Domain dossier

Potassium and creatinine monitoring during RAASi titration

RAASi therapy is limited in practice by the monitoring it requires: a drug that improves outcomes in heart failure and CKD also raises serum potassium, so titration is gated on repeat venous draws. The question this domain holds context on is what that monitoring actually costs, where it happens, and what has repeatedly gone wrong with the obvious shortcut — capillary sampling.

Read the thinness before reading the content. This is a seed dossier, assembled from evidence the repo had already retrieved in the 2026-08-25 Stage 0 pass, not from a research pass of its own. Two sections carry real sourced content; six are stamped unsearched. That is the honest state, and a later reader must not mistake this file for coverage — see §10.

This file carries no confidence tags. Every line below is context with a source attached. To affect a score, a line must be copied into a candidate's ## Claims as unverified via engine/scripts/append_research_claims.py and pass a checker there.

---

1. Condition & epidemiology

as_of: 2026-08-25 (bulk section) / 2026-08-26 (hyperkalemia-prevalence row added) / 2026-08-26, second pass (US HF/CKD comorbidity and RAASi-utilization rows added below) · sources searched 2026-08-26 (second pass): connectors literature broadened-term retries + literature-detail on PMID 33989713, 40972017, 35114384; WebFetch of niddk.nih.gov kidney-disease statistics page (2023 ADR-sourced figures, not 2024) · not searched: incidence (as opposed to point-prevalence), demographic concentration by race/age/region

2026-08-25 state, unchanged below this line: Not searched. Prevalence and incidence of CKD and chronic heart failure, the share on RAASi, the share whose titration is interrupted by hyperkalemia, and the demographic concentration of all of it are open.

Note for whoever works this: the pseudohyperkalemia sweep in §10 returned 111 matches and was read only for its capillary-sampling content. It is likely to carry epidemiology too.

2026-08-26 addition — hyperkalemia prevalence in a CKD stage 3+ Medicare-age cohort:

Verbatim: "The prevalence of hyperkalemia in the 3 validation data sets ranged from 2.6% (n = 1282 of 50 099; Minnesota) to 4.8% (n = 287 of 6011; Florida)." — PMID 30942845

Read this narrowly: it is hyperkalemia prevalence in a Mayo Clinic CKD stage ≥3 population with a paired ECG-and-potassium-draw design, not a US general-population or Medicare-wide estimate, and it says nothing about incidence, HF/CKD comorbidity share, or RAASi utilization.

USRDS 2024 Annual Data Report — methodology confirmed, no beneficiary count extracted: Fetched directly (not a search-engine summary) at https://usrds-adr-api.niddk.nih.gov/api/referenceTables?year=2024&referenceTable=Reference_Table_Methods, a "CKD Reference Table Methods" document. It confirms the methodology a prior candidate claim referenced, but a beneficiary count itself lives in a separate reference-table spreadsheet not retrieved here — so the count claimed in candidate idea-011 claim 10 ("beneficiary count from USRDS 2024 ADR") remains genuinely un-extracted, not merely un-cited.

"Reference Tables B.1–B.6 present estimated point prevalent (December 31) counts of all fee-for-service non-ESRD Medicare population, based on the 5% Medicare sample, for adults aged 18 and older... The disease conditions of CKD, heart failure (HF), and diabetes mellitus (DM)... were determined from claims in the reported year... Counts were multiplied by 20 to represent 100% of the Medicare population meeting the cohort definition." — USRDS 2024 ADR, CKD Reference Table Methods, Reference Table B

A web search surfaced a "15.3% CKD prevalence in Medicare FFS beneficiaries, 2021" and "19.4% in Medicare Advantage" figure attributed to this ADR, but that figure was never independently retrieved or quoted from the primary report by this pass — it is a search-engine synthesis, not a source of record, and is deliberately not carried into this dossier or into any candidate claim. A later reader should re-fetch the actual Reference Table B spreadsheet (not yet located at a stable URL) before treating that number as established.

