Jul 23]. Available from:
https://www.cdc.gov/nchs/data/series/sr_03/sr03-046-508.pdf