Consider a woman we will call Ingrid. She is 76, lives outside Gothenburg, and carries a diagnosis list that would have been a death sentence a generation ago: heart failure, type 2 diabetes, chronic kidney disease, atrial fibrillation, early cognitive decline. Over the past twelve months she has seen a general practitioner, a cardiologist, a nephrologist, a diabetes nurse and an anticoagulation clinic. She has visited the emergency department twice and been admitted once. She takes fourteen medications prescribed by clinicians who have never spoken to each other. Her hospital care is run by the region; her home care is run by the municipality, under a different budget and a different law. The only person who has attended every single encounter is her daughter, who carries a ring binder.
Nothing in this story involves a failure of medicine. Every clinician was competent. Every guideline was followed. Her admission, though, was probably avoidable: one physician adjusted her diuretic, and no one checked her potassium, because checking it was nobody's job. The failure was not clinical. It was architectural.
Multiply Ingrid by tens of millions of patients across Europe and North America and you arrive at the defining, largely undiagnosed problem of modern health systems. For fifty years we have invested magnificently in what medicine can do. We have invested almost nothing in holding it together.
1. The paradox of productive medicine
By any measure of capability, healthcare's last half-century has been a triumph. Health spending has roughly doubled as a share of GDP since 1970, to around nine to ten per cent across the OECD and seventeen in the United States. The physician workforce has grown and specialised: American medicine alone now recognises well over a hundred specialties and subspecialties. Diagnostic imaging that did not exist in 1975 is now routine. More than twenty thousand medicines are approved for use. Genomic testing, biologics, minimally invasive surgery, hundreds of thousands of health apps: the production function of medicine has expanded on every axis.
Health systems celebrate each of these additions as progress, and individually, each one is. Here is the paradox: every new capability is also a new dependency. Every additional specialist, service, diagnostic and digital tool adds a node to the network a patient must traverse, and every node multiplies the connections between nodes.
The mathematics are unforgiving. In any system, the number of potential interfaces grows far faster than the number of participants: with n actors involved in a patient's care, the potential lines of communication approach n(n-1)/2. Four actors produce six interfaces. Twelve actors produce sixty-six. Capacity grows linearly; the coordination surface grows quadratically. Healthcare has spent five decades adding actors and congratulating itself on the nodes, while the interfaces, where the actual work of keeping a patient safe now happens, were left to fax machines, referral letters and daughters with ring binders.
Software engineering learned this lesson painfully. Brooks's law, formulated in 1975, holds that adding people to a late software project makes it later, because the communication overhead of each new person outweighs their added output. Healthcare has its own version: adding a ninth specialist to a complex patient can subtract value, because the marginal coordination cost exceeds the marginal clinical benefit. Yet no health system's investment logic acknowledges this. We still count what we add. We do not count what the addition costs everyone else.
Complexity science offers the right lens. Health systems are complex adaptive systems: their behaviour emerges from interactions between components, not from the components themselves, a point made in the BMJ as early as 2001 and ignored by capital planning ever since. We have relentlessly optimised the components. Complexity is the exhaust of capability: every unit of medical progress emits interactions, handoffs and decision points that someone, somewhere, must manage. For fifty years, no one was assigned.
2. Complexity is compounding faster than investment
If investment has grown impressively, patient complexity has grown faster, and it has changed in kind.
Start with who patients now are. A landmark Lancet analysis of 1.75 million Scottish primary care records found that nearly a quarter of the entire population, and the majority of people over 65, live with two or more chronic conditions. Multimorbidity begins ten to fifteen years earlier in deprived communities, and in absolute numbers most multimorbid people are of working age. This is not a Scottish curiosity; it is the demographic reality of every advanced health system. Chronic conditions now account for an estimated 70 to 80 per cent of health spending in the European Union. In the United States, the sickest five per cent of patients consume roughly half of all healthcare expenditure, and by some estimates a Medicare beneficiary with five or more chronic conditions sees around a dozen different physicians and fills some fifty prescriptions in a year.
The provider side mirrors this. Research on American primary care found that the typical Medicare beneficiary sees seven physicians across four practices annually, and that a single primary care doctor would need to coordinate with 229 other physicians in 117 practices to manage her panel. Those numbers date from before the last fifteen years of subspecialisation. They have not gone down.
Now look at what a care journey has become. The old mental model, still embedded in most pathway documents, is a line: presentation, diagnosis, treatment, recovery. The modern reality is a graph. Ingrid's journey involves primary care, emergency care, three specialist services, laboratory and imaging diagnostics, a hospital ward, municipal home care, a pharmacy, an informal caregiver and an insurer or regional payer, each holding a partial record, a partial budget and a partial mandate. Patients do not move through pathways anymore. They move through networks, and networks require orchestration that pathways never did.