2026-08-26, second pass — US HF/CKD comorbidity, resolved with a real PMID (candidate idea-011's prior pass could not re-identify its original "20-67%"/"~52%" web-search sources; this is a different, independently retrieved and resolvable finding, not a re-identification of those two papers, which remain unfound — see below):

Verbatim:

"Among 365 494 hospitalizations (age 72 ± 15 years, left ventricular ejection fraction [EF]: 43 ± 17%), median discharge eGFR was 51 ml/min/1.73 m2 ... 234,332 (64%) had eGFR <60 ml/min/1.73 m2, and 18,869 (5%) were on dialysis." — PMID 33989713

"'Triple therapy' with angiotensin-converting enzyme inhibitor/angiotensin receptor blocker/angiotensin receptor-neprilysin inhibitor + beta-blocker + mineralocorticoid receptor antagonist was used in 38%, 33%, 25%, 15%, 5%, and 3% for eGFR ≥90, 60 to 89, 45 to 59, 30 to 44, <30 ml/min/1.73 m2, and dialysis, respectively" — PMID 33989713

"At baseline, 643 (61.1%) were receiving GDMT [RASi only, n = 496; SGLT2i only, n = 25; both, n = 122; range 42.7% (USA) to 82.2% (Spain, Sweden)]." — PMID 40972017

Read this narrowly: PMID 33989713 is HF-hospitalization comorbidity (a sicker, hospitalized slice of the HF population, not all outpatient HF/CKD), and its "64%" figure is defined by discharge eGFR<60, not a formal CKD diagnosis code — a different operational definition than the original unconfirmed "20-67% comorbidity" claim, though it falls squarely inside that range and is far better sourced (n=365,494, a named US registry, resolvable PMID) than the range it replaces. PMID 40972017's GDMT figure combines RASi and SGLT2i together, not RAASi alone, and its US arm (42.7%) is the lowest of the countries reported, not "~52%, lower than Germany or France" as the original claim stated — consistent in direction (US utilization trails some European peers) but not the same number. The original two source papers behind candidate idea-011 claims 8-9 remain unidentified despite five further distinct search-term attempts this pass (queries logged in §10); this is now the second research pass to search for them specifically and come up empty on the exact papers, while finding adjacent, resolvable, arguably stronger-quality substitutes instead.

USRDS Medicare beneficiary count — retried, still not extracted from the 2024 ADR itself: The 2024 ADR's own report page (usrds-adr.niddk.nih.gov/2024/chronic-kidney-disease/1-ckd-in-the-general-population) renders as a JavaScript single-page application; a direct WebFetch of it returned only the page title, no report content — a different failure mode than the 2026-08-25 pass's API 404, but the same practical outcome (not retrievable by this pass's tools). The 2024 ADR's own citable journal article (PMID 40379355, American Journal of Kidney Diseases, 2025) carries no abstract in PubMed (literature-detail confirms this — common for ADR-summary editorials) and its publisher page returned HTTP 403 to a direct WebFetch. A different, directly-fetched federal page — NIDDK's general kidney-disease statistics page (niddk.nih.gov/health-information/health-statistics/kidney-disease, fetched successfully this pass) — attributes these figures to the 2023, not 2024, Annual Data Report: "CKD affects more than 1 in 7 U.S. adults—an estimated 35.5 million Americans" (CDC, 2023, not Medicare-specific); "Medicare spending for beneficiaries with CKD (not including ESKD) ages 66 or older was nearly $77 billion in 2021, representing 24.1% of Medicare spending in this age group" (USRDS 2023 ADR); "More than 808,000 people in the United States are living with ESKD" (USRDS 2023 ADR). None of these is the specific "beneficiary count meeting the CKD cohort definition" figure candidate idea-011 claim 10 originally sought, and none is drawn from the 2024 report specifically — so that exact figure stays genuinely un-extracted after two research passes, though real, quotable, directly-fetched federal-page content (just not the 2024-vintage, not-ESKD-excluded count) now exists in this dossier where none did before.

Gaps: RAASi utilization isolated from SGLT2i; incidence (as opposed to prevalence); demographic concentration by age/race/region; the exact 2024-vintage USRDS CKD beneficiary count (percentage or n) for the non-ESKD Medicare FFS population, as opposed to the adjacent 2023-vintage spending/mortality/ESKD-count figures now on record. The original two comorbidity/RAASi review papers behind candidate idea-011 claims 8-9 remain unidentified after two research passes.

2. Care pathway & clinical setting

as_of: 2026-08-25

Not searched. As of this date no source has been queried for this section. Where RAASi titration happens (cardiology clinic, nephrology clinic, primary care, pharmacist-led titration services), who draws the sample, what turnaround the decision needs, and what lab infrastructure is already in the room are all open.

