Skip to content

Method

How this report was built. What you can trust. What you can't.


Data Sources

SourceWhat It ProvidesConfidence
FDA MAUDE (2.99M events)Adverse event counts, deaths, injuriesHigh for counts
Company-reported proceduresDenominators for per-procedure ratesModerate (unverified)
MANIFEST-US registry (41,968 patients)FARAPULSE clinical complication ratesHigh
VARISURE Safety Survey (850 procedures)VARIPULSE post-correction safety, same-day dischargeModerate (single survey)
VOLT-AF IDE (392 patients)Abbott Volt clinical effectiveness/safetyHigh (IDE)
OMNY-AF pilot (30 patients)Omnypulse early safety/efficacyLow (pilot)
CMS Medicare data (CPT 93656)US AF ablation volume: 74,448 beneficiaries (2023)High for Medicare
LLM extraction (40% of events)Failure modes, root causes, design flawsModerate
VA procurement (USAspending.gov)Federal contract volumes, institutional adoptionModerate (federal only)
Analyst surveys (BTIG, Citi)Market share estimatesLow-moderate
2026 AF Symposium (Feb 6, 2026)Late-breaking clinical dataHigh (peer-reviewed)

Procedure Volume Estimates

Per-procedure rates in this report use these denominators:

PlatformProceduresSourceLimitation
FARAPULSE500,000+ globalBSC earnings (cumulative, early 2026)Global, not US-specific
PulseSelect~30,000 global estMedtronic PR (Sept 2024: 10K+), extrapolatedStale. Actual may differ.
VARIPULSE~40,000 globalJ&J Q4 2025 earnings ("nearly 40,000 patients")Global, not US-specific
AfferaNot disclosedCannot estimate rates
Volt~392 (IDE only)VOLT-AF IDE trialPre-commercial

Total US AF ablation: ~250,000 procedures (2025 estimate). Based on CMS CPT 93656: 74,448 Medicare beneficiaries (2023), ~12% annual growth, x2.5-3x all-payer multiplier. CMS does not distinguish PFA from RF ablation.


The Reporting Rate Comparison

MAUDE events per 1,000 procedures:

  • FARAPULSE: 4.8
  • VARIPULSE: 6.5
  • PulseSelect: 34.7

The reporting gap between FARAPULSE and VARIPULSE has narrowed from 7x (at 3K procedures) to ~1.35x (at 40K procedures). This is within normal manufacturer variation. Per-procedure comparisons are now meaningful.

PulseSelect remains a significant outlier at 34.7 per 1K — consistent with its high manufacturing defect rate.


Key Limitations

  1. Procedure denominators are company-reported. Not independently verified. Global, not US-specific. MAUDE events are predominantly US.
  2. LLM extraction covers 40% of events. Root cause classification (design_flaw, user_error, unknown) is based on the extracted subset. The 60% unextracted events could differ.
  3. PulseSelect volume is estimated. Medtronic reported "10,000+" in Sept 2024. We estimate ~30,000 by late 2025 based on event growth trajectory. Actual may differ.
  4. "Death" does not equal "caused by device." MAUDE reports events where the device was involved. A death during ablation may not be device-caused. VARIPULSE's first death was likely caused by QDOT (RF), not PFA.
  5. VARIPULSE deaths: 3 MAUDE records, 2 patients. Two Nov 28 records are the same event filed twice. Per-patient death count is 2, not 3.
  6. CMS data lags 18-24 months. Most recent confirmed: CY2023. 2024-2025 volumes are projected.
  7. Affera volumes unknown. Medtronic reports PFA revenue but not Affera-specific procedure counts.
  8. VARISURE and OMNY-AF are conference data. Not yet peer-reviewed at time of writing.

Reproducible Queries

All queries run against ScanPath ClickHouse (2.99M events). PFA head-to-head:

sql
SELECT
  CASE
    WHEN d.brand_name ILIKE '%varipulse%' THEN 'VARIPULSE'
    WHEN d.brand_name ILIKE '%farawave%'
      OR d.brand_name ILIKE '%farapulse%' THEN 'FARAPULSE'
    WHEN d.brand_name ILIKE '%affera%'
      OR d.brand_name ILIKE '%sphere%9%' THEN 'AFFERA'
    WHEN d.brand_name ILIKE '%pulseselect%' THEN 'PULSESELECT'
  END as platform,
  count() as events,
  countIf(resulted_in_death = 1) as deaths,
  countIf(root_cause = 'design_flaw') as design_flaws,
  countIf(root_cause = 'manufacturing_defect') as mfg_defects
FROM spincast.events e
JOIN spincast.devices d ON e.device_id = d.id
WHERE e.event_date >= '2024-01-01'
  AND platform IS NOT NULL
  AND d.brand_name NOT ILIKE '%glenosphere%'
  AND d.brand_name NOT ILIKE '%therasphere%'
GROUP BY platform
ORDER BY events DESC

SpecValue
DatabaseScanPath ClickHouse
Total events2.99M, deduplicated
LLM extraction40% (1.2M events)
Data throughJanuary 2026
GeneratedFebruary 2026
Key sourcesMANIFEST-US (JACC, Jan 2026), VARISURE (2026 AF Symposium), VOLT-AF IDE (2026 AF Symposium), HRS/EHRA PFA Statement (Feb 2026), CMS 2026 OPPS Final Rule