Pharmacy leaders typically commit to budgets months before they know what prices will do. Many purchasing platforms, contract leakage reports, and acquisition cost dashboards are built to answer one question: what happened? That backward-looking analysis is valuable for operational accountability, but doesn’t give you enough information to plan effectively.
With drug spend representing roughly 10% of total hospital operating expenses and nonfederal hospital pharmaceutical expenditures reaching $39B in 2024, a pricing environment that shifts faster than budget cycles demands something more than retroactive analysis.
63% of pharmacy leaders carry specific savings targets. Hitting those targets requires knowing what you’re projecting against months in advance, not just reconciling what you already spent.
Why Spend Analysis Alone Doesn’t Support Pharmacy Budget Planning
Hospitals face 10,000 weekly price updates across their formularies. By the time retroactive analysis catches a pricing shift, the margin is gone. The question spend analysis can’t answer is the one that matters most for pharmacy budget planning: given what we know today about prices, contracts, and demand, what can we expect to spend next quarter?
The Three Inputs a System-Level Pharmacy Purchasing Forecast Depends On
A reliable pharmacy purchasing forecast draws from three distinct data categories. Most teams have partial access to all three. The ones that produce forecasts that hold up under pressure use all three together.
Historical Purchasing Data at the NDC Level, Across All Facilities
Aggregate spend data hides the site-level variation that drives real cost differences across a health system. A system-level forecast needs granular purchase history by drug, by NDC, and by location. Rolled-up totals flatten meaningful differences between facilities and produce assumptions that are wrong for any individual site.
Forward Price Signals
GPO contract expiration dates, known WAC increase windows, upcoming patent expirations, and biosimilar launch timelines are all dateable months or years in advance. A forecast that doesn’t incorporate them is projecting against a static pricing environment that doesn’t exist.
Demand Variables by Site
Patient volume projections, formulary additions and removals, service line changes, and seasonal patterns all shift what a system will purchase. Volume shifts at one facility change GPO tier accumulation rates across the whole system, which means a system-wide average demand assumption can make a forecast structurally wrong before it’s even finalized.
Three demand signals in particular require active tracking rather than historical extrapolation:
- Service line volume by category, not total admissions. A system-wide patient volume projection tells you little about which drug categories face pressure. A new oncology program or expanded infusion center changes high-cost specialty demand at that site in ways that won’t appear anywhere in historical purchasing data. Those assumptions have to be built from scratch.
- Formulary committee decision timelines. A biosimilar addition approved in Q3 affects Q4 purchasing, not Q3. Mapping formulary decisions to their purchasing impact date is what separates a demand signal from a demand surprise.
- Service line contractions. Demand suppressions are as important as demand additions. If GLP-1 adoption is reducing bariatric surgery volume at a facility, the forecast for related surgical medications at that site needs a separate downward assumption, not a carryforward from prior-year spend.
Why Demand Data Is Harder to Use Than It Looks
Drug demand is highly volatile at the drug-by-dispensary level but more stable in aggregate. Forecasting at the wrong level of granularity produces projections that don’t survive contact with actual purchasing behavior.
Two patterns in particular create forecasting risk:
- Formulary transitions. A biosimilar conversion doesn’t just change price. It shifts volume distribution across NDCs and sites. A forecast built on pre-conversion demand assumptions will be wrong from day one.
- Therapeutic cascades. GLP-1 adoption illustrates the risk: formulary additions in one area suppress demand in adjacent categories (bariatric surgery volume, related injectables). Those ripple effects need to be modeled separately, not absorbed into a system-wide average.
With 88% of hospitals reporting pharmacy technician shortages, manual demand tracking across multiple facilities at the NDC level isn’t operationally viable. Teams that rely on it can’t build a forecast that stays current.
A forecast that lags real purchasing behavior by even a few weeks loses its value for pharmacy budget planning.
Contract Terms Are a Forecasting Variable, Not a Fixed Input
Most pharmacy teams treat contract terms as a background condition to manage when a contract comes up for renewal. A well-built forecast treats them as an active input, because contract events have predictable timing and material cost consequences. Drug price volatility doesn’t arrive only through market forces. Much of it arrives through contract events that were dateable all along.
The largest hospital GPOs award mostly 3-year contracts. That means an expiration date is almost always known years in advance. When a GPO contract lapses without a successor agreement in place, effective pricing reverts to WAC, a dateable, foreseeable cost jump that belongs in the forecast as a risk event, not a surprise discovered at renewal.