The one thing the retrieved evidence bears on: §5's inpatient findings are hospital inpatient data, and this domain's setting is outpatient. Whether a 30% collection-failure rate in hospitalised adults transfers to an ambulatory clinic population is exactly the kind of question this section exists to answer, and it is unanswered.

Gaps: the entire section.

3. User groups & incentives

as_of: 2026-08-25

Not searched. As of this date no source has been queried for this section.

Recorded so the shape is visible: prescriber (titrating clinician), operator (whoever lances the finger), purchaser (clinic or health system), payer, and patient are five different parties with different incentives here, and a device that saves a patient a venipuncture may cost a clinic staff time. None of that is evidenced.

Gaps: the entire section.

4. Standard of care & clinical evidence

as_of: 2026-08-25 (capillary/venous rows) / 2026-08-26 (ECG-screening row added below) · sources searched 2026-08-25: PubMed via connectors literature / literature-detail, 8 PMIDs read at abstract depth · sources searched 2026-08-26: literature-detail on PMID 30942845 · not searched: guidelines (KDIGO, ACC/AHA/HFSA), CLSI documents, FDA Dual 510(k)/CLIA-Waiver guidance, any prospective/multi-site validation of ECG-based hyperkalemia screening published after 2019

Verbatim:

"Hemolysis rates were similar between arterial (8.6 %) and venous (10.6 %) samples but were substantially higher for capillary (17.7 %) samples." — PMID 40480412

"Paired samples were comparable for all the analytes tested with the exception of bicarbonate and potassium, which were significantly different among the capillary sample group" — PMID 33009571

"Capillary and venous results were clinically comparable for 16/18 analytes when specified HIL cutoffs were applied, although potassium and chloride showed limited interchangeability" — PMID 42446434

Note what these do and do not establish. All three are about capillary sampling as a method, on laboratory and blood-gas analysers — none of them tests a strip-and-meter device, and none of them tests on-strip hemolysis rejection. They characterise the problem the domain faces; they are not evidence about any product.

ECG-based (non-invasive) hyperkalemia screening — a second technology arm

Added 2026-08-26, for candidate idea-011. Distinct from the capillary-vs-venous literature above: this is about inferring hyperkalemia risk from ECG waveform morphology, without any blood draw at all, as a screening trigger for a confirmatory lab test rather than a replacement for one.

Verbatim:

"Using ECG leads I and II, the AUC of the deep-learning model was 0.883 (95% CI, 0.873-0.893) for Minnesota, 0.860 (95% CI, 0.837-0.883) for Florida, and 0.853 (95% CI, 0.830-0.877) for Arizona." — PMID 30942845

"Using a 90% sensitivity operating point, the sensitivity was 90.2% (95% CI, 88.4%-91.7%) and specificity was 63.2% (95% CI, 62.7%-63.6%) for Minnesota; the sensitivity was 91.3% (95% CI, 87.4%-94.3%) and specificity was 54.7% (95% CI, 53.4%-56.0%) for Florida." — PMID 30942845

Read this narrowly, the same discipline as the capillary rows above: this is a retrospective, single-health-system, algorithm-development-and-internal-validation study — three sites but all Mayo Clinic, all trained and tuned by the same group, no external (non-Mayo) validation, no prospective trial. The abstract's own conclusion says "Prospective studies are warranted," which this dossier has not found evidence were ever published or completed. At the 90%-sensitivity operating point, specificity is only ~55-63%, meaning a majority of ECG-flagged patients without true hyperkalemia would still need — and get — a confirmatory draw under this candidate's own intended-use framing; this is evidence for the screening-adjunct design (it does not claim to replace a draw) and evidence against treating the algorithm as a standalone diagnostic.

No cleared product implements this algorithm (see §5 and §8) and no guideline mentions ECG-based hyperkalemia screening as of this pass — not found, not merely not searched, since the KDIGO/ACC/AHA/HFSA guideline search itself has not yet been run (see Gaps).

Gaps: no guideline position retrieved (for either capillary sampling or ECG-based screening); no evidence on whether suppressing a hemolysed result changes clinical decisions; nothing on turnaround-time requirements; no external/non-Mayo validation of the ECG-screening approach located; no prospective trial located.

