Structured wound care data is wound documentation captured as discrete, coded, consistently defined fields (wound type, measurements, tissue type, product applied, care setting, dates) rather than free-text notes. That structure, not record volume, determines whether a dataset can support a defensible conclusion. A million narrative notes may answer fewer questions than fifty thousand well-formed episodes.
This article sets out what qualifies as structured, where wound documentation typically breaks down, what a usable longitudinal dataset requires, and what manufacturers, payers, and providers can and cannot conclude from it.
What Counts as Structured Wound Care Data?
At the field level, structured wound data means each clinical element lives in its own discrete, defined slot: etiology (DFU, VLU, pressure injury), anatomical location, length, width, and depth, calculated wound area, tissue type, exudate, product name and identifier, application date, and care setting. Each field has an allowed set of values or a defined unit. A query can then count, filter, and compare without anyone reading a note.
The distinction is easiest to see with one visit documented three ways.
- Free text: “Pt returns for f/u of R plantar foot wound, looks better, about the size of a quarter, applied graft, will see in a week.”
- Semi-structured: a template with a wound section containing a typed line such as “Wound 1: R plantar, 2.1 x 1.8 x 0.3 cm, granular, moderate drainage” and a free-text plan. The content is organized but still stored as a string.
- Structured: etiology = DFU; location = right plantar forefoot; length = 2.1 cm; width = 1.8 cm; depth = 0.3 cm; tissue = granulation; exudate = moderate; product = a single mapped identifier; application date = a date field; setting = hospital-based wound center.
Semi-structured data is the most common middle ground, and it can sometimes be parsed, but parsing introduces error that the structured version never incurs.
Where standard vocabularies help and where they stop
Standard coding gives structure a shared language. ICD-10-CM captures diagnosis, including diabetic foot ulcers with site and severity components, pressure ulcer stage, and venous ulcer specificity. HCPCS Q-codes identify skin substitute products for billing, and SNOMED CT offers clinical concepts for findings and procedures. These systems anchor etiology and billing events consistently across sites.
They stop short in predictable places. Most Q-codes are product-specific, but they do not carry the size, lot, or formulation detail an effectiveness analysis may need, and the not-otherwise-specified code (Q4100) can cover materially different products. Payment now blurs product distinctions further: since January 1, 2026, Medicare pays most skin substitutes as incident-to supplies at a single flat per-cm² rate, regardless of brand. ICD-10-CM does not record wound area, tissue composition, or change over time. Wound-bed detail and measurement trajectories must come from clinical fields that sit outside the code sets, which is why coding alone does not make a dataset analyzable.
Why Does Unstructured Wound Documentation Limit Real-World Evidence?
The failure modes are mundane and cumulative. Measurements sit inside narrative, so they cannot be extracted without natural language processing, which brings its own error rate. Copy-forward notes carry last week’s findings into this week’s record, creating apparent stability or progress that was never observed. Units drift between centimeters, millimeters, and inches. Product names arrive as brand, generic, abbreviation, or misspelling, so one product can appear under a dozen strings and be undercounted in every one.
Misclassification biases the cohort before analysis starts
Cohort definitions depend on etiology and healing status. When those are not recorded explicitly, analysts infer them from diagnosis codes, note language, or location. A foot wound in a patient with diabetes is not necessarily a DFU; a lower-leg wound in a patient with venous disease is not necessarily a VLU. Inferred etiology mixes populations with different expected healing trajectories, and the resulting bias is rarely visible in the final tables. Healing status has the same problem: “improving” in a narrative is not a measurement, and “healed” may be recorded late or not at all when a patient leaves care.
Claims data answers a different question
Claims are structured and broad, and they are valuable for what they capture: billed products, procedure and diagnosis codes, dates of service, rendering setting, and payer. They do not show wound size trajectory, tissue status, or why a clinician chose one product over another. A claim for a skin substitute application tells you an application occurred, not whether the wound was 2 cm² or 20 cm², or whether it was shrinking. Treating claims as outcome evidence is a common misconception, and so is the assumption that EHR data is analyzable simply because it is electronic.
Documentation gaps are widely discussed in the wound care literature, but completeness varies by site, template, and specialty. This article does not cite a prevalence figure for them, because any such number is only meaningful for the specific setting and study that produced it. Where a decision depends on a completeness rate, measure it in the dataset you intend to use.
What Does a Usable Structured Wound Dataset Look Like in Practice?
For comparative analysis, the minimum data elements are baseline wound area, wound age at baseline, etiology, relevant comorbidities, serial measurements at reasonably consistent intervals, product and application frequency, care setting, and an explicit outcome definition. Remove any of these and a specific class of question becomes unanswerable. Without wound age, for example, a product used on chronic wounds cannot be fairly compared with one used on newer ones.
Normalization
Normalization is the process of making records from different sources comparable. In wound data it typically involves four steps:
- Map every product string to a single product identifier.
- Standardize units and recompute area from length and width using one stated method.
- De-duplicate encounters that were recorded twice or carried forward without a new assessment.
- Decide how to handle missing visits, and document that decision.
A worked illustration
Suppose a DFU episode has a baseline (day 0) area of 4.0 cm² calculated as length times width (2.5 x 1.6 cm). Measurements follow at day 7 (3.6 cm²), day 14 (3.1 cm²), day 21 (2.9 cm²), and day 28 (2.4 cm²). Percent area reduction at 4 weeks is (4.0 minus 2.4) divided by 4.0, or 40%.
The denominator and exclusions matter as much as the arithmetic. If the analysis includes only episodes with a measurement within a window around day 28, say days 25 to 31, then episodes with no measurement in that window drop out, and those patients may differ systematically from those who stayed in care. If the analysis includes all episodes with a baseline measurement, the missing ones must be imputed or counted as non-assessed. Either choice is defensible, provided it is stated.
