How Low Can You Go
Platelets, Penguins & Ghost Policies
An evidence-based recalibration of procedural platelet count thresholds
No verdict yet. Just the facts as they were presented.
No verdict yet. Just the facts as they were presented.
Scan to join the quiz
Results
Ghost Policies: What Are They?
A clinical practice rule — whether unwritten or buried in an unreviewed protocol — found in no published guideline, sustained by habit, enforced as if it were institutional policy.
Example: “Plt < 80K? No epidural.”
What other ghost policies do you know?
Well-Recognised Ghost Policies
in Anaesthesiology
- “Nil by mouth after midnight”
Rigid 6–8 h fasting — despite clear fluids clearing the stomach in <2 h (ASA 2017/2023; ESA-IC 2022) - “Hold all regular meds — patient is NPO”
ASA 2012: essential meds should continue with sips of water. β-blocker withdrawal → 2–4× perioperative MI risk - “Cricoid pressure prevents aspiration”
Sellick 1961 (n=26), never proven in an RCT. NAP4: aspiration in 50% of airway deaths regardless. Birenbaum 2019 RCT: no difference - “Preload with crystalloid before spinal”
Ni 2017 meta-analysis (10 RCTs, n=824): preload 57.8% hypotension vs coload 47.1% (OR 1.62). Vasopressors are what matter - “Trendelenburg for hypotension”
Geerts 2012: increases preload for ≤1 min — baroreceptor resetting eliminates the effect. Leg raise more effective, plus no ventilation impairment or ICP rise - “Aspirin = no neuraxial”
ASRA 2025: NSAIDs “do not create a level of risk that will interfere with neuraxial block.” Confusion with P2Y12 inhibitors (clopidogrel) which DO require 5–7 day hold
More Ghost Policies
- “Routine coagulation screen before neuraxial”
ASA guidelines: history & exam sufficient. van Veen 2010: routine PT/APTT has near-zero predictive value. Chee 2008: “not recommended” - “Spinal contraindicated in severe aortic stenosis”
Titrated epidural or low-dose spinal may be safer than GA — individualised risk assessment, not an absolute - “No adrenaline in digital blocks”
Ilicki 2015: 2,797 blocks, zero ischaemic complications. The 48 pre-1950 cases were procaine (pH ~1), not adrenaline. 1,111 modern cases — zero necrosis - “Ketamine contraindicated in raised ICP”
Zeiler 2018 systematic review: no ICP elevation in TBI. 1970s studies had uncontrolled CO₂ — ICP rise was CO₂-mediated - “No Hartmann's in renal failure — the potassium will cause hyperkalaemia”
Contains only 5 mmol/L K+ — less than a banana. SMART secondary analysis 2021: patients WITH hyperkalemia had less progression to severe K+ and half the need for RRT with balanced crystalloids vs saline. The “safe” saline makes it worse (hyperchloremic acidosis → transcellular K+ shift)
How Ghost was born
Bromage PR 1993 — Neurologic Complications of Regional Anesthesia for Obstetrics · In: Shnider SM, Levinson G, eds. Anesthesia for Obstetrics. 3rd ed. Baltimore: Williams & Wilkins, 1993:443–4
Early textbook authority in obstetric anaesthesia. Widely referenced by subsequent authors and guideline committees as the foundational word on neuraxial safety in thrombocytopenia.
Historical Anaesthesia Texts — The <100,000 Limit
The <100,000/µL threshold was largely popularised in the 1990s by textbook authors and early authorities like Dr Charles Bromage. These texts carried enormous weight — before systematic reviews, before MPOG, before the Rule of Three — and became the de facto standard for a generation of anaesthetists.
Conservative Abundance — Caution Without Denominator
Early guidelines favoured extreme caution because the consequence of an epidural haematoma — nerve damage or paralysis — is severe, even though the actual bleeding risk is very low. Expert opinion filled the evidence vacuum. A thousand platelets above 100K was safety; one below was perceived negligence.
