What Claims Data Can Reveal About HCP Clinical Activity

Two ophthalmologists can share a taxonomy code, a hospital affiliation, and a ZIP code while running very different practices.

One spends most operating days on cataract surgery. The other runs a retina service built around intravitreal injections and retinal imaging. A specialty directory files them side by side, and a commercial team launching a retinal therapy would treat them as equal targets until a rep spends a morning in each waiting room and learns otherwise.

The same split shows up between neurologists who share a title while one focuses on migraine and the other on neurodegenerative disease, and in the wide range of service volumes that CMS utilization files record under a single specialty code.

A rep planning the week from a specialty list cannot see those differences until the visit, and HCP claims data shows them before anyone gets in the car.

Every billed service generates a record carrying a procedure code, the diagnosis that justified it, a date, a place of service, and the National Provider Identifier of the clinician who performed it. Aggregated across payors and months, those records describe HCP clinical activity with a precision that specialty labels and self-reported profiles cannot reach.

That precision has conditions attached. Claims are billing documents, and the rules that govern how they are coded also govern what they can prove. Commercial teams that read HCP claims data well know which codes carry real signal about a physician’s focus, and they know exactly where the record goes quiet.

What HCP Claims Data Can Reveal About Clinical Activity

Most provider data describes identity. Licensing records, training history, and directory listings establish who a clinician is, and they change slowly. Clinical activity moves faster.

A surgeon adds a new technique, shifts cases to an ambulatory surgery center, or joins a larger group, and none of it registers in a credential file. Claims fill that gap in three ways, by recording activity as it is billed, by describing how a practice operates day to day, and by placing each provider in a regional context.

Billing Records Capture Activity That Credential Files Miss

Claims data in healthcare captures activity because a claim is generated each time care is billed. The FDA’s framework for its Real-World Evidence program names medical claims and billing data alongside electronic health records and registries as primary sources of real-world data, which says a great deal about how much routine care leaves a billing trace.

For a commercial team, that trace answers the questions that actually decide where reps spend their week, such as whether a surgeon performs the procedure a device supports and at what volume.

The code sets themselves keep pace with clinical practice. The AMA’s CPT 2026 release added 288 new codes, and Category III codes for emerging services made up 27% of them. Once a new technique has a code, its adoption becomes visible in physician claims data, provider by provider, long before it appears in a published case series.

Practice consolidation adds a second reason to work at the provider level. According to the AMA’s Physician Practice Benchmark Survey, the share of physicians in private practice fell from 60.1% in 2012 to 42.2% in 2024.

As more clinicians practice inside large hospital-owned or multi-specialty groups, the organization name says less about any one physician’s work. NPI-level claims separate the individual from the group, so a team can see which of forty cardiologists in a system actually performs structural heart procedures.

Volume, Patient Counts, and Setting Show How a Practice Actually Runs

Read together, the fields on a claim build a working profile of a provider’s practice. Each dimension answers a different commercial question, and the useful insight usually comes from combining them.

Procedure volume is the most direct measure. A count of billed CPT or HCPCS codes over a defined period shows how often a provider performs a given service, which makes it possible to rank surgeons by total knee arthroplasty volume (CPT 27447) or dermatologists by Mohs surgery volume (CPT 17311) instead of inferring activity from specialty alone.

Patient counts change how that volume reads. A retina specialist with 1,200 intravitreal injection claims across 150 patients runs a recurring treatment practice, with each patient returning on a schedule. A surgeon with 1,200 claims across 1,100 patients performs largely one-time procedures and depends on a steady flow of new referrals to sustain that number. The distinction shapes how a manufacturer sizes demand for a consumable compared with a capital device, even when headline volume looks identical.

Place-of-service codes show where care happens. CMS maintains them as a two-digit code set that professional claims use to identify the setting where a service was rendered. The same procedure billed in a physician office, a hospital outpatient department, or an ambulatory surgery center carries different purchasing implications, since the buyer, the contracting process, and the value analysis committee all change with the setting.

Reimbursement shifts as well, because Medicare uses the place-of-service code to decide whether many physician services are paid at the facility or nonfacility rate.

The AMA’s 2026 overhaul of lower extremity revascularization codes was explicitly framed around care moving toward outpatient settings, and claims are where that migration becomes measurable for an individual provider.

