By Brent Dover, Chief Executive Officer, Carta Healthcare
Some costs in healthcare are loud. They show up in headlines, board meetings, and budget battles. Others are quiet, absorbed into the normal way of doing business until someone adds them up. Manual clinical data abstraction is one of the quiet ones, and the sum is larger than most people expect.
Across U.S. health systems, manual abstraction consumes an estimated ten to fifteen billion dollars every year. It is one of the most labor-intensive workflows in quality operations. And unlike many costs, it does not get more efficient as an organization grows. It tends to get heavier.
Start with the unit economics of the work itself. A standard case requires roughly thirty minutes to abstract. A complex one can consume five to six hours of a skilled professional's time. Multiply that across a registry program at a large health system and the labor reaches well beyond eleven thousand hours a year for a single registry, before counting the rest of the portfolio.
The biggest time sink is patient history. Abstractors scroll manually through years of encounters, outside records, and procedure reports to locate the dates, prior diagnoses, and comorbidities a registry requires. This is painstaking, attention-heavy work, and it does not scale by working faster. There is a floor to how quickly a person can responsibly reconstruct a patient's clinical story from fragmented documentation.
If the only problem were the hourly cost, abstraction would be a line item to negotiate. The real issue is structural. In a manual model, growing a clinical service line means growing abstraction headcount in rough proportion to case volume. More cardiac procedures mean more cardiac abstraction hours. More registry participation means more people. The cost curve rises with ambition, which means the very success a quality program is built to support also makes it more expensive to sustain.
That dynamic makes program expansion operationally fragile. A service line that wants to grow has to justify not only clinical and operational investment but a corresponding increase in abstraction capacity. When that capacity is hard to add, expansion slows, and a quality infrastructure meant to enable growth quietly becomes a constraint on it.
Volume is rarely smooth. It spikes with seasonality, new programs, and referral patterns. In a manual model, a spike turns directly into a backlog, because there is no slack to absorb it. Backlogs are not just a scheduling annoyance. They create data gaps that affect audit readiness and threaten submission deadlines. The cost of the manual model is highest at exactly the moment an organization can least afford it, when volume surges and the calendar does not move.
There is a workforce dimension to this as well. Skilled abstractors are expensive and difficult to retain. Turnover does more than create vacancies. It introduces inconsistency in data quality across registries and facilities, because judgment built up over years walks out the door and has to be rebuilt. A model that depends entirely on a scarce and mobile workforce inherits the volatility of that workforce.
The intuitive response to all of this is to hire more abstractors, and at some scale that response stops being viable. The talent pool is limited, training takes time, and each addition raises the coordination and quality-control burden. Linear cost growth is tolerable for a while and then it is not, particularly for organizations expanding their registry footprint to capture more quality and reimbursement opportunity. The manual model does not break loudly. It simply makes growth progressively more expensive until the math discourages the growth.
The point of naming these costs is not to argue that abstraction is unimportant. It is the opposite. Abstraction matters enough that it should not be the thing that limits how much quality work an organization can take on. The goal of bringing AI into the workflow is not merely to shave hours off individual cases, though it does that. It is to change the shape of the cost curve, so that expanding a service line or adding a registry no longer requires a proportional expansion of manual labor.
The reason this constraint is worth removing becomes vivid when you imagine it gone. When the cost of abstraction no longer rises in lockstep with case volume, a quality program starts to behave differently. Expanding registry participation stops being a staffing negotiation. Adding a service line stops requiring a proportional increase in manual hours. The organization can pursue the quality and reimbursement opportunities it wants on the strength of its clinical strategy rather than the size of its abstraction workforce.
That shift shows up in real patterns of behavior. Health systems that break the link between volume and headcount tend to expand the scope of what they abstract, taking on more registries because the marginal cost of doing so has changed. Supporting a wide range of registries, from cardiovascular and vascular to trauma, oncology, and surgical quality programs, becomes a matter of capacity rather than a matter of hiring. The constraint that quietly capped ambition becomes optionality instead.
There is also a resilience dividend. A model that does not depend entirely on a scarce and mobile workforce is less exposed to the volatility of that workforce. Turnover still happens, but it no longer translates directly into inconsistency across registries and facilities, because the institutional knowledge is not carried solely in individual heads. The point is not that people matter less. It is that the program no longer breaks when a key person leaves, which is its own form of cost avoided.
Reframed this way, abstraction stops being a cost center to minimize and becomes a capacity to manage. The strategic question is no longer how to spend as little as possible on a necessary chore. It is how much quality and reimbursement opportunity the organization can pursue once the work no longer scales in lockstep with people. That is a more ambitious question, and it is the right one. A health system that can grow its registry footprint without growing its abstraction staff in proportion has turned a constraint into a lever, and a lever is what lets a quality program expand on its own terms rather than on the terms of its headcount.
That is the difference between a cost you reduce and a constraint you remove. Trimming minutes per case is useful. Breaking the link between case volume and headcount is transformative, because it lets a quality program grow on the strength of its clinical ambition rather than the size of its abstraction staff. The ten to fifteen billion dollar figure is the headline. The constraint hidden inside it is the part worth fixing.