How Ghost was born
Rasmus 1989 — Unrecognized thrombocytopenia and regional anesthesia · Obstet Gynecol 1989;73:943–6 · PMID 2726116
2,929 consecutive parturients reviewed. 24 had platelet counts <100,000/µL (range 15,000–99,000). 14 received regional anaesthesia — including patients as low as 15K. Zero permanent neurologic sequelae.
Beilin 1997 — Safe epidural analgesia with Plt 69–98K · Anesth Analg 1997 (n=30) · PMID 9249118
30 parturients with platelet counts 69–98 × 10⁹/L received epidural analgesia. Zero epidural haematomas. Authors cautioned: small sample, not a safety guarantee.
van Veen 2010 — The risk of spinal haematoma following neuraxial anaesthesia or lumbar puncture in thrombocytopenic individuals · Br J Haematol 2010;148:15–25 · PMID 19775301
"In the absence of additional risk factors, a platelet count of 80 × 10⁹/L is a safe count for placing an epidural or spinal anaesthetic."
Authors explicitly stated this was expert opinion based on small case series — not an absolute cutoff. Over time, "safe count" was transformed into "do not proceed below 80K."
How Ghost was born
NICE NG24 2015 — Blood Transfusion (1.7.4) · November 2015
"Consider prophylactic platelet transfusions to raise the platelet count above 50×10⁹/L in people who are having invasive procedures or surgery."
A transfusion trigger — not a procedural contraindication. "Consider" ≠ "do not proceed."
BCSH 2017 — Guidelines for the use of platelet transfusions · Br J Haematol 2017;176:365–394
"lumbar puncture when the platelet count is ≥40×10⁹/L (2C)."
"venous central lines … when the platelet count is >20×10⁹/L (1B)."
BCSH 2017 itself sets thresholds well below 50K — LP at 40K, CVC at 20K. The 50K ghost is stricter than the guideline it claims to follow.
How They Survive
I was trained to do it like this
Cultural transmission
We've always done it this way
Institutional inertia
Someone told me it was the rule
Oral tradition
The Cost of Ghost Policies
Clinical Harm
- Denying neuraxial → more general anaesthetics
- GA in obstetrics: significant morbidity and mortality
- Unnecessary treatment
- Delays, prolonged admissions, wasted blood products
NZ Medicolegal Risk
- HDC Right 4(2): services must comply with professional standards → MCNZ requires evidence-based practice
- Right 6: patients entitled to know guidelines support proceeding at ≥70K
- Bolitho: practice contradicted by all published guidelines cannot withstand logical scrutiny
- “You must always be prepared to explain your decisions” — MCNZ Good Medical Practice
What Guidelines Actually Say
| Guideline | Recommendation | Reference |
|---|---|---|
| SOAP 2021 | ≥70K → Proceed | 50–70K → Competing risks may justify | <50K → Avoid | SOAP 2021 Interdisciplinary Consensus Statement on Neuraxial Procedures in Obstetric Patients with Thrombocytopenia. Anesth Analg 2021;132:1531–1544. PMID: 33861047 |
| AAGBI / RA-UK 2013 | ITP/Gestational: 50–75K increased risk, 20–50K high risk | Pre-eclampsia: <75K very high risk (if abnormal coags) | Harrop-Griffiths W, et al. Regional anaesthesia and patients with abnormalities of coagulation. Anaesthesia 2013;68:966–972. PMID: 23905877 |
| ANZ ITP 2022 | ≥70 × 10⁹/L is reasonable for neuraxial (GRADE 2D) | Choi PY-I, et al. Australian and New Zealand consensus statement on the management of immune thrombocytopenia in pregnancy. Med J Aust 2022;217:43–51. PMID: 34628650 |
| Bauer Meta-analysis 2019 | Inflection point at 75K — below this, risk increases (19,000+ procedures) | Bauer ME, et al. Lumbar neuraxial procedures in thrombocytopenic patients: systematic review and meta-analysis. J Clin Anesth 2019;57:103–110. PMID: 31810860 |
| MPOG 2017 | 573 neuraxial with Plt <100K (MPOG) + 951 literature = 1,524 combined — zero epidural haematomas. Stratified: ≥70K ≤0.2%, 50–69K ≤3.4%, <50K ≤11% | Lee LO, et al. Risk of Epidural Hematoma after Neuraxial Techniques in Thrombocytopenic Parturients. Anesthesiology 2017;126:1053–1063. PMID: 28383323 |