Trajectory completes the picture. Comparing a provider’s volume quarter over quarter separates physicians who are growing a procedure from those who are winding it down.

For HCP claims data analysis, a provider at moderate volume with a steep upward slope is often a better prospect than a high-volume provider whose numbers have been flat for three years.

Regional Baselines Change What a Provider’s Volume Means

Clinical activity is not evenly distributed, and the unevenness rarely tracks population alone. A cross-sectional analysis of Medicare beneficiaries found a 6.2-fold difference in thyroidectomy rates across US hospital referral regions, and the variation did not correlate with surgeon availability, socioeconomic status, or access to care. The researchers attributed it largely to local practice patterns.

That finding has a direct implication for reading HCP claims data. A surgeon’s volume means little without a regional baseline.

Two hundred procedures a year might place a provider in the top tier of one region and in the middle of another, and a national ranking flattens that difference. Commercial teams get a more accurate picture by comparing each provider against peers in the same market, then asking why a region runs high or low before treating it as an opportunity.

Territory design benefits most from geographic claims patterns. Territories drawn by ZIP code or population tend to produce uneven workloads, because procedure volume clusters around particular health systems and referral hubs. Mapping claims volume onto geography, then accounting for real driving distance between accounts, produces territories balanced by clinical opportunity.

For field teams covering rural regions, drive time often determines whether a high-volume account can be serviced at all.

Using Diagnosis and Procedure Claims to Understand HCP Focus

Procedure codes record what a clinician did during an encounter, and the diagnosis codes on the same claim record the medical reason for doing it.

Reading the two together shows which patient populations sit behind a provider’s procedure volume, and that detail often separates a qualified target from a merely plausible one.

Diagnosis Detail Determines Whether a Provider Matches the Indication

Retina practice offers a clean illustration. Intravitreal injection (CPT 67028) appears in the claims of nearly every retina specialist, so procedure volume on its own cannot tell a manufacturer which ones treat the condition its therapy is indicated for.

The diagnosis codes attached to those injections can. Injections linked to exudative age-related macular degeneration (ICD-10-CM H35.32-) indicate a different patient population than injections linked to diabetic retinopathy codes under E11.3-, and a team launching a therapy for one indication needs to know where that population concentrates.

HCPCS Level II drug codes, such as J0178 for aflibercept, add a further layer by showing which agent the provider actually administers.

The level of diagnostic detail matters as much as the pairing. AHRQ’s Clinical Classifications Software Refined groups more than 70,000 ICD-10-CM diagnosis codes into over 530 clinical categories. Those categories work well for sizing a broad therapeutic area. They work poorly when a product’s indication sits at the level of a single code, because a CCSR category can blend several conditions with very different treatment paths.

Medical device companies and specialty diagnostics labs in particular need individual ICD-10 codes, since their indications are written that way.

Coding rules also shape which diagnoses appear at all. The ICD-10-CM Official Guidelines for Coding and Reporting instruct outpatient coders not to code diagnoses documented as probable, suspected, or rule-out, and to code the presenting signs and symptoms instead.

So a physician working up a suspected condition in a clinic often bills symptom codes and diagnostic test codes, and the confirmed diagnosis only appears later, sometimes under a different provider.

For a diagnostics lab, this changes the targeting logic entirely. The physicians ordering the most diagnostic workups are often better found through symptom codes and test CPT codes than through confirmed diagnosis codes.

Billed Activity Shows Which Specialties Actually Perform a Procedure

Specialty taxonomy is self-reported when a provider enrolls for an NPI, and it records training, often from years earlier. Claims record what each specialty actually does, which frequently differs from what the taxonomy implies.

Researchers at the National Cancer Institute and CMS compared CMS provider files against the AMA Physician Masterfile and found that NPPES taxonomy codes identified only 59.8% of surgical oncologists, and most of those the CMS files missed were coded simply as surgeons.

The same researchers noted that cancer studies often work around the problem by classifying physicians on what they bill, treating a provider with chemotherapy claims as a medical oncologist regardless of the specialty code on file.

Many procedures cross specialty lines. Peripheral vascular interventions are performed by interventional cardiologists, vascular surgeons, and interventional radiologists, each with different referral patterns and different purchasing relationships. A device team that targets only one of those specialties because of an assumption about who “owns” the procedure leaves a large share of the volume unaddressed. Mapping billed CPT volume by specialty shows how a procedure actually distributes across specialties, and the answer often reorders a launch plan.