5. Reference products — and why people stop using them

as_of: 2026-08-25 (capillary-sampling subsection) / 2026-08-26 (RAASi-discontinuation subsection added below) · sources searched 2026-08-25: the three papers in §4 · sources searched 2026-08-26: connectors literature sweeps + literature-detail on 5 PMIDs (35085685, 41126484, 42506781, 40369054, 40788620, 34428591) · not searched: cleared point-of-care potassium devices, their IFUs, MAUDE, recalls, commercial-distribution status

The incumbent here is not a product but a practice: venous draw sent to a lab. The relevant abandonment evidence is therefore about the alternative practice — capillary fingerstick sampling — which has been tried in adult populations and repeatedly retreated from.

Capillary fingerstick sampling (as a practice, not a product)

• What it is: lancet collection of capillary whole blood, in place of venipuncture.
• Price / what it realizes per unit: not established.
• Evidence behind it: §4 above.
• Adoption: not established. One paper's title states the conclusion directly.
• Abandonment record — why people stop:

• Verbatim:

"Preanalytical Challenges During Capillary Fingerstick Sampling Preclude Its Widespread Use in Adult Hospitalized Patients" — title, PMID 33009571

"Patients required multiple fingers be lanced in 15% of cases"; "Capillary sample rejection rates due to insufficient volumes were as high as 30% in the initial 30 patients enrolled." — PMID 33009571

"Fingertip collections yielded larger volumes and fewer collection failures than TAP Micro Select but showed higher hemolysis." — PMID 42446434

Read this carefully in both directions. That people retreat from capillary potassium is evidence the failure mode is real and unsolved — which is what a candidate proposing to solve it needs. It is not evidence that anyone will adopt a device that claims to solve it, and it is not evidence that on-strip hemolysis rejection works. Abandonment of the incumbent is never demand for the replacement.

Gaps: no cleared point-of-care potassium product has been enumerated, so there is no product-level abandonment evidence at all here — only practice-level. MAUDE has no connector in this repo (docs/backlog.md 1.9).

RAAS inhibitor therapy — discontinuation and dose-reduction attributable to hyperkalemia

Added 2026-08-26. This is the incumbent drug therapy whose titration idea-011's device supports, and its own abandonment record is the revealed-behaviour twin for the class-3 question "would clinicians act on an ECG-based hyperkalemia flag?" (see §9 and docs/research-lane-spec.md §3.1). What clinicians have actually done, on record, when a potassium reading comes back high:

• What it is: ACE inhibitors, ARBs, ARNIs, and mineralocorticoid receptor antagonists (MRAs) — the guideline-directed therapy this device's titration workflow supports.
• Evidence behind continuing it: improved cardiovascular/renal outcomes is the premise of the whole domain; not re-litigated here.
• Abandonment record — why people stop (or are stopped by a clinician):

• Verbatim:

"RAASi therapy was discontinued in 6.2%, and down-titrated in 4.5% of patients due to hyperkalemia." — PMID 41126484

"After developing hyperkalemia, 346 (15%) discontinued RASi... Discontinuing RASi was associated with a 16% [95% CI 2-33%] higher hazard of mortality than continuing RASi." — PMID 40369054

"In the 30 days following the first hyperkalaemic event, RASi/MRA was down-titrated or discontinued in 3.6% of cases." — PMID 40788620

"Over median follow-up of 18 months, frequency of drug discontinuation of ACEis/ARBs, ARNIs and MRAs was 12.7%... 10.4%... and 20.4%... Chronic kidney disease was the only independent predictor of increased risk of discontinuation of each of the RAAS inhibitor classes." — PMID 34428591

Read this carefully, same discipline as the capillary-sampling rows above. This is evidence that clinicians do act on a potassium result by down-titrating or discontinuing RAASi at a real, measured, non-trivial rate — the revealed behaviour behind the unanswerable "would clinicians act on a hyperkalemia flag?" But note precisely what it does and does not establish for idea-011: every study that isolates hyperkalemia as the stated reason (41126484, 42506781, 40369054, 40788620) is non-US; the one US study located (34428591, CHAMP-HF) gives a general discontinuation rate with CKD as a predictor, not hyperkalemia isolated as the stated reason. And none of these studies involve an ECG-based flag at all — every one is a response to an actual lab potassium value, which is exactly the behaviour this candidate's own design (ECG flag → confirmatory draw → then a lab value) is built to preserve, not replace. Whether clinicians would act as promptly on an ECG-triggered flag as they do on a lab value remains genuinely unanswered — that half of the original question stays class 3, permanently.