Clinical science itself compounds the problem, because evidence is produced one disease at a time. A famous JAMA analysis modelled a 79-year-old woman with five common conditions and found that faithfully following every relevant clinical guideline would put her on twelve medications taken in nineteen daily doses, with multiple built-in conflicts. Each guideline was rational. The sum was not. No committee anywhere is responsible for the sum.
Researchers in the "minimally disruptive medicine" tradition have given this a name: treatment burden. For complex chronic patients, the workload of being a patient, the appointments, the referrals, the prescription logistics, the phone queues, the repeated retelling of one's own history, approaches a part-time job. Which exposes the quiet scandal at the heart of modern care: health systems have outsourced the integration function to the least resourced, least trained, most vulnerable actor in the network. The sickest patients are asked to serve as their own general contractors.
The conclusion is uncomfortable but unavoidable. The binding challenge in advanced health systems is no longer treatment. For almost any single condition, an effective intervention exists somewhere in the network. The challenge is orchestration: getting the right sequence of a dozen interventions, across a dozen institutions, to happen coherently for one human being.
3. The missing investment
Follow the capital, and the asymmetry becomes almost embarrassing.
Over the past two decades, advanced systems have poured money into nodes. Denmark committed a multi-billion-euro programme to build a generation of superhospitals. Hospital groups across Europe and North America have spent nine- and ten-figure sums on electronic health record implementations that primarily serve a single institution. Scanner fleets have expanded, robotic surgery platforms have proliferated, medical schools have grown, and thousands of digital health point solutions have been procured, each adding another node, another login, another partial view. Every one of these investments is individually defensible. Collectively they share a blind spot: not one of them answers the questions that determine whether the whole works.
Who coordinates Ingrid's care? Who owns the transition from ward to home, the single most dangerous week in her year? Who manages the handoff between cardiology and nephrology when their prescriptions collide? Who is accountable for continuity across the twelve months, and who can even see the whole journey? In most systems the honest answer is nobody. Or everybody, which in accountability terms is the same thing.
Economics explains how we got here. Ronald Coase observed in 1937 that firms exist to internalise coordination costs: when transactions are complex and interdependent, doing them inside one organisation beats contracting them across a market. Healthcare has spent decades drifting in the opposite direction. Care migrated out of the single institution, into day surgery, ambulatory clinics, home care, subspecialised centres and now virtual providers. The drift was clinically and economically sensible. But we dismantled the hospital's implicit coordination machinery, one roof, one chart, one team who knew the patient, and replaced it with nothing. We deconstructed the firm without building the network.
Payment systems then locked the gap in place. DRGs pay for episodes. Fee-for-service pays for activities. Even most capitation covers a single provider's slice. Every dominant payment model in the developed world pays for nodes; none pays for the journey. What is not paid for is not owned, and what is not owned is not invested in.
Management accounting delivered the final blow. In hospital cost structures, coordination is classified as overhead: an indirect cost, the first line trimmed in every efficiency programme. Discharge planners, case managers and care coordinators sit low in the status hierarchy, are chronically understaffed, and are measured, when measured at all, on throughput rather than outcomes. A hospital board will approve three hundred million euros for a building without existential debate, then agonise over three million for a coordination function, because one becomes an asset on the balance sheet and the other is an expense on the income statement. There is no line item for the in-between.
This is the missing investment, and it is a category error, not a funding shortfall. Coordination has been treated as an administrative burden to be minimised rather than a strategic capability to be built. Imagine a logistics company that invested only in warehouses and trucks, never in routing, and you have a fair portrait of how health systems have allocated capital since the 1990s.
4. The hidden cost of fragmentation
If coordination is the missing investment, fragmentation is the compounding interest, and the bill is larger than almost any executive assumes, because it is scattered across budgets that never meet.
The interface is the unit of failure. A study in the Annals of Internal Medicine found that roughly one in five patients suffers an adverse event within three weeks of hospital discharge, most of them medication-related, and a majority judged preventable or ameliorable. Systematic reviews find unintended medication discrepancies in somewhere between a third and two-thirds of patients at care transitions. The classic New England Journal of Medicine analysis of Medicare data found nearly one in five discharged patients readmitted within thirty days, and, most damning, that half of the medical patients readmitted had seen no physician at all between discharge and readmission. They did not fall through a crack. They fell through the space between two organisations, each of which had done its own job.
Add duplicate diagnostics, repeated because images and results do not travel. Add diagnostic delay, the referrals that die in transit and the abnormal results no one actions, a leading source of malpractice claims. Add the invisible tax on clinicians, who now function as human middleware: chasing results, re-entering data, reconciling records, managing inboxes, an established driver of the burnout that is itself shrinking capacity. Add the patient experience of narrating one's own history for the ninth time to the ninth stranger.