Mid-contract changes present a less obvious but equally important forecasting challenge:
| Contract event | Frequency | Budget impact |
| 340B ceiling price recalculation | Quarterly | Changes effective cost of eligible drugs each quarter |
| GPO tier threshold resets | Per contract terms | Missing a tier forfeits rebates and volume-based pricing |
| Manufacturer price adjustments | Ongoing | Alters WAC-based pricing independent of contract expiration |
Specialty drugs, oncology treatments, and cell and gene therapies now drive the majority of pharmacy budget variability. Contract term tracking effort should concentrate on those categories. A single biosimilar conversion or patent expiration in a high-spend therapeutic class can shift a system’s annual forecast by millions.
Teams that embed GPO expiration calendars and patent timelines into their forecast can plan renegotiation windows proactively. Teams that don’t embed those timelines find out at renewal.
How Purchasing Data Becomes a Forward Projection
The mechanic that turns purchasing data into a forward projection starts with segmentation. Total pharmacy spend is not a useful unit of projection. Different drug categories have fundamentally different price behaviors, and a single inflation assumption applied across the entire formulary will be wrong for most of it.
A workable segmentation looks something like this:
| Drug category | Price behavior | Projection approach |
| Stable generics | Predictable, narrow variance | Historical trend with limited upside |
| Volatile generics | Exposed to supply chain disruption | Wider price-range assumptions; flag offshore API exposure |
| High-cost specialty | Driven by pipeline and contract events | Model against known biosimilar entry dates and GPO terms |
| Shortage-prone drugs | Subject to spot-market premiums | Build in shortage-risk buffer; providers report 6–10% annual budget increases from shortage-driven costs |
Once categories are segmented, the projection layers in two additional inputs.
First, demand volume projections applied by site, not system-wide, are multiplied against the price projections for each category. Second, site-level results roll up to the system level. The site-level step matters because system averages hide the facilities doing the most damage to a forecast.
A single oncology-heavy site running significantly higher specialty volume than the system average can push the ceiling scenario from a planning assumption to a near-certainty if its demand inputs aren’t modeled separately. This layering produces a spend range, not a point estimate, which is what makes the output usable for pharmacy budget planning under volatile conditions.
The output should be a rolling 3–6 month forward view by drug category, updated at least monthly. Specialty and oncology categories require more frequent review as pipeline events and contract milestones approach. Hospitals face daily price changes across thousands of NDCs a forecast updated quarterly is already outdated when it’s published.
Bluesight’s purchasing optimization tools are built to surface those changes in real time, giving pharmacy leaders the data they need to keep projections current.
What Makes a Pharmacy Purchasing Forecast Hold Up
A forecast that holds up under drug price volatility isn’t just one that’s accurate on average. It’s one that survives a January WAC increase, a mid-year shortage event, or a GPO contract lapse without requiring a full rebuild. Three practices determine whether it does.
Build Scenario Ranges, Not Point Estimates
A single projected price per drug has no buffer when reality moves. The structural fix is scenario ranges rather than point estimates.
A three-scenario framework gives finance something it can actually plan against:
- Floor scenario. GPO pricing holds, no shortage events, biosimilar uptake proceeds on schedule.
- Base scenario. Moderate drug price inflation, one or two isolated shortage events, minor formulary adjustments.
- Ceiling scenario. WAC-exposed categories reprice mid-year, a GPO contract lapses, a shortage in a high-spend category forces spot purchasing.
Finance can plan variance against a range. A point estimate that turns out to be wrong gives them nothing to work with.
Stress-Test Before It Goes to Finance
Before the forecast reaches the finance team, test it against known risk events: contracts expiring within the forecast window, drugs on the active shortage list, and single-source generics with supply chain exposure. These aren’t hypothetical; they’re identifiable from existing data, and surfacing them before submission is the difference between a forecast that prompts a conversation and one that causes a revision.
Close the Validation Loop Each Quarter
After each quarter, compare forecasted spend to actual spend by drug category. Where the model over-projected, tighten the assumption. Where it under-projected, identify whether the miss was a price event, a demand shift, or a contract change; adjust accordingly. A forecast that is never validated against outcomes doesn’t improve.
Account for the Full System, Not Just the Average
For multi-facility health systems, resilience requires one additional layer: site-level demand inputs and contract compliance tracking at each facility. A system-wide average projection masks site-level GPO tier failures until they surface in a budget variance report. By then, the quarter is already over.
Vizient projects 3.84% pharmacy price inflation through mid-2026. External benchmarks like that serve as a useful sanity check on internal projections, particularly for categories where internal price-signal data is thin. If a category’s internal projection diverges significantly from a credible external benchmark, that gap is worth understanding before the forecast is finalized.
Pharmacy leaders building a forward-looking purchasing plan need more than a model. They need contract and pricing visibility, real-time data, and the ability to detect deviations from expected spend before they compound. That’s where current-state analysis and forward-looking forecasting work together. One keeps the model honest; the other keeps the budget protected.
Bluesight’s purchasing optimization platform gives pharmacy and supply chain leaders the data infrastructure to do both. See how it works.