Gaps (as of 2026-08-25/26, first pass): no US study isolating "discontinued specifically because of hyperkalemia" (rather than general discontinuation with CKD as a predictor) was located; no study on time-to-action after an ECG-type flag exists, or could exist, absent primary interview/prospective research.

2026-08-26, second research pass — a US Medicare complication-rate finding, plus a

second non-US discontinuation cohort

as_of: 2026-08-26, second pass · sources searched: connectors literature broadened retries (see §1) + literature-detail on PMID 35114384, 39052473

A US, Medicare-specific finding — closer to filling the gap above, but not the same question. PMID 35114384 is a US Medicare fee-for-service claims cohort study, not a discontinuation study: it reports the rate of hyperkalemia hospitalization (a downstream complication, not a discontinuation or down-titration decision) following MRA (spironolactone) initiation in HFpEF.

Verbatim:

"A total of 621,171 patients with HFpEF (mean age 80 ± 8 years, 62.9% female) were included. We identified 40,241 initiations of spironolactone with initiation rate/100 person-years of 16.8 ... in 2013 and increasing to 19.9 ... in 2018 ... we identified a total of 164 hyperkalemia hospitalization with stable incidence rates per 1000 person-years between 2013 (12.0 ...) and 2018 (10.6 ...)." — PMID 35114384

Read this narrowly: this is a hospitalization rate, not a discontinuation or down-titration rate — it does not fill the "US study isolating discontinuation specifically because of hyperkalemia" gap noted above, and the paper's own conclusion is that hyperkalemia hospitalizations stayed flat even as initiation rose (a reassurance finding about spironolactone safety post-TOPCAT, not a demand signal for a monitoring device). It is, however, a genuine US Medicare beneficiary count (621,171 HFpEF patients, claims-phenotyped) directly usable as TAM context — see §1.

A second non-US retrospective cohort, larger than the ones already on record. PMID 39052473 (DISCOVER CKD retrospective arm) is a UK/Japan claims-based comparison, much larger than the Saudi/Japanese/Canadian cohorts already in this section:

Verbatim:

"In the UK Clinical Practice Research Datalink, 37,713 patients with HK and 142,703 patients without HK were included for analysis (HK prevalence 20.9%). In the Japan Medical Data Vision, 5924 patients with HK and 74,272 patients without HK were included for analysis (HK prevalence 7.4%) ... The proportions of patients with CKD and HK who stopped treatment with diuretics, K+ binders, and RAASi during follow-up were 48.7%, 76.5%, and 50.6%, respectively, in the United Kingdom, and 22.9%, 53.6%, and 29.2%, respectively, in Japan." — PMID 39052473

Read this the same way as the cohorts above: another non-US data point (this dossier still has zero US studies isolating "discontinued specifically because of hyperkalemia," as opposed to general discontinuation or hospitalization), but a materially larger and more directly comparable pair of national claims databases than what was on record before, and it adds a treatment-pathway dimension (potassium-binder uptake) the earlier cohorts did not report at all.

Gaps, updated 2026-08-26 second pass: still no US study isolating "discontinued specifically because of hyperkalemia" — PMID 35114384 is the closest US data gets, and it measures hospitalization, not discontinuation; no study on time-to-action after an ECG-type flag exists, or could exist, absent primary interview/prospective research.

6. Reimbursement landscape

as_of: 2026-08-25 (lab-test billing question) / 2026-08-26 (Breakthrough-pathway and HCPCS-fetch notes added below)

Not searched (lab-test billing). As of 2026-08-25 no source has been queried for this section. The CMS Clinical Laboratory Fee Schedule, the CLIA waived-test list, the codes a point-of-care potassium and creatinine test would bill under, and whose budget an outpatient clinic pays them from are all open.

Note the standing constraint: CPT descriptors are AMA-licensed and must never be reproduced here. Scope to HCPCS plus published payer policy.

