This summer, host cities across the country will take on something rare: a dense season of major events—the World Cup, the nation’s 250th, the July 4th corridor—arriving close together and drawing enormous crowds. Most of the planning attention, rightly, goes to the events themselves. But moments like these also do something quieter and more useful. They put the systems underneath them through their paces, and they reveal which ones were built for that and which ones were not. Most of what we call supply chain systems do not hold up in the conditions of a real surge.
For years, we’ve operated under a convenient mental model that splits the health system into two distinct worlds. On one side are clinical systems: the ones that admit, diagnose, treat, and discharge. On the other are back-office systems: the machinery that keeps the institution running—materials management, distribution, finance, procurement, and contracting. It’s a clean division; one side delivers care, the other runs the business. But that distinction was always more administrative than real, and the next demand shock makes it impossible to ignore.
By clinical system, I don’t mean only the EHR or the bedside documentation tool. I mean any system whose failure can delay a procedure, force an unsafe substitution, interrupt therapy, or consume the clinical labor that should be spent on patients.
When the Network Breaks Down
Because a surge hits every campus at once and coordinates across an entire region, systems are forced to function as one care machine. Decisions depend on data that has to be current, trusted, and immediately usable. There is no pause between treating and resupplying, no buffer where lagging systems can catch up. And that’s where the cracks begin to show—because for all the progress IDNs have made in seeing the patient, they still struggle to see themselves in real time.
Take a critical consumable—blood-warming sets, red rubber catheters, a backordered closure device. Data on where that item sits, how much is on hand, what condition the par is in, and whether it’s been consumed is spread across an ERP, a distributor portal, Epic, and a half-dozen unit-level closets that don’t reconcile. At any given moment, the network does not actually know how many of those items it has or where they are. The moment a real backorder bites, materials managers stop trusting the system of record. They revert to phone calls, spreadsheets, and walking the floor.
A recall raises the stakes on the same blindness. The notice arrives with lot numbers, and the question becomes immediate: which closets, carts, and case picks hold affected product, and has any of it already reached a patient? The system of record cannot say. It takes days of chasing invoices and checking shelves to settle a question patient safety needs answered in hours.
Over the past decade, health systems have integrated the clinical record to make a physician at the bedside dramatically more effective—orders, results, decision support, all in one pane. Ambient listening now captures clinical documentation automatically. AI vision models read X-rays and augment radiologists in real time. Technology has been pushed all the way through to the point of care. During that same time, the supply chain systems that support those same clinicians have barely evolved. We have done almost nothing comparable on the supply side. In many ways those systems have become less integrated, intermediated by data lakes and BI tools that separate the source systems from the point of care.
Inside a surge, the back-office systems fail to provide the very data they were designed to track: how many ventilators are functional, where the float staff are, whether the central sterile pipeline can keep pace, and whether the substitutes on hand are actually clinically equivalent. The lesson is that the systems supporting materials, labor, and capital equipment cannot live on a back-office island. They are a subset of clinical systems, inextricably linked to the network’s core mission of delivering care.
Know the Demand and Know Yourself
We often talk about supply disruption as something imposed from outside—a manufacturer recall, an allocation, a hurricane closing a distribution center. The harder truth is that a meaningful portion of that fog is self-inflicted.
We have spent a decade investing in the first half of the equation. Our ability to sense external demand and external risk—census forecasting, recall feeds, market intelligence—has improved markedly. But our ability to understand our own posture—true on-hand inventory, real substitution depth, actual burn rate by location—lags far behind.
You see it most clearly at the moment of decision. A value analysis lead is approving a conversion across eight facilities. The clinical evidence is current. The contract pricing is live. The demand forecast is continuously updated. That entire side of the equation has been engineered for speed. But when attention turns inward, the picture degrades. On-hand counts trail reality. Backorder status reflects yesterday’s portal pull. Par levels are fragmented across facilities. Everything about the market is immediate; everything about ourselves is historical.
That discrepancy is a direct result of how the systems were designed. Conway’s Law has shaped the IDN more than any formal architecture. Systems mirror the way hospitals are organized, so we ended up with different replenishment models, different clinical equipment choices, and different supply strategies, hospital to hospital, unit to unit, each optimized for its own workflow and vendor ecosystem, each evolving on its own timeline. That made sense when every unit managed its own supply closet. It does not hold in a world where demand is distributed across a network, lead times are compressed, supply chains stretch across continents, and outcomes depend on how well every part of the enterprise works together.
And the items themselves are never just items. A blood pressure cuff is not a line on a purchase order. It comes with compatible monitors, the right adapters, the clinical workflows built around it, the procedures for changeover, and the way it feeds into charting. Swapping one cuff for another means rethinking the entire ecosystem around it: corequisite parts, training, integration. There is far more to changing out an item than changing a part number. That complexity is exactly what gets erased when supply chain is treated as nothing more than moving boxes.
The fragmentation is not only organizational; it is technical. Pulling data out of Epic, in and out of Workday, through the ERP and the financial systems means crossing disparate platforms, each with its own schema, its own update cadence, its own authentication. Bridging any two of them means navigating firewalls, facility-specific permissions, and separate IS&T, clinical, supply chain, and regulatory teams before anything moves. What materials managers are left with is another login and another tool bolted onto the side: manual exports, manual imports, reconciliation spreadsheets, rather than a single workflow of execution. We built systems that mirror our fragmentation, then bolted data bridges on top, when we should have designed for integrated execution from the start.
