Healthcare Needs a Moneyball Moment

Story ByGuest

•

19 hours ago

•

3 Mins Read

We don't need more metrics. We need to determine which ones actually predict winning.

By- Dr. Feby Abraham | Founder and CEO, Entelix Health. Advisor at ScaleHealthTech. Former Health System Chief Strategy and Innovation Officer. Former McKinsey Partner

In 2002, the Oakland A's had one of the smallest payrolls in Major League Baseball and one of the best records in the game. Most people remember the story from Michael Lewis's book and the Brad Pitt film: Billy Beane replaced the instincts of scouts with statistics.

That is not quite what happened, and the difference matters for what we are about to explore.

Beane did not discover that batting average was useless, and he did not invent the analysis. Bill James and a generation of outsiders had been building it for years, and on-base percentage had been sitting in plain sight in the same box scores for decades. What Beane did was act on predictive relationships the market had not yet priced correctly. Every club was paying a premium for batting average and home runs. He bought the players the rest of the league had undervalued against a better measure.

Other industries have had versions of that moment. Retail stopped managing to same-store sales and started managing to customer lifetime value. Software stopped counting license bookings and started counting net revenue retention.

Healthcare's version remains unfinished. It is not that we lack predictive measures. We have built a great many of them: risk adjustment, HEDIS, bundled payments, total cost of care, population health analytics. The problem is that they rarely displace volume-based measures when the decisions actually matter. They inform. They almost never govern. Healthcare has assembled most of the components of Moneyball without ever integrating them into the scoreboard that governs capital, talent and rewards.

We are extraordinarily good at optimizing the wrong objective function

Walk into any health system and you will find dashboards: encounters, RVUs, revenue per bed, occupancy, market share, NPS, readmissions, risk-adjusted mortality, days in accounts receivable. Hundreds of indicators, tracked with real rigor.

None of them are wrong. Each measures something genuine. The problem is subtler and more dangerous than measuring the wrong things: an organization can become world-class at optimizing every one of them without meaningfully improving what the enterprise exists to produce.

That gap between what we optimize and what we are trying to predict is healthcare's mispricing. And it compounds. Capital flows toward the assets that score well on today's board. Talent is promoted for the behaviors it recognizes. Partnerships get counted rather than valued. The scoreboard quietly writes the strategy.

Step one: Define Winning

Baseball has an advantage: winning is unambiguous. Healthcare's objective function is multidimensional, and the five things health systems say they care about do not carry equal weight. Treating them as parallel goals gives a board no way to reason about trade-offs. They are better understood as a causal architecture.

Ultimate outcomes: healthy life, affordability, equity. What the enterprise exists to produce. The U.S. ranks last among the 10 high-income nations studied on access, equity and outcomes while spending more than twice per capita.
Enterprise requirements: sustainable economics and workforce capacity. Not the purpose, but the preconditions. 197 rural hospitals have closed or converted since 2005, and the U.S. faces a projected shortfall of up to 86,000 physicians by 2036.
Leading mechanisms: trust, continuity, access, engagement. The levers that move the outcomes, and the earliest signals that they are moving. Today they are weak: U.S. consumers are about 60% loyal, and frequent users of care split it across an average of 4.2 provider networks.

Enterprise requirements are constraints to satisfy, not goals to maximize. Leading mechanisms are where a scoreboard earns its keep, because they are observable long before outcomes are.

Introducing the Healthcare Sabermetrics Scoreboard

This is the exercise: not a longer dashboard, but a different one.

Three things deserve emphasis. First, almost none of these are inventions. Time-in-therapeutic range, healthy-at-home days, continuity ratios, share of care and risk-adjusted panel outcomes can be computed today from data health systems already hold. Like on-base percentage, they already exist. They are simply not on the board.

Second, the right-hand column matters most. A metric that cannot answer "what am I trying to predict?" is a number, not a scoreboard. The discipline of naming the prediction is what separates sabermetrics from analytics.

Third, that discipline cuts both ways. Applied honestly, it exposes this table as a starting point rather than a finished framework. Some entries are outcomes, some are proxies, and only some are true leading indicators. Risk-adjusted total cost of care is largely an outcome. Likelihood-to-return is a plausible improvement on NPS but not yet a proven one. Measures such as likelihood-to-return and share of care are also valuable only when paired with outcomes and appropriate utilization; pursued alone, they can reproduce the same volume incentives under different names. Before any measure governs capital, it should survive five questions:

Predicts what, specifically?
Over what horizon?
For which population and business model?
On what evidence, in our own data?
Is it actionable before the outcome occurs, or only explanatory afterward?

Too many of healthcare's most influential enterprise metrics fail the last question. That is the real indictment.

So, what is Healthcare's on-base Percentage?

There may not be one. A cancer center, a safety net system and a full-risk primary care platform are not playing the same season. But for an enterprise that intends to carry risk, the organizing concept is close to:

This is an objective function, not an accounting measure. It sits above a portfolio of components, sustained disease control, continuity, consumer trust, appropriate utilization and risk-adjusted total cost, which feed it rather than multiply into it. The aim is not to manufacture a falsely precise enterprise number. It is to create a governing objective against which a small portfolio of measurable outcomes and leading indicators can be tested.