2026-08-26 addition — CMS's RAPID coverage pathway (Breakthrough-designated devices): Candidate idea-011's own Verifier pass established (via connectors policy 2026-16368, Federal Register document number 2026-16368 / 91 FR 51710, published 2026-08-11) that CMS has proposed — as a notice with an open comment period closing 2026-10-13, not a final rule — a "Regulatory Alignment for Predictable and Immediate Device" coverage pathway limited to Class II/III FDA Breakthrough-designated devices at the IDE-presubmission stage, requiring "separately payable" status, and explicitly excluding IVDs by name. Carried here because it is domain-relevant to any future non-invasive/screening candidate in this space, not only idea-011: an ECG-based (non-IVD) hyperkalemia or potassium-adjacent screening device with Breakthrough designation could plausibly route through this pathway if finalized; a device that measures potassium directly (an IVD) could not. Re-check document 2026-16368's status against the 2026-10-13 comment deadline before relying on this.

2026-08-26 addition — HCPCS Level II file: located, not yet loadable. A real, currently-live CMS URL for the January 2026 HCPCS Level II Alpha-Numeric file was located and confirmed reachable (https://www.cms.gov/files/zip/january-2026-alpha-numeric-hcpcs-file.zip, confirmed via engine/scripts/fetch_data_file.py --dry-run: HTTP 200, 2,519,574 bytes). It could not be loaded into data/hcpcs_level_ii.csv, because it is a ZIP archive and engine/scripts/fetch_data_file.py validates and writes CSV/TSV/JSON bodies only — it has no unzip step. This is a distinct, more specific finding than "no CMS files present": the source exists and is reachable, and the blocker is a capability gap in the fetch tool, not an access or licensing gap. Separately, www.cms.gov itself returned HTTP 403 to a plain page fetch (distinct host behaviour from the fda.gov 401 pattern already documented in this repo, but the same practical effect: an automated page-level fetch is not reliable here). No HCPCS code number is recorded in this dossier as a result — secondary sources (AAPC, vendor blogs) name G0511 and G0320-G0322 as remote-monitoring-adjacent HCPCS codes, but those were never independently verified against the primary CMS file or a resolvable government document by this pass, so they are deliberately not written down as findings here (a search-engine synthesis is not a source of record).

Gaps: the entire lab-billing question. hcpcs, payment, coverage and procedures remain blind — data/ still holds no CMS files (docs/backlog.md 1.1) — but the HCPCS gap specifically now has a concrete next step (extend the fetcher to handle a ZIP body, or fetch by hand and unzip) rather than being an open-ended absence.

7. Technology trajectory

as_of: 2026-08-25 (lab-strip technologies) / 2026-08-26 (ECG-wearable note) / 2026-08-26, second pass (broadened predicate search added below)

Not searched (lab-strip technologies). As of 2026-08-25 no source has been queried for this section. What has changed in ion-selective electrode miniaturisation, in enzymatic creatinine chemistry, and in on-strip optical hemoglobin detection over the last decade is open, as is the rate of new entry.

Constraint to carry: a counted series — clearances per year, entrants per year — cannot carry a confidence tag anywhere in this repo (docs/backlog.md 1.10), so anything of that shape lives here as context and cannot be scored from.

2026-08-26 addition — ECG wearables are an active, clearing technology, separately from the hyperkalemia-screening question. python3 -m connectors 510k "KardiaMobile" --applicant "AliveCor" confirms AliveCor's KardiaMobile 6L was cleared as a 510(k) (K220350, decided 2022-05-25) — a general ECG-recording consumer device, not indicated for hyperkalemia screening. This establishes that the underlying hardware category (adhesive/handheld consumer ECG) clears routinely; it is the hyperkalemia-inference algorithm layer specifically that has no located clearance (see §8).

2026-08-26, second pass — a genuinely broadened predicate search, across the whole regulatory shape a hyperkalemia-screening algorithm would need, not just AliveCor's name. The prior pass checked whether AliveCor's own device had cleared. This pass asked a wider question: does any company's ECG-based/arrhythmia-detection device carry a cleared indication touching electrolytes, potassium, or hyperkalemia, under any name?