Healing endpoints also vary by source. Definitions of complete healing, such as the required duration of confirmed closure, and the choice of percent area reduction thresholds differ across guidelines and trials. The FDA’s 2006 chronic wound guidance, for example, defines complete closure as full re-epithelialization without drainage or dressing requirements, confirmed at two consecutive visits two weeks apart, and the Wound Care Collaborative Community has proposed an addendum to that guidance. Whichever reference you adopt, state its definition in the methods.
Which Questions Can Structured Wound Care Data Answer for Manufacturers, Payers, and Providers?
Manufacturers
Post-market clinical follow-up under EU MDR requires manufacturers to proactively collect and evaluate clinical data on their devices as part of post-market surveillance; check the regulation’s current text and your notified body’s expectations for the exact requirements. Structured real-world data can feed that process by showing how a product is used and what is recorded after use. Beyond PMCF, utilization by wound type and setting shows where a product is actually applied, and co-use patterns show which products appear alongside or in place of yours. The competitor that shows up in the data is often not the one in the strategy deck.
Payers and policy staff
Skin substitute coverage has been shaped by LCDs and payment policy that keep changing. The DFU and VLU skin substitute LCDs scheduled to take effect January 1, 2026 were withdrawn on December 24, 2025, leaving the legacy Novitas, First Coast, and CGS policies in place and other MACs deciding coverage case by case. Meanwhile, the CY 2026 Physician Fee Schedule moved most products to a flat per-cm² payment, and the CY 2027 proposed rule builds on that approach (see Dr. Caroline Fife’s review of the proposed rule). Confirm current status before building an argument on it. Structured data can test whether the clinical criteria in a policy match what is documented in practice, such as wound size and duration at first application. It can also show application counts per episode, the interval between applications, and the share of episodes with no recorded measurement trajectory, all of which are useful for utilization audits.
Providers and purchasers
Benchmarking across SNF, home visit, and hospital-based wound center settings shows how product selection and application patterns differ by site of care. Value analysis committees can use the same data to compare products within a wound type on a like-for-like basis, subject to the confounding limits discussed below.
How Intellicure Analytics organizes this
The Wound Care Industry Dashboard organizes real-world utilization data, drawn from structured documentation captured at the point of care in Intellicure’s wound care EHR, by product, wound type, and care setting, including product leaderboards and trend views. These charts describe what is used, where, and how that is changing; they are utilization views, not outcome claims. Product Ranking Reports go a step further, comparing healing rates and cost efficiency across products, and they carry the same confounding limits discussed below. Any causal effectiveness question should go through a defined study design.
What Are the Limits and Bias Risks of Structured Real-World Wound Data?
Structure removes one class of problem and leaves several others intact.
Confounding by indication. Clinicians choose products based on wound severity, perfusion, infection history, and payer rules. If a product is reserved for harder wounds, an unadjusted comparison will make it look worse than it is; if it is used on wounds already progressing, it will look better. Adjustment on baseline area, wound age, etiology, and comorbidities reduces this, but it cannot remove factors that were never recorded.
Selection and site bias. Contributing sites may not represent national practice. A dataset weighted toward hospital-based centers says less about SNF or home-visit use than its headline size suggests. Check composition before generalizing.
Censoring and missing follow-up. Censoring means an episode ends in observation before the outcome is known, because a patient transferred, was lost to follow-up, or changed sites. If censoring correlates with outcome, as it often does when patients who heal or deteriorate leave care, simple healing rates are biased. Time-to-event methods that handle censoring explicitly are the usual remedy.
Structured is not accurate. A populated field can hold a default value, a stale entry, or a measurement taken with an inconsistent technique. Ruler-based and image-based area estimates do not always agree, and a field that is 100% complete can still be wrong. Completeness and validity are separate checks.
Limits on causal claims. Real-world data supports comparative effectiveness work, surveillance, and utilization analysis. It does not replace randomized trials when the question is causal, particularly for a new indication, a regulatory claim, or a small expected effect that confounding could swamp. When the decision hinges on proving that a product caused a difference, or when key covariates are not captured routinely, a prospective study is the better tool. Real-world data is often most useful for deciding what to prospectively study.
How Should You Evaluate a Structured Wound Data Source Before Relying on It?
Put these questions to any vendor, registry, or internal dataset:
- Provenance: Where do records originate, and were fields captured at the point of care or derived afterward?
- Coverage: How many sites and which care settings? Which wound types?
- Product detail: Are CTPs and CAMPs identified at the product level or only by code group?
- Completeness by field: What share of episodes have baseline area, wound age, and serial measurements?
- Update cadence: How current is the data, and how are late entries handled?
- Privacy: What is the de-identification method, and how does it align with HIPAA?
- Validation: Is there evidence of accuracy checks, and are the methods published or open to review?
Prefer sources whose methods are peer-reviewed or fully documented, and ask for the limitations in writing. A vendor that cannot state its own gaps has not measured them.
Then match the dataset to the decision. Market sizing needs broad product and setting coverage but tolerates coarse clinical detail. PMCF needs longitudinal follow-up on your own device, with defined endpoints. Coverage evidence needs wound characteristics and application patterns that map onto policy criteria. Competitive intelligence needs product-level co-use and displacement, not outcomes. A dataset excellent for one of these can be unfit for another.
Matching Data Structure to the Decision You Need to Defend
The value of wound data depends on three things: how finely the fields are structured, how plainly the limitations are stated, and how well the dataset fits the specific decision in front of you. Volume alone does not substitute for any of them.
Review the Intellicure Analytics dashboard against your use case, or request a methodology discussion to test whether the fields, coverage, and granularity support the question you need to answer. Learn more about our services