For over a decade, bundled payments have been hailed as one of the most promising tools for fixing American healthcare. The logic is elegant: instead of paying providers for every test, visit, and procedure, pay a single fixed price for an entire episode of care—a knee replacement, say, from surgery through rehabilitation. Efficient providers keep the savings. Inefficient ones absorb the losses.

It sounds like exactly the kind of incentive realignment health economists have been calling for since the 1970s. And Medicare, along with private insurers, has run increasingly ambitious bundling experiments, covering everything from cardiac care to maternity.

Yet despite the theoretical elegance and policy enthusiasm, bundled payments have not transformed healthcare delivery the way advocates predicted. Spending curves remain stubbornly upward. Fee-for-service remains dominant. Why has such a sensible idea produced such modest results? The answer lies in three challenges—technical, behavioral, and empirical—that reveal a deeper truth about healthcare payment reform.

The first problem is deceptively simple: what exactly belongs in a bundle? A hip replacement sounds like a discrete event, but the clinical reality is messier. Does the bundle start with the first specialist consultation or the day of surgery? Does it end at discharge, at 30 days, at 90 days? What about the physical therapy the patient needed anyway for an unrelated back issue?

Each boundary decision has enormous financial consequences. Set the episode too narrowly, and providers simply shift costs outside the window—discharging patients early, then billing separately for readmissions. Set it too broadly, and providers bear risk for complications they cannot reasonably control, like a patient's diabetes flaring up months later.

Medicare's Comprehensive Care for Joint Replacement program, for instance, uses a 90-day episode. That choice wasn't arbitrary; it reflected extensive negotiation over which complications are plausibly related to surgical care versus which belong to the patient's underlying health. Every bundle requires hundreds of such judgments, and each one creates opportunities for gaming.

The result is that bundle design becomes a perpetual cat-and-mouse game between regulators refining definitions and providers finding edges. Unlike a restaurant menu, where a pasta dish is obviously one item, medical care resists clean unit definitions. The human body doesn't deliver illness in tidy packages.

Takeaway

The elegance of any payment model depends entirely on how cleanly you can define the product being purchased—and healthcare stubbornly resists clean definitions.

The second challenge is more troubling because it cuts against the policy's equity goals. When providers accept a fixed price for an episode, they bear the financial risk of complications. This creates a powerful, if often unconscious, incentive to avoid patients most likely to have complications—the sick, the frail, the socially disadvantaged.

Health economists call this cream-skimming or adverse selection. A surgeon paid a bundled rate for knee replacements has every reason to prefer a healthy 62-year-old weekend athlete over a 78-year-old with diabetes, obesity, and limited home support. The second patient will likely cost more, recover slower, and threaten the bundle's margins.

Research has documented these patterns. Studies of Medicare's joint replacement bundles found evidence that some hospitals shifted sicker patients to providers outside the bundled program. In cardiac bundles, analysts have noted subtle shifts in patient mix toward lower-risk cases. Risk adjustment helps, but no formula fully captures the dozens of factors that make one patient harder to treat than another.

The policy irony is sharp: a reform meant to improve efficiency may worsen access for the patients who need care most. Bundled payments can deliver better outcomes on average while simultaneously making the system less welcoming to the sickest patients—a troubling trade-off that averages conceal.

Takeaway

Whenever you transfer financial risk to providers, you also give them reason to be selective about which patients they accept. Efficiency and equity can pull in opposite directions.

After a decade of implementation, the evidence on bundled payments is surprisingly mixed. Some programs show modest savings—typically 1 to 3 percent per episode—driven largely by reduced post-acute care use, particularly shorter stays in skilled nursing facilities. These are real gains, but hardly the transformation advocates envisioned.

More importantly, many savings may be illusory at the system level. If hospitals discharge joint replacement patients home instead of to rehab facilities, Medicare saves money on that bundle. But if those patients then use more outpatient therapy or experience more falls, total spending may shift rather than shrink. Evaluations that focus on the bundle itself can miss what happens around its edges.

Randomized and quasi-experimental studies of Medicare's bundled programs have found meaningful reductions in specific episodes, especially orthopedic procedures with predictable trajectories. But extensions to medical conditions like pneumonia or heart failure have produced weaker results. Illness with variable courses resists the tidy pricing that bundles require.

Perhaps most tellingly, bundled payments have not bent the overall healthcare cost curve. National spending continues rising at rates driven by factors—drug prices, administrative overhead, labor costs, technology diffusion—that no payment reform touching 2 percent of episodes can meaningfully address.

Takeaway

Payment reforms can change provider behavior at the margins without changing the fundamental economics of a system. Structural costs require structural answers.

Bundled payments aren't a failure. They've produced real efficiencies in well-defined surgical episodes and pushed providers to think about care as a continuum rather than a series of billable events. That's genuine progress.

But the experience reveals an important lesson about healthcare reform: elegant incentives cannot overcome messy realities. Payment models must contend with clinical complexity, patient heterogeneity, and the full ecosystem of providers around any single episode.

The next generation of payment reform will likely need to combine bundles with population-level accountability, better risk adjustment, and direct regulation of the cost drivers that payment models alone cannot touch. Progress in healthcare tends to come not from silver bullets, but from patient accumulation of imperfect tools.