The Big Numbers
Moen 2004
Sweden, 1990–1999 · Nationwide registry
200,000
obstetric epidurals
- 2 spinal haematomas in the entire obstetric population
- Both patients had HELLP syndrome — systemic coagulopathy, not isolated thrombocytopenia
- Across ALL populations (1.7 million neuraxial blocks): 33 spinal haematomas total — coagulopathy in 11, IV heparin in 18, thrombocytopenia in 5, ankylosing spondylitis in 5
Moen V, Dahlgren N, Irestedt L. Severe Neurological Complications after Central Neuraxial Blockades in Sweden 1990–1999. Anesthesiology 2004;101:950–959. PMID: 15448529
NAP3 2009
United Kingdom · 1-year prospective audit
320,425
obstetric neuraxial blocks
- Zero spinal haematomas in the obstetric population
- Largest prospective audit of neuraxial complications ever conducted
- Confirmed obstetric patients are a lower-risk population for epidural haematoma
- Benchmark for the "observing zero" statistical problem
Cook TM, Counsell D, Wildsmith JAW. Major complications of central neuraxial block: report on the Third National Audit Project of the Royal College of Anaesthetists. Br J Anaesth 2009;102:179–190. PMID: 19139027
Zero Observations — But what are the odds?
Interpreting “Zero”: The Rule of Three
The Rule of Three
- Hanley & Lippman-Hand, JAMA 1983 — a landmark paper addressing exactly this problem
- Calculates the maximum possible risk with 95% confidence when zero events are observed
- Formula: Maximum Risk ≈ 3 / n (where n = sample size)
In plain English: If you observe zero events, the true rate could still be as high as 3/n — you just got lucky. With 95% confidence, it's no higher than that. The "3" comes from −ln(0.05) ≈ 2.9957.
“If you flip a coin 3 times and get zero heads, you'd have no confidence the coin was fair. But flip it 300 times and still get zero heads — you're confident the probability of heads is less than about 1%.”
Applying It to the Obstetric Data
Lee et al. (2017): MPOG 573 + systematic review 951 = 1,524 thrombocytopenic parturients (combined). Zero epidural haematomas.
| Plt Range | n | Max Risk (95% CI) |
|---|---|---|
| 0–49,000 | 27 | ≤11% (~1:9) |
| 50–69,000 | 89 | ≤3.4% (~1:30) |
| 70–100,000 | 1,286 | ≤0.2% (~1:429) |
The ≥70K group has a maximum SEH risk of 0.2% — ≤1:429. This is where the evidence-based consensus threshold comes from. Not 80K. Not 100K. 70K.
Below 70K, uncertainty widens fast — exactly why guidelines recommend individualised assessment rather than a single cutoff.
Source: Hanley JA, Lippman-Hand A. If Nothing Goes Wrong, Is Everything All Right? Interpreting Zero Numerators. JAMA 1983;249:1743–1745.
Rule of Three: Limitations & Why We Still Use It
Critical Limitations
- It's an approximation. The “3” is a convenient rounding of −ln(0.05) = 2.9957
- Assumes perfect data. Doesn't account for selection bias (academic centres only) or publication bias (zero-event studies less likely published)
- Ignores prior knowledge. A Bayesian approach would start with our existing belief that SEH is rare — the true estimated risk would be even lower
Why We Still Use It
- Simple and universal. All major papers in this field use it — creates a common language for discussing risk
- Usefully conservative. The critiques suggest the true risk is likely even lower than the Rule-of-Three ceiling
- Clinically actionable. ≤0.2% max SEH risk for Plt ≥70K (≤1:429); 3.4% for 50–69K; 11% for <50K. GA carries its own real risks (failed intubation 1:224–1:808). Both matter — ghost policies ignore both sides of the ledger.
By using the 0.2% figure, we are presenting a conservative, worst-case scenario that is still incredibly reassuring compared to the alternatives.