Two searches, run directly against openFDA rather than through the single-top-match connectors 510k CLI (which only returns one record and would miss a real predicate hiding lower in a result list):

1. `requests.get('https://api.fda.gov/device/510k.json', params={'search': 'device_name:hyperkalemia'}) → **zero results, HTTP 404 with NOT_FOUND`** — a genuine empty-set answer (not the "endpoint doesn't exist" 404, see connectors/fda.py's EndpointUnavailable distinction), across the entire 510(k) database, not just AliveCor's applicant name. The same unqualified full-text query for bare hyperkalemia (no field restriction) also returns zero. Caveat: openFDA's 510(k) index does not carry full free-text of the Indications for Use statement — only structured fields (device_name, applicant, statement_or_summary, which holds only the literal word "Summary" or "Statement", not the actual text) — so this rules out device naming referencing hyperkalemia, not every possible IFU phrasing buried in a summary PDF this pass did not individually open.
2. device/classification search for regulation_number:870.2380 (the "Machine Learning-Based Notification Software" cardiovascular family openFDA associates with AI-ECG screening algorithms, discovered by tracing the product codes behind real companies' cleared AI-ECG devices — Anumana's and Eko's low-ejection-fraction algorithms, K232699/K233409, both product code QYE) returns **all ten established product codes in this family**: atrial fibrillation risk (SBQ), hypertension (SFR), coronary artery disease (QXX), general cardiovascular (QXO), cardiac hypertrophy (SIK), valvular regurgitation (SIJ), cardiac amyloidosis (SHP), aortic stenosis (SII), reduced ejection fraction (QYE), pulmonary hypertension (SAT). **None is for hyperkalemia, electrolyte imbalance, or potassium.** A broader unrestricted search for `device_name: "Notification Software"` across all regulations (12 total hits) adds only "Loss of Pulse Notification Software" (SDY, reg. 870.2790) and "Radiological Computer-Assisted Triage and Notification Software" (QAS, reg. 892.2080) — neither electrolyte-related.

Why this matters, structurally. The QYE family shows the actual pathway shape for this device category: a first company establishes a De Novo grant that creates a new product code for a specific ECG-AI "notification" target (e.g. reduced ejection fraction), and subsequent companies then 510(k) against that code as a predicate — five 510(k)s (K233409, K232699, K250119, K250649, K250652) now sit under QYE alone. No equivalent founding grant exists for hyperkalemia or potassium, so there is no product-code family for a follow-on 510(k) to piggyback against — this device would be the founding entrant, not a follow-on, if it is first to market. This sharpens (not contradicts) the prior pass's finding: the regulatory question for idea-011 is not "510(k) or De Novo" in the abstract, it is "De Novo, with no existing predicate-code family in this specific indication space to shorten the path," which is a harder and more specific claim than "no predicate found under AliveCor's name."

A secondary, structural curiosity worth flagging: python3 -m connectors regulation 870.2380 --expect "reduced ejection fraction" returns refuted — "There is no section 21 CFR 870.2380 in the Code of Federal Regulations as of 2026-08-19" — even though openFDA's own device/classification endpoint reports regulation_number: "870.2380" for all ten codes above. This is a discrepancy between FDA's internal classification database's regulation_number field and what is actually codified in the eCFR at the time of this pass; a future researcher citing "21 CFR 870.2380" as though it were confirmed CFR text would be citing something the eCFR connector cannot find. Not resolved further this pass — flagged so nobody re-derives false confidence from the classification database's field name alone.

Gaps: lab-strip technology trajectory (ISE miniaturisation, creatinine chemistry, optical hemoglobin detection) — the entire original section, still unsearched. Also unsearched: the individual Indications for Use PDFs behind the QYE-family and other AI-ECG 510(k)s (clearance-summary, per-K-number), which could in principle mention a secondary electrolyte-adjacent claim not captured by device_name alone.

8. Market players & dynamics

as_of: 2026-08-25 (lab-device players) / 2026-08-26 (AliveCor entry) / 2026-08-26, second pass (Anumana/Eko/Cardiologs surfaced as adjacent, not competing, entrants — see §7)

Not searched (lab-device players). As of 2026-08-25 no source has been queried for this section. The 2026-08-25 Stage 0 pass ran 6 clearances and 2 recalls sweeps on this question and their records were logged as signals, but no player-level picture was assembled from them and none of it has been read into this file.

