Market access decisions are ultimately evidence decisions. Even the best value narrative will be challenged by empirical questions: How many patients qualify under realistic criteria? What is the true comparator in routine practice? How quickly do patients progress through lines of therapy? What does resource use look like in the payer’s budget window? And after launch, why are claims being denied, where are delays occurring, and which segments are most affected?
10.1 Claims, EHR, and Registry Data: What They Are and How They’re Used
Most market access analytics relies on three clinical data families: claims, electronic health records (EHR), and registries. They are complementary. Claims are strongest for paid utilization and costs. EHR is strongest for clinical detail and pathway reality. Registries are strongest for purpose-built follow-up and long-term outcomes. Choosing the wrong family is a common reason payer-facing evidence fails under review.
Claims data: administrative records generated when payers adjudicate payment. Claims typically include coded diagnoses, procedures, drug identifiers, dates of service, provider and facility identifiers, and financial fields such as billed, allowed, and paid amounts. Claims can be pharmacy claims (dispensed drugs) or medical claims (procedures, facility services, and clinician-administered products in certain billing models).
Claims are foundational for access because it shows what the system paid for at scale. That makes it highly useful for questions about utilization, persistence, switching, site of care, and total cost of care.
Use claims data when your question is primarily about affordability or utilization in the payer’s world. Common access applications include:
- Population sizing: diagnosed prevalence, treated prevalence, and baseline therapy patterns by payer segment.
- Comparator realism: what patients actually receive and how quickly they change therapy.
- Budget impact inputs: baseline resource use and cost distributions anchored in paid amounts.
- Post-launch access monitoring: inflection points in paid utilization after policy changes or competitor entry.
Claims also have predictable limitations. It often lacks clinical nuance such as severity, lab values, biomarkers, imaging results, and documented reasons for switching or discontinuation. Coding can be inconsistent across providers and settings. Many outcomes are not explicit and must be inferred from proxies, which introduces measurement error. When claims are used for outcomes, label endpoints as proxies, document definitions clearly, and validate against a subset with clinical detail when possible.
EHR data: clinical records generated at the point of care, typically including diagnoses, medications prescribed, lab values, vitals, and other clinical observations. EHR is powerful for access because it can define eligibility and response in ways claims cannot, especially for biomarker-based and severity-based criteria.
EHR is most valuable for market access when payers will question generalizability or will impose criteria that depend on clinical variables. Typical applications include:
- Eligibility definition: sizing biomarker-defined or severity-defined segments aligned to likely reimbursed criteria.
- Pathway mapping: describing how patients are diagnosed, referred, and treated in routine practice.
- Real-world outcomes: evaluating clinical endpoints recorded in care settings, including response markers where available.
EHR limitations must be managed explicitly. Coverage is rarely population-wide and is often concentrated in certain health systems or regions. Documentation is variable and missingness can be high. Longitudinal follow-up can break when patients change providers. For payer-facing use, include a representativeness statement, define missingness handling, and be transparent about which variables are structured versus derived from unstructured notes.
Registries: structured collections of disease, therapy, or procedure data, typically designed with a defined dataset, follow-up schedule, and governance model. Registries are valuable when long-term outcomes are central to access durability or when conditional reimbursement requires ongoing evidence generation.
Use registries when the payer’s binding uncertainty is durability, rare disease natural history, or the need to validate appropriate use over time. Common registry use cases include:
- Natural history context: baseline outcomes under current care when trials are small or comparators are weak.
- Durability evidence: follow-up beyond trial windows to support reassessment and reduce uncertainty penalties.
- Appropriate-use validation: demonstrating that funded criteria are applied as intended and that utilization does not drift.
Registries also carry execution risk. They take time to stand up, require sustained participation, and can suffer from selection bias. If a registry is used to support reimbursement commitments, endpoints, adjudication rules, reporting timelines, and data ownership must be defined upfront and must be feasible for sites and payers.
A practical selection rule is to match data to the decision lever. If the lever is payer spend and paid utilization, start with claims. If the lever is eligibility criteria and clinical appropriateness, add EHR or registry data. If the lever is durability uncertainty, prioritize registry-like follow-up. When possible, use linked datasets to combine clinical detail with paid-cost realism, but only when linkage quality is documented and stable.
10.2 Sales, Channel, and Distribution Data
Market access outcomes are visible not only in payer policy but also in channel behavior. Channel and distribution data helps you translate “covered” into “started,” diagnose friction between prescription and therapy receipt, and manage contract operations and gross-to-net. It is often the fastest signal that access is breaking down, because denials, rework, and abandonment show up here before they show up in revenue.