Critique informed by: Winkler RL, et al. J Am Stat Assoc 2002. Bayesian perspective: Gelman A, et al. Bayesian Data Analysis 3rd ed. 2013, Chapter 2.8.
The Hidden Risk: GA vs Neuraxial
General Anaesthesia (Obstetric)
1:224–1:808
Failed tracheal intubation in obstetric GA (range: Quinn BJA 2013 → DREAMY 2021)
Of those failed intubations, case fatality rate:
1:90
Maternal deaths per failed intubation (Kinsella, IJOA 2015)
Absolute risk of maternal death from GA (airway):
≈1:20,000–1:72,000
= failed intubation rate × case fatality rate
- Aspiration: full stomach in labour
- Hypertensive response to laryngoscopy
- Uterine atony from volatile agents
- No neuraxial opioids for post-op analgesia
Spinal Epidural Haematoma
~1:200,000
General obstetric population (Moen, Anesthesiology 2004)
320,425
NAP3 2009: UK obstetric neuraxial blocks — zero spinal haematomas
Zero cases
In 573 neuraxials with Plt <100K (MPOG 2017)
Mitigations:
- 95% of SEH present symptoms within 48h
- Early decompressive laminectomy (<8h) is critical
- Enhanced neurological monitoring detects signs
- Low-concentration LA + opioid preserves motor function
Zero penguins in the car park ≠ no penguins in the world
Why Platelet Count Can Mislead
- Platelet count measures quantity, NOT function
- ITP / gestational: low count but young, large, hyperfunctional platelets — count underestimates haemostatic capacity
- Cirrhosis / liver disease: normal count may mask severe platelet dysfunction
- CKD / uraemia: normal count + dysfunctional platelets
- No correlation between platelet count and surgical bleeding (Slichter, Transfus Med Rev 2004)
TEG / ROTEM: Looking Beyond the Count
Clinical Role
- SOAP 2021: insufficient evidence for routine use — consider in equivocal cases
- PFA-100: insufficient evidence (SOAP, AAGBI)
- Best used when count is borderline and aetiology uncertain
Normal TEG/ROTEM is reassuring; abnormal should prompt caution
Workup for the Low Platelet Count Patient
History & Exam
- Bleeding history (epistaxis, gingival, menorrhagia, bruising)
- Medications (heparin, NSAIDs, aspirin, anticonvulsants)
- Previous platelet counts — acute vs chronic?
- Hepatosplenomegaly, lymphadenopathy, petechiae
- Pregnancy: pre-eclampsia features, BP, urinalysis
Laboratory Workup
- FBC + blood film (pseudothrombocytopenia? schistocytes?)
- Coagulation screen (PT, aPTT, fibrinogen)
- Liver function tests, creatinine
- TEG/ROTEM — if available and count borderline
Pattern Recognition
| Aetiology | Count | Function | Bleeding Risk |
|---|---|---|---|
| Gestational | ↓ | Normal/↑ | Low |
| ITP | ↓↓ | Normal/↑ | Variable |
| Pre-eclampsia/HELLP | ↓ | ↓ | High |
| Liver disease | Variable | ↓↓ | High |
| DIC | ↓↓ | ↓↓ | Very high |
Decision Framework
- ≥70K + no coagulopathy + reassuring history → proceed
- 50–70K + ITP/gestational → risk-benefit discussion
- <50K or concerning aetiology → consult haematology
NZ & Australian Guidelines
ANZCA
- Endorses NBA Patient Blood Management Guidelines (2023–24)
- NBA Module 2: Perioperative (ANZCA endorsed)
- NBA Module 5: Obstetrics (ANZCA endorsed)
- No standalone ANZCA platelet threshold guideline exists
- References AAGBI/RA-UK 2013 in practice
ANZ ITP Consensus 2022
- ≥70 × 10⁹/L is reasonable for neuraxial anaesthesia
- GRADE 2D (weak recommendation, low quality evidence)
- Consider haematology referral for counts <50K
- IVIg or corticosteroids can boost pre-delivery counts
“Individual risk-benefit assessment is essential”
NZ Blood Service
- NZBS follows ANZSBT guidance
- Not procedure-specific thresholds
- Prophylactic: Plt <10K (no bleeding) or <20K (with risk factors)
- Therapeutic: active bleeding + Plt <50K
ANZCA has no independent platelet threshold guideline — it endorses NBA and ANZ ITP consensus. AAGBI 2013 and SOAP 2021 are the most clinically useful references.