Then add it up. A comprehensive JAMA analysis put annual waste in US healthcare at 760 to 935 billion dollars, roughly a quarter of all spending, with "failure of care coordination" alone at 27 to 78 billion. That line item flatters the system. Failures of care delivery (over 100 billion), a substantial share of overtreatment and low-value care, and part of the 266 billion in administrative complexity are all downstream of fragmented journeys. The OECD, using different methods, estimates that around one fifth of European health spending is wasted. Whatever the precise figure, the direction is clear: modern health systems lose more value between providers than inside them. The white space on the org chart is the most expensive real estate in healthcare.
Why does so obviously costly a problem persist? Because fragmentation is an externality machine. The hospital that underinvests in discharge does not pay for the readmission; until penalty regimes arrived, it billed for it. The specialist who repeats a scan does not bear its cost. The benefits of coordination accrue to payers, patients and society; the costs fall on individual providers with no mechanism to recoup them. Under fee-for-service, a readmission is not a failure. It is revenue. Fragmentation, in other words, is not a malfunction of the system. In many markets it is the system, functioning exactly as financed.
There is a useful name for what has accumulated: coordination debt. Like technical debt in software, it is incurred every time a new capability ships without the connective work that should accompany it, and it compounds quietly, its interest paid daily in readmissions, duplication, delay and burnout. Also like technical debt, it appears on no balance sheet, right up until the moment it dominates the income statement.
5. Coordination is the new scarce resource
Every era of healthcare has been defined by its binding constraint, and strategy has always followed the constraint.
In the post-war decades, the constraint was capacity: too few beds, too few doctors. The rational response was to build and train, and systems did, on a heroic scale. From the 1980s onward the constraint shifted to capability: what could medicine actually treat? The rational response was technology and therapeutic innovation, and the pharmaceutical, device and diagnostic industries delivered. Both constraints have now loosened dramatically. In advanced systems, for most conditions, an effective intervention exists and is, in principle, reachable.
The constraint has moved again, and this time strategy has not followed. Eliyahu Goldratt's theory of constraints holds that a system's throughput is set by its bottleneck, and that improving anything other than the bottleneck creates not output but inventory. Look at healthcare's inventory: waiting lists, patients boarding in emergency departments, pending referrals, unread results, discharge-ready patients occupying acute beds because home care cannot be arranged. This is work-in-progress piling up at a constraint, and the constraint is not clinical capacity. Operating theatres sit idle while coordination fails; specialist appointments are consumed by patients who needed a different specialist. The bottleneck is the system's ability to manage interconnected journeys, and an hour lost at the bottleneck, as Goldratt taught, is an hour lost for the whole system.
Herbert Simon saw the deeper economics in 1971: a wealth of information creates a poverty of attention. Healthcare now generates more data per patient than any institution can absorb, while shared attention, someone actually holding the whole of Ingrid in view, has become the rarest commodity in the system. The scarce resource is no longer the physician, the facility or the therapy. It is coordinated attention at journey level.
Other industries hit this wall decades ago and responded in a strikingly consistent way: they built coordination infrastructure as a layer above their assets. Aviation is the canonical case. Global traffic grew to roughly forty million flights a year while fatal accident rates fell to parts per million, not because flying became simpler, but because the industry built ICAO standards, common operating procedures and air traffic control: a neutral orchestration layer that no airline owns and every airline obeys. Container shipping did the same at the interface itself: Malcolm McLean's standardised box collapsed cargo handling costs by more than ninety per cent and detonated an explosion in world trade, because it made every handoff between ship, truck and train frictionless. Modern logistics runs on control towers and real-time visibility platforms; a courier can track a thirty-euro parcel across three continents with more continuity than a hospital tracks a thirty-thousand-euro patient across a corridor.
Healthcare is nearly unique among complex industries in having multiplied its nodes for half a century without building the control layer. And the instinctive rebuttal, that interoperability will fix this, misses the point. Data exchange is necessary; it is not coordination. Giving every pilot the same radio frequency does not conjure air traffic control. Someone still has to watch the whole sky, sequence the traffic and own the outcome. Interoperability moves information. Orchestration moves patients.
This is why coordination is becoming, quite literally, healthcare's scarce resource: it is the input the system cannot currently manufacture at scale, and the one that determines the yield of every other input. In any production system, the bottleneck sets the price. Whoever relieves it captures the value.
6. The future winners will be orchestrators
The evidence that coordination pays is no longer speculative; it has simply been treated as a collection of charming exceptions rather than as a strategy.