The same logic applies within a specialty. Two providers listed under internal medicine might run a hospitalist practice and an outpatient practice heavily focused on diabetes management.

Two orthopedic surgeons might split between sports medicine and joint replacement. Healthcare claims analysis makes those distinctions visible because subspecialty focus shows up in procedure and diagnosis mix even when the taxonomy code stays generic.

Claims also surface clinicians that specialty-based lists tend to miss. Nurse practitioners and physician assistants bill under their own NPIs for many office-based services, including injections, minor procedures, and chronic disease management.

In some practices they carry a substantial share of the patient volume for a given condition. A target list filtered to physicians alone can look complete while omitting the clinicians who see those patients most often.

What Claims Data Can and Cannot Tell Commercial Teams

Medical claims data is a record of what was billed. That definition explains both its strength and its blind spots, and teams that understand the blind spots make better use of the strength. Three limitations matter most for commercial work.

Coverage Gaps Can Make an Active Provider Look Quiet

Care that never generates a claim never appears in claims data. According to the CDC’s National Center for Health Statistics, 27.2 million Americans of all ages, or 8.2%, had no health insurance in 2024.

Cash-pay services, many aesthetic procedures, and care delivered through direct-pay arrangements fall outside the record as well. Providers whose practices lean on those patients will look smaller in claims than they are.

Payor coverage creates a larger distortion. According to KFF, 55% of eligible Medicare beneficiaries, or 35.2 million people, are enrolled in Medicare Advantage plans in 2026.

A dataset built only on traditional fee-for-service Medicare therefore misses most of the Medicare population, along with every commercially insured and Medicaid patient. A provider in a high Medicare Advantage market can appear nearly inactive in fee-for-service files while running a busy practice.

Diagnosis Codes Are Less Complete Outside the Hospital

Procedure codes tend to be reliable because payment depends on them. Diagnosis codes are less consistent, particularly in outpatient settings.

A study comparing Medicare claims with laboratory-confirmed Clostridioides difficile cases found that claims diagnosis codes captured 81% of infections overall, with sensitivity reaching 93% for hospitalized patients and falling to 56.2% for patients who were not hospitalized.

Combined with the outpatient rule against coding suspected conditions, this means diagnosis-based counts for clinic-managed conditions usually understate the true patient population. They remain useful for ranking providers against one another, since the undercount affects similar providers similarly, but they should not be read as exact prevalence.

Reporting Lag and Missing Motive Limit What Recent Claims Can Prove

Claims also arrive on a delay. Medicare allows providers to file claims up to 12 months after the date of service, so the most recent quarters in any claims dataset are incomplete when first reported and fill in over time. A sudden drop in a provider’s latest quarter is often a reporting lag rather than a change in practice.

Even complete, timely claims describe what a provider did without explaining why. They do not capture a physician’s clinical reasoning, product preferences, contracting constraints, or openness to a new approach.

A high-volume surgeon may be locked into a competitor’s system through a hospital agreement, and nothing on a claim will say so. Claims intelligence tells a team where to look and what to ask, and the rep’s first conversation supplies the context the claim leaves out.

How Alpha Sophia Helps Commercial Teams Explore HCP Claims Data

The limitations above point to what a useful claims platform has to provide. It needs all-payor coverage so providers are not misread through a single payor’s lens, code-level detail so indications can be matched precisely, and a way to turn that detail into lists and territories without a data engineering team in the middle.

Alpha Sophia is built around those requirements, as a reference layer that commercial teams query and enrich from while keeping ownership of interpretation, targeting decisions, and CRM workflows.

All-Payor Claims Anchored to the NPI Keep Provider Views Complete

Alpha Sophia provides all-payor US medical claims covering more than 3.9 million providers across commercial plans, Medicare (including Medicare Advantage), and Medicaid.

Every record is anchored to the provider’s NPI, which keeps a physician’s activity attached to the individual even when that physician practices across several sites. Procedure data covers CPT and HCPCS Level II codes, and diagnosis data is available at the individual ICD-10 code level as well as through CCSR categories, so teams can size a therapeutic area broadly and then narrow to a single indication.