Practical Decision Algorithm for Neuraxial
Assess Aetiology
ITP / Gestational? Pre-eclampsia, DIC, sepsis, liver disease?
Check Trend
Stable or rising → more permissive. Falling rapidly → more cautious.
Platelet Count
≥70K → Proceed (SOAP/AAGBI)
50–70K → Risk-benefit discussion
<50K → Avoid; consult haematology
Risk-Benefit Analysis
GA risk vs SEH risk. Difficult airway? Obesity? Labour? GA may be the riskier option.
Mitigations
Experienced operator · Low-concentration LA + opioid · Enhanced neuro monitoring × 48h
Patient education — red flags (95% present ≤48h):
Leg weakness (59%) · Back pain (41%) · Bladder/bowel dysfunction (27%) · Numbness/tingling (23%) · Saddle anaesthesia (9%)
Adapted from: SOAP 2021, AAGBI 2013, Bauer et al. 2019. Always document risk-benefit discussion and monitoring plan.
Replace Ghost Policies With Evidence
| # | Recommendation | Rationale |
|---|---|---|
| 1 | Individualised Risk-Benefit Assessment | Replace rigid thresholds with documented clinical judgement |
| 2 | ≥70K = Evidence-Based Green Zone | SOAP, AAGBI, ANZ ITP consensus all support neuraxial at ≥70K |
| 3 | GA Is NOT the “Safe” Default | Failed intubation 1:224–1:808; case fatality per failed attempt 1:90 — GA has real, quantifiable risks that should factor into the decision alongside SEH risk |
| 4 | Platelet Count ≠ Function | ITP: low count, normal function. Pre-eclampsia: opposite. Aetiology matters |
| 5 | Document & Audit | Record aetiology, trend, guidelines, risk-benefit. Audit neuraxial denial rates |
| 6 | Mitigate Bleeding Risk | US-guided technique + experienced operator + atraumatic needle + low-concentration LA |
Procedure Thresholds at a Glance
Invasive Procedures
| Procedure | Plt/μL | Source |
|---|---|---|
| CVC (US-guided, IJ/SCV) | 20,000 | BCSH 2016; PACER: ≥30K for tunneled/haem |
| Lumbar Puncture | 40,000 | van Veen 2009; BCSH 2016 |
| Arterial Line (radial) | No specific | SIR — compressible site |
| PICC Line | 20,000 | Similar to CVC; compressible |
| Intercostal Drain | 50,000 | NICE NG24 |
| General / Abdominal Surgery | 50,000 | NICE NG24; 50–75K for high-risk |
Peripheral Nerve Blocks
| Risk Tier | Plt/μL | Examples |
|---|---|---|
| ● Low (superficial, compressible) | ≥50K | Fascia iliaca, femoral, axillary, popliteal, ankle/wrist |
| ● Intermediate (deeper, partial) | ≥50–75K | Interscalene, supraclavicular, infraclavicular, adductor canal |
| ● High (deep, non-compressible) | ≥75K (treat as neuraxial) | Lumbar plexus, deep cervical, stellate, coeliac |
Sources: PACER — van Baarle et al. NEJM 2023. PNB — AAGBI/RA-UK 2013 Table 2; ASRA 5th Ed 2025. Invasive — BCSH 2016; NICE NG24; SIR 2019.