Kaiser Permanente's integrated model, one record, one budget, aligned incentives across the journey, has for decades outperformed fragmented peers on chronic disease outcomes. Alaska's Nuka System of Care rebuilt an entire delivery system around relationship and journey ownership, and saw emergency department use fall by roughly forty per cent and hospital admissions by about a third over its transformation. In the Netherlands, Buurtzorg's self-managing neighbourhood teams turned radical continuity into an operating model; independent assessments found it needed markedly fewer care hours per client while topping national satisfaction rankings. In Germany's Kinzig valley, Gesundes Kinzigtal contracted to hold a population's journey economics through shared savings and produced measurable health and cost gains. In the United States, capitated high-touch primary care groups such as ChenMed and Oak Street built businesses in which keeping complex seniors out of hospital is the revenue model rather than a happy accident. The pattern across all of them is identical: someone, finally, owned the journey and its economics. A journey P&L existed.
Policy is converging on the same conclusion from above. England folded its purchaser-provider competition experiment and made 42 integrated care systems statutory in 2022, an institutional admission that competition between fragments had failed. Germany's hospital reform is forcing concentration and, through the belated national rollout of the electronic patient record, laying digital rails. Sweden's "God och nära vård" reform attempts to re-anchor journeys in primary care across its 21 regions and 290 municipalities. The Netherlands pioneered bundled payments for chronic care. Portugal is confronting the coordination consequences of more than a million residents without an assigned family doctor. And in the American market, the most telling signal of all: UnitedHealth's Optum has quietly become the largest employer of physicians in the United States. Payers are verticalising into orchestration because they have understood, before most providers, where the power in a networked system sits. The entities gaining ground are not those with the most beds. They are those closest to owning the journey.
What does competing on coordination actually require? Four layers, which together might be called the coordination stack. The first is visibility: a single longitudinal state of the patient that survives organisational boundaries, the shared picture of the sky. The second is navigation: a person and increasingly a system accountable for guiding each patient to the right next node, so that the general contractor role is finally lifted off the patient. The third is orchestration: the active management of handoffs, medications, tasks and logistics across providers, care traffic control in the most literal sense. The fourth is accountability: contracts, measurement and incentives set at journey level, outcomes and total cost per journey rather than per encounter. Most systems today possess fragments of the first layer and aspirations toward the fourth. The second and third layers are the void, and the void is the opportunity.
Follow the logic forward and the shape of future healthcare organisations changes. The winners of the next two decades will look less like service providers and more like network orchestrators: entities whose core asset is not a building but an operating picture, whose core competence is not a procedure but continuity, and whose strategic question has shifted from "what do we own?" to "what do we coordinate?" Hospitals become high-acuity nodes within journeys designed and managed elsewhere, much as airports remain essential yet do not run the airspace. For investors, the implication is that coordination infrastructure is an emerging category in the way logistics software and payment rails once were: horizontal, unglamorous, indispensable. The durable moat in digital health will not be the ten-thousandth point solution, each of which adds another node to the graph, but the layer that binds the nodes. And in networked markets, one rule holds with monotonous regularity: the orchestrator, not the asset owner, ends up setting the terms.
7. The strategic conclusion
Strip the argument to its studs and it reads like this. Healthcare's binding constraint has moved, from capacity, to capability, to coordination, and capital allocation has not moved with it. The greatest gap in modern healthcare is not money. Advanced systems already spend ten to twelve per cent of GDP, seventeen in the American case, and the marginal euro poured into capacity now buys less health than at any point in history. The marginal euro invested in coordination, by contrast, is very likely the highest-yield investment available in the sector, because it does not add capacity: it activates capacity that has already been paid for.
Every executive team can test itself in an afternoon. Take the last decade of capital allocation and sort it into two piles: investments in nodes, buildings, devices, systems that serve one institution, and investments in the network, anything that manages the journey across institutions. For most organisations the split is not 80/20. It approaches 100/0. Then ask the five questions this article began with. Who coordinates? Who owns transitions? Who manages handoffs? Who ensures continuity? Who sees the entire journey? An organisation without confident answers does not have a coordination strategy. It has a coordination hope, and its patients' daughters have ring binders.
None of this diminishes what medicine has built. It reframes it. Fifty years of investment created a system capable of treating almost every condition, and no system capable of guiding a human being through the complexity that capability created. We industrialised the encounter and left the journey to chance. The value is now leaking out precisely where no one is looking: in the white space between excellent institutions.
Which is why the next healthcare transformation will not resemble the previous ones. It will not arrive as a molecule, a scanner or a model. It will arrive as infrastructure: the connective layer that turns institutional excellence into system performance, as air traffic control once turned dangerous individual flights into the safest complex system humanity operates. Building it is not an IT project or an administrative tidy-up. It is the founding of a new strategic capability, and, plausibly, of healthcare's next great industry.
The organisations that see this first, providers who take ownership of journeys, payers who contract for them, investors who fund the rails, policymakers who pay for the in-between, will define the coming era of care delivery, just as the builders of hospitals defined the last one. For every board, the question is no longer whether to invest in coordination. It is whether to build the control tower, or wait to be routed by someone else's.