Drug shortages persist because the supply chain is still invisible

At the end of 2025, the FDA said it had prevented 330 shortages, yet 93 shortages were still open at year-end and 1,424 potential shortage notifications had arrived from 167 manufacturers. ASHP said 223 active shortages were still open in Q2 2026, and 77 percent of them began in 2022 or later. Fifteen percent involved controlled substances, which is one reason shortages hit pain control, anesthesia, and behavioral health care so hard.

The claim is simple: drug shortages keep recurring because no one sees the whole supply chain, and the fix is a shared visibility architecture that joins manufacturing, distribution, hospital inventory, and clinical substitution decisions into one operating picture. Without that, every participant makes local decisions from partial data, and local efficiency keeps defeating system resilience.

Shortages survive because the signal arrives in fragments

The drug shortage problem has survived for years because its failure modes are distributed. A plant has a quality event, a batch is delayed, a raw material slips, a distributor sees fill rates fall, a hospital starts substituting, and the regulator hears about it through another channel. Each party knows something. None of them sees enough.

That fragmentation is not just a technology problem. It is also a process problem, a data-latency problem, and an incentive problem. The manufacturer may know capacity is constrained, but the distributor sees only orders and allocations, the hospital sees only its own shelves, and the group purchasing organization sees contract terms without live clinical consumption. By the time those signals align, the shortage is already in motion.

The FDA’s 2025 report underscores how much effort is spent on mitigation after the fact. The FDA said it used 78 instances of regulatory flexibility and discretion and still ended the year with ongoing shortages. That is active management, not market self-correction.

A shared operating picture is the missing control plane

The useful model is not another dashboard. It is a shared visibility architecture built as a control plane across parties that already generate the right data, just not in a common form. Manufacturer ERP, manufacturing execution system, quality management system, batch release status, and maintenance events need to sit beside distributor inventory, allocation, order fill rate, backorders, and shipment status, then be reconciled against hospital pharmacy inventory, purchasing, formulary rules, and medication use.

Once those streams are connected, the system can do work that current shortage lists cannot. It can flag a single-source injectable when fill rates begin to drift, surface a region where par levels are collapsing, and route a substitution review before the last unit is gone. It can also connect shortage notifications to the downstream sites most likely to feel the hit first.

A reference design would use an event stream for live shortage and inventory signals, a document store for notices and supporting evidence, a retrieval index for current shortage guidance and therapeutic alternatives, a rules engine for formulary and substitution policy, and an audit log for every alert, override, and approval. That is not a promise of perfect prediction. It is a way to make the relevant facts visible before the problem becomes a bedside surprise.

The value of aggregation is already visible in the institutions tracking the problem. The FDA drug shortages database and ASHP live shortage tracking both maintain current shortage views, but each still reflects only part of the operating picture. A shared architecture extends that reference data layer into a working network of decisions.

For readers who want the implementation angle, our AI Governance work shows how shared decision rules and auditability belong in the same design, not bolted on later.

What better visibility changes in daily operations

The point of the architecture is not to centralize all stock in one place. It is to make exceptions legible and fast. A pharmacy buyer should not be hunting through email for a shortage notice while a clinical leader is separately deciding whether to substitute one product for another. Those two decisions need to land in the same review queue with the same underlying facts.

That means several workflows become machine-assisted but still human-governed. A multi-party alert can go out when a manufacturer signals disruption, a distributor allocation changes, or a hospital’s consumption pattern moves outside expected bounds. The alert can then trigger inventory confirmation, therapeutic interchange review, and escalation to purchasing and clinical leadership. No one should have to retype the same shortage into three systems.

Prediction has a role, but a modest one. Models that weigh supplier concentration, historical interruption patterns, demand volatility, quality-event history, and lead-time changes can triage where limited attention goes first. They do not prevent sterile manufacturing failures, and they do not repair delayed remediation. They do, however, help scarce monitoring resources focus on the products that will hurt most when they fail.

That matters most for categories where the downstream cost is clinical, not just operational. ASHP said nearly half of new shortages involved products with a single manufacturer. When one source dominates, a small disruption can become a national problem quickly, especially in controlled substances and sterile injectables.

There are real limits, and they are not technical alone

Shared visibility does not solve economics. Low-margin sterile injectables still have weak incentives for redundant capacity, and a shared architecture cannot create a second plant where none exists. It also cannot make quality remediation faster when facility issues or sterile failures require months or years to fix.

There are governance limits too. Manufacturers will treat capacity and allocation data as sensitive, and hospitals will not hand over clinical data without clear access controls, audit logging, and role-based permissioning. Substitution rules must stay under pharmacy governance, with clinical review before they are automated. If a system cannot explain who changed what, and why, it should not be driving medication decisions.

There is also an equity problem. If allocation logic simply favors the largest systems or the loudest buyers, the architecture will reproduce the same shortages in a more orderly way. The design has to account for smaller hospitals, rural providers, and safety-net institutions when supply is rationed.

For organizations that want the operational boundary conditions before buying tools, our Healthcare page is the right next stop. The point is not software first. The point is which decisions must be shared, and which must stay local.

The decision is whether to keep managing scarcity blind

The persistent shortage numbers from the FDA and ASHP make one thing hard to deny: the current model is still too segmented to absorb routine disruption. If manufacturers, distributors, hospitals, and regulators remain stuck in separate queues, the supply chain will keep discovering scarcity only after it has already spread.

The practical choice is whether to build the shared visibility layer now, around the systems of record that already exist, or keep funding exception handling as if it were strategy. QueryNow works with teams that want the first path. Tell us the workflow you want gone, we build it in your environment in two weeks, and you pay $10,000 only after it meets the acceptance criteria you signed off on. Start there at /build.