References
[1] SOAP 2021 — Neuraxial procedures in thrombocytopenic obstetric patients. Anesth Analg. PMID 33861047
[2] AAGBI/RA-UK 2013 — Regional anaesthesia & coagulation abnormalities. Anaesthesia. PMID 23905877
[3] Bauer 2019 — Neuraxial procedures in thrombocytopenia: meta-analysis. J Clin Anesth. PMID 31810860
[4] van Veen 2010 — Spinal haematoma risk in thrombocytopenia. Br J Haematol. PMID 19775301
[5] Lee MPOG 2017 — Epidural hematoma in thrombocytopenic parturients. Anesthesiology. PMID 28383323
[6] van Baarle PACER 2023 — Platelet transfusion before CVC placement. NEJM. PMID 37224197
[7] Estcourt Cochrane 2016/2018 — No RCTs on platelet thresholds for neuraxial. PMID 29709077
[8] NICE NG24 2015 — Blood transfusion guideline. nice.org.uk/guidance/ng24
[9] BCSH 2017 — Guidelines for platelet transfusions. Br J Haematol. PMID 28009056
[9a] BCSH 2003 — Guidelines for the use of platelet transfusions. Br J Haematol 2003;122:10–23. PMID 12823341
[10] ANZ ITP 2022 — Management of adult ITP in Australia & NZ. Med J Aust. PMID 34628650
[11] NBA PBM 2023–24 — Patient blood management guidelines. blood.gov.au
[12] NAP3 2009 — Major complications of neuraxial block in UK. Br J Anaesth. PMID 19139027
[13] Bhardwaj 2017 — Coagulopathies: TEG, ROTEM, Sonoclot analysis. Ann Card Anaesth. PMID 28393783
[14] Kinsella 2015 — Failed intubation in obstetric GA: review. Int J Obstet Anesth. PMID 26303751
[14a] Quinn 2013 — Failed intubation: UK case-control. Br J Anaesth. PMID 22986421
[14b] DREAMY 2021 — GA for obstetric surgery in England. Anaesthesia. PMID 32959372
[15] Moen 2004 — Neurological complications after neuraxial block. Anesthesiology. PMID 15448529
[16] SIR 2019 — Periprocedural bleeding risk management. J Vasc Interv Radiol. PMID 31229333
[17] Hanley 1983 — Rule of Three: interpreting zero numerators. JAMA. PMID 6827763
[18] Slichter 2004 — Platelet count & bleeding risk. Transfus Med Rev. PMID 15248165
[19] Lind 1991 — Bleeding time does not predict surgical bleeding. Blood. PMID 2043759
Abbreviation Key
Guidelines & Organisations
SOAP — Society for Obstetric Anesthesia and Perinatology
AAGBI — Association of Anaesthetists of GB & Ireland
ANZCA — Australian & NZ College of Anaesthetists
ANZ ITP — ANZ Immune Thrombocytopenia Consensus
BCSH — British Committee for Standards in Haematology
NICE — National Institute for Health and Care Excellence
NBA — National Blood Authority (Australia)
NZBS — New Zealand Blood Service
ANZSBT — ANZ Society of Blood Transfusion
SIR — Society of Interventional Radiology
MCNZ — Medical Council of New Zealand
HDC — Health and Disability Commissioner
Regulatory
HDC Code — NZ Code of Health and Disability Services Consumers' Rights
Bolitho — Bolitho v City and Hackney HA [1997] (logical basis test)
Databases & Trials
MPOG — Multicenter Perioperative Outcomes Group
PACER — Platelet Transfusion before CVC Placement trial
NAP3 — Third National Audit Project (RCoA)
DREAMY — GA for Obstetric Surgery in England study
Cochrane — Cochrane Database of Systematic Reviews
Clinical Terms
ITP — Immune Thrombocytopenia
SEH — Spinal Epidural Haematoma
GA — General Anaesthesia
TRALI — Transfusion-Related Acute Lung Injury
TACO — Transfusion-Associated Circulatory Overload
TEG — Thromboelastography
ROTEM — Rotational Thromboelastometry
PFA-100 — Platelet Function Analyzer-100
RCT — Randomised Controlled Trial
CVC — Central Venous Catheter
HELLP — Haemolysis, Elevated Liver enzymes, Low Platelets
DIC — Disseminated Intravascular Coagulation
Thank You
Questions & Discussion
"If you see zero penguins in 1,286 commutes, you're ≤0.2% confident of seeing one tomorrow."
Roman Vasilev · Specialist Anaesthetist · Palmerston North Hospital
Prepared June 2026 — evidence current as of May 2026