To choose between clinical data abstraction companies and an in-house team, compare the full cost per chart on both sides rather than the vendor's rate alone. Then insist on contract terms for turnaround, error handling, a business associate agreement and data return, and plan how you will audit the company's work once charts start coming back.
What clinical data abstraction companies deliver
Clinical data abstraction companies sell trained people who read your charts and fill in the data elements you need, usually for a registry, a quality program, research or a payer review. You send charts or give remote access to your records system. They return completed forms or files, often with a turnaround date and a quality report attached.
The work itself is the same work an in-house abstractor does. A person reads the chart, finds each element the data dictionary names, records the value and notes where it came from. What a medical record abstractor does each day, and the skills the role needs, are covered in what a medical record abstractor does.
Medical data abstraction companies differ mainly in how they package that work:
- Project-based. A fixed set of charts and elements, priced per chart or per project, with an end date.
- Ongoing. A steady monthly volume, often for a registry with regular submission deadlines.
- Staff extension. Abstractors who work inside your systems under your procedures, billed by the hour.
- Backlog clearing. A short burst of capacity to catch up before a deadline.
Ask each company which of these it offers and which it does most. A company built for steady registry work may not suit a one-off research project, and the reverse.
Also ask what the company does not do. Some return only the completed forms and leave submission to you. Some do not handle chart retrieval from outside facilities. Some abstract only from electronic records and not from paper or scanned charts. Each gap is work your team still owns, and it belongs in your cost comparison.
Registry work has its own rules for how elements are defined and how agreement between abstractors is checked. The guide to clinical registry abstraction covers those rules and the audits registries run.
Outsourced vs in-house chart abstraction costs
Outsourced vs in-house chart abstraction costs differ in where they hide. An outside company's cost is mostly its rate per chart or per hour, plus your time managing it. An in-house team's cost is salaries, training, tools and the time lost to turnover. Neither side is cheaper by default; only a full worksheet tells you.
The outsourced side has costs beyond the invoice. Someone on your team writes the specifications, answers the company's questions, sends charts, checks returned work and fixes errors. Charts the company cannot finish, because pages are missing or access failed, come back to you. If you switch companies later, you pay for setup again.
The in-house side has costs beyond salary. New abstractors need training before their work can be trusted. Experienced abstractors leave, and their replacements start slow. Volume rises and falls, so a team sized for the busy months sits idle in quiet ones. Someone has to check the team's work as well.
Cost comparison worksheet
Fill in your own numbers for each line. The example values are invented for illustration and are not estimates of real costs.
| Line item | Outsourced (example) | In-house (example) |
|---|---|---|
| Charts per year | 1,200 | 1,200 |
| Vendor rate per chart | $A x 1,200 | None |
| Abstractor salaries and benefits | None | Salary x number of staff |
| Recruiting and training | None | Hours to train each new hire x hourly cost |
| Project management on your side | Hours per month x hourly cost | Hours per month x hourly cost |
| Specification writing and updates | Hours per data dictionary change | Hours per data dictionary change |
| Quality checks on returned work | Sampled charts x minutes each | Sampled charts x minutes each |
| Rework of errors | Charts returned x minutes each | Charts redone x minutes each |
| Charts the vendor could not finish | Count x minutes to finish in-house | None |
| Software and access | Remote access setup | Abstraction or review software |
| Setup if you switch later | One-time cost | None |
| Total per year | Sum | Sum |
| Cost per chart | Total / charts | Total / charts |
Run the worksheet twice: once at your current volume and once at the volume you expect next year. The answer can change as volume moves, because outsourced costs scale with charts while in-house costs move in steps as you hire.
Some costs do not fit in a cell. An in-house team can answer a question about a chart the same afternoon, while an outside company works through tickets and agreed hours. An in-house team builds knowledge of your charts and your physicians' habits over years; a company may rotate abstractors across clients. Against that, a company can absorb a sudden spike in volume that would swamp a small team. Write these trade-offs next to the totals so the decision is not made on the numbers alone.
Which contract terms should abstraction companies agree to?