Syndicated sales and volume data: aggregated measures of product movement derived from wholesalers, pharmacies, and other intermediaries. These datasets help benchmark competitive trends and detect inflection points that may signal policy shifts, tender outcomes, or site-of-care changes.
Specialty pharmacy and hub workflow data: operational data describing referral intake, benefit verification, prior authorization steps, time to dispense, abandonment, and reasons for delay. These data are especially useful for quantifying access friction and for prioritizing fixes by payer segment and by step in the journey.
Distribution and inventory signals: shipment and channel mix data across wholesalers, specialty distributors, hospitals, and direct channels. For clinician-administered products, ordering patterns can reveal adoption and can also reveal cash-flow concerns when reimbursement timing is uncertain.
Use channel data to answer questions that payer policy alone cannot resolve:
- Journey diagnosis: where patients drop out between prescription, authorization, and therapy start.
- Denial intelligence: which payer segments drive denials and what documentation issues are most frequent.
- Operational forecasting: how long policy wins take to translate into starts and how lag varies by segment.
Interpret channel data with care. Shipment is not the same as consumption. Hub data may not cover all patients if multiple support pathways exist, and definitions vary by service vendor. Syndicated sales often lags and does not explain net economics. Use channel data as an early warning system, then triangulate with claims and policy data to confirm root causes and quantify the payer segments driving variance.
To make channel data actionable, standardize a small set of operational definitions across vendors. Differences in how “start,” “approval,” or “abandonment” are defined can create false signals and misdirect payer engagement.
- Time to therapy: days from prescription to dispense or administration.
- Denial rate: share of cases denied, with reasons grouped into a stable taxonomy.
- Rework rate: share of cases requiring resubmission due to documentation or criteria mismatch.
- Abandonment: share of approved cases not initiated, separated into cost-related and other drivers.
- Channel mix: distribution of volume across specialty, retail, hospital, and direct channels.
10.3 Pricing and Reimbursement Databases
Pricing and reimbursement data translates your value story into the economic reality that determines access. Market access teams use pricing and reimbursement databases to set corridors, manage global spillovers, prepare negotiation positions, and ensure provider economics are viable. The most important discipline is separating four concepts: list prices, net terms, reimbursement rates, and patient cost exposure.
Price and reimbursement debates fail when teams mix perspectives. A payer thinks in net budget impact, a provider thinks in reimbursement and cash flow, and a patient thinks in out-of-pocket exposure. Good data practice keeps these views separate, then links them explicitly in the model.
List price databases: sources that track published prices used as reference points in many markets. List prices matter for international reference pricing, public transparency, and certain payment methodologies, even when they are not the realized net outcome.
Net price signals: internal and external indicators of discounts, rebates, tenders, and other concessions. In many systems the true net is confidential, so teams use directional signals. These signals are useful for planning and for corridor setting, but they require governance and careful labeling to avoid treating estimates as facts.
Coverage and formulary policy databases: repositories that track coverage status, restrictions, tiering, renewal rules, and site-of-care policies. These datasets support payer segmentation and early warning monitoring because access tightens first through criteria changes and friction rules.
Payment and coding references: sources that track billing codes, payment schedules, and policy rules that determine whether providers can bill and get paid reliably. These data are critical for clinician-administered products and for any therapy where site of care drives cost and adoption behavior.
Pricing and reimbursement data supports a focused set of access decisions:
- Corridor setting: triangulating precedent, affordability constraints, and spillover risks.
- Provider economics: modeling margin and cash-flow implications by site of care and payment rule.
- Policy monitoring: detecting restriction drift and identifying which accounts require engagement.
Because pricing and policy data can be sensitive, define internal guardrails: who can access directional net estimates, what can be shared externally, and what requires review. Maintain a short, time-stamped pricing and policy fact book per priority market, separating public facts from internal estimates and labeling assumptions clearly.
10.4 Patient-Level and RWE Data Providers and Partnerships
Real-world evidence has become a core input to reimbursement, reassessment, and contracting, especially when uncertainty at launch is material. Market access teams therefore need a working understanding of the provider ecosystem that supplies patient-level data and the partnership models that enable durable evidence generation.
At a high level, providers fall into four common categories. Claims and multi-source aggregators: large-scale datasets used for treatment patterns, cost, and utilization. EHR-focused and therapeutic-area datasets: deep clinical detail, often concentrated in specific care settings. Research networks and federated platforms: access to cohorts across institutions with standardized query models. Linkage and privacy-enabling services: capabilities that support patient matching, tokenization, de-identification, and privacy-preserving linkage between claims, EHR, labs, and other sources so analyses can combine clinical detail with cost realism.
