Can BPC-157 Consumption Be Reliably Tracked in Real-World Datasets Despite Having No National Drug Code in 2026?
BPC-157 consumption can be partially tracked in real-world datasets in 2026, but only through unstructured clinical-note surveillance — not through standard prescription or claims channels. Because BPC-157 carries no National Drug Code and no approved label, every conventional pharmacy-claims query returns zero results. LLM-curated notes can surface confirmed users, but only those who disclosed use to a documenting clinician.
Why Does the Absence of an NDC Make BPC-157 Invisible to Standard Dataset Queries in 2026?
The NDC is the eleven-digit identifier linking every FDA-approved drug to its dispensing record in pharmacy claims, EHR medication lists, and payer adjudication systems. BPC-157 has never received FDA approval, so no NDC exists. Any dataset query filtering on NDC returns a structural zero for BPC-157, covering virtually all PBM claims databases and most structured EHR medication fields.
This is not a data-quality problem; it is a structural feature of how drug identification works in US health data infrastructure. The FDA's 2021 guidance on real-world data from EHRs explicitly notes that compounded preparations lacking an NDC require proxy measures or multi-component definitions to be identified in structured datasets.
BPC-157 sits at the extreme end of that problem: it is compounded, unapproved, and predominantly sourced through grey-market channels that generate no billing record. 503B outsourcing facilities are the only compounding pathway that generates any NDC-adjacent record, as they must report products to the FDA NDC Directory under a distinct marketing category. However, BPC-157 was removed from the FDA's 503A Category 2 bulk drug substances list in April 2026, constraining licensed compounding pathways.
As of late 2026, the dominant sourcing pathway for BPC-157 remains grey-market, generating no structured dispensing record in the US health data ecosystem. Any dataset claiming to measure BPC-157 consumption through NDC-based queries is measuring nothing. The compound is structurally invisible to the most widely used pharmacoepidemiological tools.
What Methodology Did the 2026 Venkatakrishnan Preprint Use to Identify 1,039 Confirmed Users?
Venkatakrishnan and colleagues screened approximately 15 million de-identified clinical records from the Mayo Clinic Platform in a 2026 preprint. A keyword search across unstructured notes flagged 1,536 records. LLM curation plus physician validation then confirmed actual use in 1,039 patients — a 67.6% confirmation rate — making this the first published real-world consumption dataset for BPC-157.
The keyword-plus-LLM pipeline is a deliberate workaround for the NDC gap. Because no structured medication field will ever contain BPC-157 under current regulatory conditions, the only machine-readable signal is free-text documentation in clinical notes — progress notes, intake forms, and patient-reported medication lists transcribed by clinicians. The LLM layer was necessary to distinguish confirmed use from incidental mentions, differential diagnoses, or clinician warnings about the compound.
Physician validation was the third stage, applied to resolve ambiguous LLM outputs. The 67.6% confirmation rate from flagged to confirmed records implies that roughly one-third of clinical notes mentioning BPC-157 did not constitute confirmed consumption. These were likely clinical discussions, adverse-event warnings, or patient inquiries rather than documented use.
The dataset produced a 33-fold increase in newly documented quarterly users between 2020 and 2026, with pain management as the most frequently documented indication. These figures represent a lower bound on actual consumption: the method captures only patients who disclosed use to a clinician who documented it in a note retained in the Mayo Clinic Platform's de-identified corpus.
How Do the Available Tracking Methods Compare for BPC-157 in Real-World Datasets?
Six tracking approaches are available for BPC-157 in real-world datasets as of 2026. LLM-curated clinical note surveillance yields the highest confirmed-user count but requires a large de-identified EHR corpus. NDC-based pharmacy claims return a structural zero. ICD-10 proxy codes, patient registries, social media pharmacovigilance, and ingredient-level compounding billing each carry distinct capture-versus-verification tradeoffs summarised below.
| Tracking Method | Data Source | BPC-157 Capture | Verification Level | Coverage Limitation |
|---|---|---|---|---|
| LLM-curated clinical note surveillance | De-identified EHR free-text (e.g., Mayo Clinic Platform) | 1,039 confirmed users from ~15M records | LLM + physician validation | Captures only patients who disclosed use to a documenting clinician |
| NDC-based pharmacy claims query | PBM claims, Medicaid/Medicare, retail pharmacy | Structural zero — no NDC exists for BPC-157 |
N/A | Completely blind to unapproved, grey-market, and compounded compounds without NDC |
| ICD-10 diagnosis-code proxy | Claims databases, EHR structured fields | Indeterminate — codes capture indication, not compound | Low specificity | Cannot distinguish BPC-157 from any other treatment for the same diagnosis |
| Patient-reported outcome registry | Survey platforms, self-experimentation trackers | Variable; self-selected population | Self-reported, unverified | Severe selection bias; no denominator; no clinical corroboration |
| Social media pharmacovigilance | Reddit, forums, X (Twitter) | High volume of mentions; no confirmed-use filter | None — anecdotal | Cannot distinguish use from discussion; no adverse-event causation |
| Ingredient-level compounding billing | 503B outsourcing facility NDC Directory reports | Zero — BPC-157 not on 503A/503B list as of late 2026 |
Regulatory filing | Requires active compounding authorization; not applicable under current regulatory status |
What Is the Capture-Floor Problem and Why Does It Structurally Undercount BPC-157 Use?