Provider profiles layer that activity with taxonomy, organizational affiliations, open payments, publications, and clinical trial participation.

Code-Level Filters and Cohort Analysis Turn Claims Into Working Lists

Commercial teams filter providers by any combination of CPT and HCPCS codes, ICD-10 diagnoses, specialty, location, and volume, then rank the results by the procedure or diagnosis that matters to the product.

Cohort analysis lets teams compare groups of providers side by side, for example high-growth versus flat-volume surgeons in the same specialty, to see which segment warrants investment.

Territory Manager maps claims-based opportunity onto geography, with driving distance in miles, route planning, and heat maps for designing territories around clinical volume. Results export to Excel, sync natively with HubSpot, and reach Salesforce and other systems through export or the Alpha Sophia API.

Plain-English Questions Work Through the Assistant or Your Own AI Tools

Teams that prefer asking questions over building filters can use the in-app Alpha Sophia Assistant, which converts a plain-English request into a filtered provider list and shows the criteria it applied so the result can be checked and refined.

Teams already working in Claude, ChatGPT, or Cursor can connect those tools to Alpha Sophia through the Model Context Protocol and query the same claims data without leaving their workspace.

For repeatable processes, autonomous workflows can size a market, build a target list, and pass the output to connected tools such as a CRM.

In every mode, the answers come from Alpha Sophia’s claims data rather than from a model’s training set, a point covered in more detail in Alpha Sophia’s piece on agentic AI for HCP engagement. That matters because a general-purpose model without access to provider records generates plausible answers instead of retrieving real ones.

Conclusion

HCP claims data turns a provider directory into a description of clinical practice. Procedure and diagnosis codes together show what a clinician does, how often, and for which patients, while place-of-service and geographic patterns locate that activity and the organizations that control purchasing around it.

The same data has firm limits. Coverage gaps, outpatient coding rules, and reporting lag all shape what appears in the record, and none of it explains a physician’s reasoning.

Teams that account for those limits use claims to decide where field time goes and which questions to bring into each account, and they treat sudden changes in the data as prompts for verification before acting on them.

Alpha Sophia gives commercial teams all-payor, NPI-anchored claims data with code-level filtering, cohort analysis, territory design, and AI-assisted access. Book a demo to see how your target procedures and diagnoses map across the providers in your market.

FAQs

What is HCP claims data?
HCP claims data is the record of services that healthcare providers bill to insurers, linked to each provider’s National Provider Identifier. Each claim includes procedure codes, diagnosis codes, dates, and place of service. Aggregated over time, it shows what individual providers actually do in clinical practice.

What can claims data reveal about HCP clinical activity?
Claims data reveals which procedures a provider performs, how often, for how many patients, and for which diagnoses. It also shows where care is delivered, which payors cover the patients, and whether a provider’s volume is growing or declining. Together these signals show a provider’s real clinical focus beyond their listed specialty.

How do pharma companies use claims data?
Pharma companies use claims data to identify providers who treat patients with specific diagnoses relevant to a therapy’s indication. Diagnosis and procedure patterns help define the prescriber audience, prioritize field outreach, and support launch planning and market access work. Claims also help medical affairs teams find active clinicians for engagement and clinical trial recruitment.

How do MedTech companies use claims data?
MedTech companies use claims data to find surgeons and proceduralists who perform the procedures their devices support, ranked by volume. Place-of-service data shows whether those procedures happen in hospitals, outpatient departments, or ambulatory surgery centers, which shapes the sales approach. Geographic claims patterns also guide territory design and distributor oversight.

What is the difference between claims data and clinical data?
Claims data comes from billing records and captures what services were provided, for which coded diagnoses, and by whom. Clinical data comes from medical records and includes lab values, imaging results, physician notes, and clinical reasoning that claims do not contain. Claims offer broad coverage across providers, while clinical data offers depth on individual patients.

How does Alpha Sophia provide access to HCP claims data?
Alpha Sophia provides all-payor US medical claims data for more than 3.9 million providers, anchored to each provider’s NPI. Teams can explore it through filters and cohort analysis in the platform, through the in-app AI Assistant, or by connecting their own AI tools over the Model Context Protocol. Data can be exported, synced to HubSpot, or accessed through the Alpha Sophia API.