Abstraction companies should agree to contract terms on turnaround, error handling, privacy and data return. Turnaround says when work is due. Error handling defines an error and what happens when one is found. A business associate agreement governs the protected health information they touch. Data return says what you get back when the contract ends.
Turnaround
Write the turnaround as a measurable promise: a number of business days from when a chart is available to when the completed abstraction is returned. Say what happens when the deadline is missed, whether that is a fee credit, priority handling or the right to end the contract. Tie the clock to something both sides can see, such as the date a chart appears in a shared queue.
Accuracy and error handling
A promise about accuracy means little unless the contract defines it. Agree on what counts as an error: a wrong value, a missing value, a value from the wrong encounter. Agree on how it is measured, such as re-checking a sample of charts each month, and who does the re-checking. Agree on what happens when the error rate on a sample is too high: free rework, a credit, or a review of the company's process. Choose the target rate yourself, based on what your program needs.
Business associate agreement
A company that abstracts your charts handles protected health information on your behalf, so it often fits the HIPAA definition of a business associate in 45 CFR 160.103, which names data analysis and quality assurance among the covered functions. The required terms of a business associate agreement are set out in 45 CFR 164.504(e) on the eCFR. Among them, the business associate must report "any use or disclosure of the information not provided for by its contract," and must ensure that any subcontractors that handle the information "agree to the same restrictions and conditions." Ask directly whether the company uses subcontractors, including abstractors in other locations, and get the answer in writing. More on how these agreements work is in the HIPAA and PHI category. That category explains what HHS publishes; it is not legal advice for your contract.
Data return
The same regulation requires the agreement to provide that, at termination, the business associate will "return or destroy all protected health information" it still holds, "if feasible," and keep no copies. Your contract should go further for your own data: say in what format the completed abstractions, notes and audit records come back, and how quickly.
How do you audit an abstraction company's quality?
You audit an abstraction company's quality by re-abstracting a sample of its finished charts yourself and comparing the answers element by element. Do it on a schedule, with a sample large enough to cover each element and each abstractor, and track the results over time. A quality report from the company is a starting point, not a substitute.
A workable audit routine:
- Pick the sample. Each month, select a set of completed charts at random, plus any charts flagged as unusual. Make sure every abstractor the company assigns to you appears in the sample over time.
- Re-abstract blind. Have your own reviewer fill in the same elements without looking at the company's answers.
- Compare. List every element where the two answers differ. For each, go to the chart and decide which is right.
- Classify. Record each confirmed error by element, by abstractor and by type: wrong value, missing value or wrong source.
- Report back. Send the company the findings and ask for a cause and a fix for any pattern.
Patterns matter more than single errors. One element that fails across many charts usually points to an unclear definition, which may be your specification's fault rather than the company's. One abstractor whose errors cluster points to training. A rise in errors after a staffing change at the company is worth raising quickly.
How to measure agreement between two abstractors, including the statistics registries use, is covered in the guide to how agreement between abstractors is measured.
Software can help an in-house team or an auditor here. Schema Care reads medical records into fields with a confidence score and a page link for each, and holds uncertain fields for a person to approve or correct; the threshold demo on a fictional sample shows how the score decides what goes to review.
Abstraction vendor lock-in and data ownership
Abstraction vendor lock-in happens when your abstracted data, notes and history live in a company's system in a form you cannot take with you. When the contract ends, you may lose the audit trail behind past submissions, or have to pay to get your own data back. Settle ownership and export before the first chart goes out.
The risk is easy to miss at the start. Early on, the company's portal holds everything and works well. Years later, that portal may hold the only record of who abstracted a chart, which source page supported a value, and what was corrected after review. If a registry or payer later questions a value, you need that history.
Clauses to ask for:
- Ownership. All abstracted data and work products belong to you.
- Export on request. You can get a full export at any time, not only at the end, in a documented format you can open without the company's software.
- What is included. The export covers values, source references, reviewer notes, correction history and dates.
- Exit help. The company will support a handover to a new provider or to your own team for a set period.
- Destruction proof. After return, the company confirms in writing that it has destroyed what it still holds, consistent with the business associate agreement.
Test the export early. Ask for a sample export in the first month and check that you can read it, match it to your charts and load it where you need it. A clause you have never tested is a promise, not a plan.