In practice, many access organizations evaluate well-known providers in these categories, such as IQVIA, Optum, Merative (including MarketScan), Komodo Health, Flatiron Health in oncology, and network platforms such as TriNetX. The specific best choice depends on the question, the market, and the level of payer scrutiny, so focus on capabilities: representativeness, clinical depth, linkage quality, timeliness, and auditability.
Partnerships can be as important as vendors. Payers, integrated delivery systems, academic consortia, specialty networks, and disease foundations may control data that is uniquely relevant to access. Rare disease foundations often sponsor registries that become central to durability narratives. Centers of excellence can enable structured outcomes tracking for high-cost therapies where routine claims do not capture the endpoint. Design partnerships around a specific payer uncertainty and a measurement plan that defines the endpoint, the data capture workflow, reporting cadence, and how results will be used in coverage discussions.
Different evidence purposes require different data arrangements. Descriptive analytics: baseline patterns and burden to support budget impact and pathway mapping. Comparative effectiveness: relative outcomes in practice, requiring careful cohort design and confounding control. Operational outcomes tracking: measurement for contracts and reassessments, requiring clear endpoint definitions, capture feasibility, and dispute handling.
Provider selection should explicitly test three risks. Representativeness risk: conclusions do not generalize to routine care. Measurement validity risk: endpoints rely on weak proxies. Operational feasibility risk: required data cannot be captured reliably in the real pathway. If any of these risks are high, payers will respond with skepticism, narrower criteria, or a demand for safeguards.
10.5 Evaluating and Selecting Data Providers for Your Use Case
Choosing data providers is a strategic decision because it affects credibility, cycle time, and cost across the portfolio. The wrong choice delays work and produces analyses that cannot withstand payer review. The right choice creates repeatable analytics, faster decision cycles, and better post-launch control.
Start with a tight use case definition. Use case: the specific decision you need to influence and the minimum data required to change that decision. Examples include reimbursed population sizing, budget impact under alternative criteria, real-world outcomes in a restricted subgroup, denial-driver diagnosis, or outcomes-linked measurement. For each use case, define the required variables, the required follow-up duration, the required segmentation, and the tolerance for uncertainty.
Then evaluate candidate providers across eight practical dimensions:
- Coverage: the dataset includes the geographies, payer segments, and care settings that matter for the decision.
- Eligibility fidelity: the dataset can identify the target reimbursed population with the variables required.
- Endpoint measurability: the dataset captures the outcome in a credible way, not only through fragile proxies.
- Longitudinality: patients can be followed long enough to answer the payer’s durability or persistence question.
- Timeliness: data lag and refresh cadence fit the decision window and post-launch monitoring needs.
- Linkage quality: linkage methods and match rates are documented and stable over time.
- Data governance: privacy protections, permissible uses, and auditability align with internal policies and external requirements.
- Total cost of ownership: licensing, refresh, analytic support, and internal data management effort are understood.
Convert evaluation into a fast selection process. Step 1: write a one-page requirements summary with must-haves and kill criteria. Step 2: run a feasibility test using de-identified cohort queries or summary statistics to confirm that key variables exist and are populated at usable rates. Step 3: review the data dictionary and provenance documentation, focusing on how fields are created, transformed, and updated. Step 4: agree up front on deliverables, publication rights if relevant, refresh cadence, and the level of analytic support the provider will supply.
Kill criteria should be explicit. If you cannot reliably define the reimbursed population, you cannot support criteria discussions. If the endpoint cannot be measured credibly, you cannot support comparative claims. If follow-up is too short, you cannot support durability narratives. Kill criteria protect you from sunk-cost bias and prevent months of model building on an unstable foundation.
Decide early whether you need global consistency or local optimization. Global teams benefit from a limited number of standardized datasets and methods that enable comparability and reduce rework. Local teams often need local datasets that reflect local coding and practice patterns. A common effective approach is a hybrid: a global backbone dataset for benchmarking and corridor work, plus local augmentations where comparator and pathway differences materially change reimbursement arguments.
For any dataset used in payer-facing work, require a minimum documentation pack: a data dictionary, a cohort definition, a refresh cadence, and a brief limitations statement. This prevents over-claiming and reduces rework during HTA review and payer medical policy challenges.
Finally, treat data as an operational asset. Assign a data owner, define refresh cadence, implement routine quality checks, and maintain a short “assumptions and limitations” page for each dataset used in payer-facing work. Strong data that cannot be explained transparently performs like weak data in a payer committee. The goal is not to accumulate datasets; it is to build a small, trusted evidence base that supports reimbursement decisions, accelerates negotiations, and improves post-launch performance management.