2026-08-26 addition — AliveCor / KardiaK, an entrant with a stalled or non-public regulatory status in the ECG-hyperkalemia niche specifically. AliveCor announced (company press release, syndicated by CNBC, MedTech Dive, BioSpace, PRNewswire, Wareable — all reporting the same September 2018 announcement, not independent corroboration of the underlying fact) that its KardiaK Software Platform received FDA Breakthrough Device Designation for ECG-based, bloodless hyperkalemia screening. Two things this dossier independently established about checking that claim further:

1. FDA structurally will not confirm or deny pre-authorization designation status. FDA's own Breakthrough Devices Program page states: "the FDA generally cannot publicly disclose whether a sponsor has submitted a Breakthrough Device designation request for a device or whether the FDA has granted or denied the request, unless the sponsor decides to make that information available to the public," and FDA "maintain[s] on its website a list of devices granted Breakthrough Device designation **that have received marketing authorization**" only — i.e., a device only appears on FDA's own public list after clearance or De Novo grant, never before. Source: https://www.fda.gov/medical-devices/how-study-and-market-your-device/breakthrough-devices-program (fetched successfully this pass; note the general caveat already on record in this repo that fda.gov gates automated requests intermittently, so treat this as a single successful retrieval, not a guaranteed-repeatable one).
2. No 510(k) or De Novo grant for KardiaK has been located (candidate idea-011 claim 2; python3 -m connectors 510k "KardiaMobile" --applicant "AliveCor" finds only the unrelated K220350 6-lead consumer ECG clearance, decided 2022-05-25, with no hyperkalemia indication).

Put together: the absence of a public registry record for KardiaK's designation is not evidence the designation is false — it is the expected state of the world given FDA's own disclosure policy for a device that (per point 2) has not been market-authorized. A future researcher should not read "no FDA registry hit" as refuting the AliveCor claim; the correct reading is that this specific fact class (pre-authorization Breakthrough status) has no registry check available to anyone, ever, until authorization happens or the company discloses further. This is a structural limit on class-1 verification for this fact, not a gap in this repo's tooling.

2026-08-26, second pass — no other entrant found in the ECG-hyperkalemia niche specifically, after a broadened search. Companies actively clearing AI-ECG "notification software" devices (Anumana, Eko Health, Cardiologs — see §7) are all working the adjacent-but-distinct low-ejection-fraction/AFib/structural-disease niches, not hyperkalemia; none of their cleared device names, applicant-name search results, or the QYE-family product-code definitions reference potassium or electrolytes. This is a real finding (a search was run and came back empty), not an unsearched gap: as of this pass, AliveCor/KardiaK remains the only named entrant in ECG-based hyperkalemia screening specifically, cleared or not.

Gaps: the entire lab-device competitive landscape (point-of-care potassium/creatinine players, entries, exits, funding) remains unsearched.

9. Open questions — and what would settle them

10. Provenance & staleness

The Stage 0 sweeps this file draws on were limit-capped, and their counts are floors: pseudohyperkalemia 50 of 111, point-of-care potassium AND (accuracy OR agreement OR method comparison) 50 of 123, hemolysis rate AND (specimen OR sample) collection 50 of 75, hyperkalemia AND (RAAS inhibitor OR mineralocorticoid receptor antagonist) AND monitoring 50 of 89.

What this dossier has never looked at, as of 2026-08-26 (second pass): the care pathway and user groups (sections 2 and 3, untouched by either pass); CKD/HF incidence and demographic concentration proper (as opposed to the prevalence anchors now present in §1); lab-test billing codes and payment (section 6's original subject); the lab-strip technology trajectory (section 7's original subject); and the point-of-care device competitive landscape (section 8's original subject, distinct from the ECG-AI competitive landscape now covered). The 2026-08-26 second pass added a broadened, exhaustive-rather- than-top-1-match regulatory predicate search (§7), two new US TAM anchors and one new international-utilization anchor (§1), and a second, larger non-US discontinuation/ hospitalization cohort pair (§5) — it did not close the care-pathway, user-groups, or lab-billing gaps, and it still could not identify the original two claims-8/9 review papers or extract a 2024-vintage USRDS beneficiary count. The first pass had added a second, ECG-based technology arm and a drug-discontinuation abandonment record on top of the 2026-08-25 capillary-sampling material. Neither pass closed the six gaps original to the seed dossier. Absence of scanning is not absence of change, and this file remains a starting point for further research passes, not the output of one.