The need for speed is no longer confined to clinical systems. The IDN largely has the data it needs, but it lacks the ability to act on it. Many of the systems underpinning supply resilience still operate on timelines measured in days or weeks. Systems that require manual intervention at every step are fundamentally misaligned with the speed of operations. Compensating with dashboards and scorecards has not resolved the mismatch—if anything, it has made it more tolerable by providing acceptable visibility during normal census, letting programs claim progress that’s good enough. Storing and displaying information is too low a bar. Critical systems must participate in operations, translating data into supply decisions.
This leads to an obvious conclusion: if a system does not function during a surge, it does not function for the health system. Compliance, reporting, and even cost savings are poor proxies for the true objective: keep patients safe and care uninterrupted. Systems that cannot support that objective under realistic conditions are liabilities.
The Ecosystem Imperative
The separation between hospital, distributor, and manufacturer compounds the problem. Hospitals increasingly want their key distributors to act as an extension of their own storeroom: captive inventory, always available. But distributors run on thin distribution margins, and buffer stock is carrying cost they cannot recover, so they hold less even as hospitals lean on them more. Each node operates with partial information, optimizing locally while the system fails globally. A manufacturer knows a product is constrained three months out, and the first the hospital hears of it is the allocation letter. A distributor manages allocation across hundreds of hospitals but treats that data as competitive advantage rather than collective intelligence. A hospital has urgent demand but no visibility into what is actually coming or when.
During normal operations, this fragmentation is tolerable. It is baked into contract terms and procurement cycles measured in weeks. But the moment real demand hits, the entire ecosystem reverts to phone calls and manual negotiation. Distributors field calls from dozens of hospitals simultaneously, all trying to pull forward shipments or find substitutes. Manufacturers triage requests based on relationships rather than clinical need. Nobody has a current picture of what’s actually available, where it is, or what’s coming.
The money mechanics make it worse. Pricing lives in tiers, volume commitments, and rebates earned on contract compliance. When a backorder forces a substitution, the hospital often buys off contract at penalty pricing, forfeits the rebate, and generates a pile of invoice exceptions for someone to reconcile by hand. The financial machinery punishes the exact flexibility that keeps care running.
The deeper issue: manufacturers and distributors have the data. Production schedules, allocation decisions, pipeline inventory, demand forecasts. All of it exists. But it’s locked behind vendor portals, shared only when contractually required, and structurally separated from the hospitals’ operational view. That has to change. The data is not the intellectual property. The service a distributor provides is the value: the allocation expertise, the risk management, the decision support built on top of that data. Holding the data back as if it were a competitive moat protects the wrong thing. Real collaboration means transparency that runs the full path, from the point of care, through the hospital supply chain, through the distributor and its DCs, all the way back to the manufacturer, with the teams along that path actually working together and sharing the signal.
What this requires is genuine system and data interoperability across manufacturer, distribution, and hospital supply chains, reaching into the clinical environment itself. Not a single proprietary system, but standards-based interoperability where data moves in real time. And substitution has to be treated as the multi-dimensional problem it actually is. You cannot swap a part number without knowing the availability of the alternative and the corequisite parts that go with it; there is no point substituting a cuff when the compatible monitors are backordered elsewhere. It is complicated, everyone in the field knows it is complicated, and the systems have to support that dynamic change rather than pretend it away.
That means visibility anchored at the point of care. Not merely whether the hospital or the distributor has material somewhere, but whether the right material is in the right place at the right time, every time, for the next case. It means understanding the clinical risk at any point of care when an item is not available. We need an item 360, but we also need a point-of-care 360, and the full path traced backward from that point of care through the distributor and DC to the manufacturer. The stress test is not a normal surge. It is a cyber attack that blinds a distributor, a mass-casualty event, or a sudden population influx like a major sporting event, with demand climbing minute by minute. Day-old information does not cut it under those conditions.
Care and Continuity
The separation between back-office systems and clinical systems is an artifact of the past and now a critical risk in the present. This requires something more fundamental than modernization in the traditional sense. It requires a shift in how these systems are conceived, built, and evaluated.
They must be tied directly to patient outcomes, fully integrated into clinical platforms, and exercised under conditions that resemble a real surge. They must operate at the speed of care, not the speed of the monthly contract cycle. Like clinical systems, they must absorb AI to accelerate and improve decisions. This is not work that spreadsheets and reporting layers can carry. It calls for a different class of capability: platforms that fuse data across disparate sources, model the supply network, and support decisions at operational speed. Whether that capability is bought, partnered for, or built in-house matters less than the requirement itself. And they must provide a continuous, accurate picture of our own posture, because without that, every other advantage begins to erode.
Build that foundation first, and the cost conversation finally becomes a rational one. Today hospitals optimize for lowest unit cost while blind to their own consumption, waste, and risk, and that blindness is precisely what manufactures fragility and drives total cost up when disruption hits. Years of unit-price pressure have already thinned some essential commodities down to one or two makers, so a single plant failure becomes a national shortage. Systems that actually see the network let us find the waste and redundancy these poor systems are themselves creating, and eliminate it. The goal is not to spend more. It is to build the operational intelligence that makes intelligent cost management possible at all.