The test is whether it changes a decision. Consider two primary care investments competing for the same capital: a fifteen-provider clinic acquisition in a growing suburb, or an equivalent sum spent embedding pharmacists, behavioral health and care management into an existing panel.

On the traditional scoreboard the acquisition wins easily. It adds covered lives, downstream referral volume and market share, and it can be modeled with confidence. On the objective function above, the question changes: which option produces more controlled disease days, fewer avoidable acute episodes, higher continuity and lower risk-adjusted cost per patient? That is a harder question. It is also the one that determines which investment creates value in a risk-bearing world, and most capital committees are never asked it.

Where are Healthcare's Undervalued Players?

The most durable lesson of Moneyball was never analytics. It was arbitrage. Beane's edge did not come from knowing on-base percentage mattered, because plenty of people knew. It came from acting on it while the market still paid for something else.

So the strategic question is not what to measure. It is: against a scoreboard of healthy life days per dollar, which assets does this market underprice?

Primary care physicians and advanced practice providers, who generate fewer RVUs while keeping populations healthier. Priced on production, valuable for prevention.
Pharmacists and community health workers, who move adherence and prevent downstream utilization, with almost no billing code to their name.
Behavioral health clinicians, under-reimbursed while sitting directly upstream of enormous medical spend.
Virtual care and hospital-at-home, which substitute operating models for capital, a trade the current scoreboard barely recognizes because it rewards the asset rather than the outcome.
AI and ambient documentation, still budgeted as an expense line when what they return is clinician time, the scarcest asset in the system.
Upstream interventions whose downstream value is poorly captured, including nutrition, sleep health and caregiver support. These interventions may materially affect chronic disease and utilization, but their value often accrues across time, conditions and organizational budgets rather than through a single reimbursable encounter. They are not necessarily inexpensive; they are difficult for the current scoreboard to value.

Beneath all of it sits the largest unpaid workforce in healthcare: patients and their caregivers, who deliver most of the actual care and receive almost none of the investment.

None of these are cheap. Several carry real cost. They are undervalued on today's scoreboard, which is a different and more actionable claim.

The Hidden Gems Index

The same logic applied inside the organization produces the most immediately usable idea here.

The unit of analysis is the individual clinician or care team. The dimensions are three: risk-adjusted panel outcomes, patient continuity and return behavior, and traditional RVU production. The hidden gems combine superior risk-adjusted outcomes with unusually strong continuity and patient trust, despite below-median RVU production. The index would require careful adjustment for panel complexity, access, panel size and referral patterns; otherwise it could mistake favorable selection or constrained capacity for superior practice.

On the current scoreboard they look like underperformers and show up in productivity reviews as a problem to manage. On the new one they are the most valuable people in the enterprise, and the index changes three concrete decisions: who is protected rather than pressured in a compensation redesign, whose practice patterns get studied and replicated, and where the most complex panels are routed.

Five Moves

Moneyball thinking, translated into an operating agenda:

Define winning explicitly. At the board level, with a stated hierarchy, before touching a dashboard.
Find the measures that actually predict it. A handful per objective, each surviving the five questions above.
Identify the assets undervalued against those measures. People, care models, technologies, partnerships.
Reallocate capital and talent toward them. Deliberately, and before the market reprices them.
Change incentives so the organization plays to the new scoreboard. Compensation, capital committee criteria, service line reviews.

Only the second move is analytics. The fourth and fifth are where most transformations quietly die.

Why the Old Scoreboard Persists

The deepest obstacle is not metric design and it is not data. The old scoreboard persists because it distributes resources, authority and status. RVUs, volume, beds, service line contribution and capital-intensive growth are embedded in compensation formulas, budget cycles, governance structures and professional standing. Every one has a constituency that organized itself around it, often over decades and often in good faith. Changing the scoreboard is not an analytics project. It is a redistribution, and it should be planned as one.

The familiar objections are real but not disqualifying. Payment still rewards the old board, so run both and be explicit about which one wins when they conflict. Attribution is hard, so start where you carry risk and attribution is clean, prove the relationship in your own data, then extend. The first proving ground should be a defined risk-bearing population, such as an attributed Medicare Advantage or employee health cohort, where outcomes, utilization and economics can be observed together. And new metrics get gamed, exactly as the old ones were, which is the argument for a small portfolio that is hard to game simultaneously: you can inflate share of care or you can improve disease control, but doing both while lowering risk-adjusted total cost is very difficult to fake.

The Bottom Line

Healthcare's central problem is no longer the absence of data. It is the scoreboard governing how that data gets used.

We measure a great deal. We reward what we measure. And then we act surprised when the system produces exactly what we rewarded.

So the question I would put to any board considering this framework is not whether they agree with it. It is what follows: what would your organization stop funding, stop rewarding, or stop building if it genuinely adopted this scoreboard? If the answer is nothing, the scoreboard has not changed.

Billy Beane's insight was not that baseball needed more statistics. Baseball already had plenty, and most of the good ones had been worked out by people no club was listening to. His insight was recognizing which statistics actually predicted winning, and which assets the rest of the market had consequently mispriced.

Healthcare's Moneyball moment arrives when we do the same.