The capture-floor problem is the systematic gap between actual consumption and any measurable signal in health data. Grey-market sourcing, the absence of any billing record, and voluntary patient disclosure each suppress the detectable signal. Even LLM-surveillance counts only users whose clinician documented the compound by name. Self-administering users with no clinical contact are invisible to every available method.
The 33-fold rise documented by Venkatakrishnan and colleagues is a lower-bound trajectory, not a complete consumption curve. The Reuters analysis that cited this preprint described it as a rise in "confirmed users" — a precise term that implicitly acknowledges the undercounting floor. Protocol designers working with self-experimenters should treat any published prevalence figure for BPC-157 as a minimum, not a census.
Three structural factors drive the capture floor downward. First, the primary sourcing pathway — grey-market research chemical vendors — generates no billing record, no prescription, and no dispensing event in any regulated data system. Second, many users actively avoid clinical disclosure due to the compound's unapproved status and WADA prohibition.
Third, even when disclosure occurs, documentation depends on individual clinician behaviour: a note reading "patient reports using a peptide for recovery" without naming BPC-157 contributes nothing to any keyword-based surveillance pipeline. The 67.6% confirmation rate from the Venkatakrishnan pipeline also implies that the 32.4% of flagged-but-unconfirmed records represent a methodological loss, not a population that did not use the compound.
What Does the NDC Gap Mean for Protocol Designers Building Consumption-Tracking Systems in 2026?
For protocol designers in 2026, the NDC gap means BPC-157 cannot be tracked through any standard pharmacy integration, claims feed, or structured medication-list field. A functional tracker must rely on self-reported compound entry with free-text name matching, timestamp logging at the dose level, and explicit disclosure prompts — no external data source will auto-populate or verify BPC-157 consumption.
The practical design implication is that BPC-157 tracking requires a different data architecture than NDC-bearing compounds. For approved drugs, a tracker can pull dispensing events from a PBM feed or structured EHR medication list and use those as ground-truth exposure records. For BPC-157, the only ground-truth source is the user's own input.
Tracker design must therefore prioritise low-friction dose logging and compound name disambiguation. The compound appears in records as "BPC157," "BPC 157," "body protection compound," and "Wolverine peptide" — four distinct strings that a naive text-match will treat as separate entities. Structured adverse-event capture should mirror the LLM-curated fields used in the Venkatakrishnan pipeline to enable future cross-dataset comparison.
Regulatory status monitoring is a second design requirement specific to BPC-157. The compound's legal sourcing pathway is actively in flux — April 2026 removal from the 503A Category 2 list, the July 2026 PCAC 8-6 vote in favour of relisting, and pending rulemaking all create a timeline where compounding status could change within the tracker's operational window. A tracker that does not surface regulatory status changes to users is providing an incomplete protocol context.
How Should the April 2026 FDA Removal and July 2026 PCAC Vote Be Integrated Into a BPC-157 Tracking Protocol?
The April 2026 FDA removal of BPC-157 from the 503A Category 2 list and the July 2026 PCAC 8-6 vote in favour of relisting are two distinct regulatory events a tracking protocol must handle separately. The removal changed the legal sourcing pathway. The PCAC vote is non-binding and does not restore authorization. Conflating them misrepresents the compound's current legal status.
For consumption tracking, the April 2026 removal has a direct data implication: any BPC-157 use documented after that date is sourced from grey-market channels or from pre-removal compounding stock, not from licensed pharmacies operating under the 503A framework. This sourcing shift matters for purity and concentration reliability — two variables that affect dose-response interpretation in any self-experimentation dataset.
The PCAC vote creates a forward-looking tracking requirement. If final rulemaking results in 503A relisting, licensed compounding pharmacies could begin dispensing BPC-157 under individual prescriptions. That transition would, for the first time, generate ingredient-level billing records that could partially close the NDC gap for users accessing the compound through licensed channels.
A well-designed tracker should be architected to incorporate that data source when it becomes available, rather than requiring a full rebuild. Until rulemaking completes, the tracking architecture remains entirely dependent on self-report. Protocol designers should document the regulatory status at the time of each tracking record so that sourcing-pathway shifts can be analysed as a variable in any longitudinal dataset. What Did the July 2026 FDA PCAC Review Conclude About BPC-157's Biopharmaceutical Data Gaps and 503A Compounding Eligibility? Why Did FDA Scientists Recommend Against Adding TB-500, BPC-157, and MOTS-C to the Compounding Greenlist in July 2026? What Do the Human Adverse Event Reports for BPC-157 Actually Show — and How Should Practitioners Interpret the FAERS Record in 2026?