Pre-Op Screening Checklist: What Every Anesthesiologist Should Verify Before Surgery

Why checklists work — and why ours still miss things

Anesthesiologists know checklists. The WHO Surgical Safety Checklist reduced major surgical complications by 36% when implemented properly. The pre-induction timeout is standard. The three-pause model — sign in, time out, sign out — is ingrained. Checklists work because they impose structure on cognitive work that is otherwise vulnerable to omission under time pressure and distraction.

But the pre-operative assessment isn't the timeout — it happens hours or days before the case, under very different conditions. There's no structured pause, no laminated card, no second voice to verify. It's a clinician working through a chart, an intake form, and a patient interview, assembling a risk picture that will drive every anesthetic decision that follows.

That's where gaps appear. Not from lack of skill, but from incomplete information, time pressure, and the absence of a structured protocol. The 11-item checklist below is what that structured protocol looks like. Each item includes what to check, why it matters clinically, the common miss rate in practice, and how AI pre-op assessment catches what manual review misses.

The pre-op screening checklist

1. 1

   Cardiac Risk Assessment

   What to check: Active cardiac conditions (unstable angina, decompensated CHF, significant arrhythmias, severe valvular disease), revised cardiac risk index (RCAI) factors, functional capacity (METs), and any recent cardiac events within 60 days.

   Why it matters: Perioperative major adverse cardiac events (MACE) account for approximately 40% of perioperative mortality. The revised Lee cardiac risk index predicts MACE risk from six independent predictors — and it can be applied entirely from structured intake data. Patients with poor functional capacity (<4 METs) and ≥2 RCAI risk factors may need further cardiology evaluation before elective cases.

   Common miss: Patients with compensated heart failure who describe fatigue as "normal for my age." History of prior MI more than 6 months ago that the patient doesn't consider worth mentioning.

   ⚠ Missed in ~18% of manual reviews ✓ AI flags all RCAI criteria automatically
2. 2

   Airway Assessment

   What to check: Prior difficult airway documentation, Mallampati class, mouth opening, neck mobility, thyromental distance, body habitus, facial hair, presence of dentures or loose teeth, and sleep apnea history.

   Why it matters: Unanticipated difficult airway is one of the most dangerous intraoperative scenarios. The NAP4 audit found that 72% of airway complications were associated with failure to anticipate difficulty. Pre-op identification of risk factors allows for advance equipment preparation, awake intubation planning, and team briefing — none of which is possible if the first indication of difficulty is failed laryngoscopy at induction.

   Common miss: Sleep apnea documented in primary care notes but not mentioned by the patient. Obesity-associated airway changes in patients who haven't had prior intubations and have no "prior difficult airway" flag in the chart.

   ⚠ OSA missed in ~22% of pre-op screens ✓ AI cross-references BMI, OSA history, and airway predictors
3. 3

   Medication Reconciliation

   What to check: Complete medication list including OTC medications, supplements, and herbal products. Specifically flag: anticoagulants and antiplatelets (hold timing), beta-blockers (continue perioperatively), ACE inhibitors/ARBs (hold morning of surgery debate), SGLT2 inhibitors (hold 3 days pre-op due to DKA risk), MAOIs, and medications that prolong QT.

   Why it matters: Drug interactions and inappropriate medication continuation/discontinuation account for a significant portion of preventable adverse perioperative events. SGLT2 inhibitors in particular remain underrecognized — the FDA issued guidance on hold protocols after perioperative DKA cases in non-diabetic patients with cardiac indications.

   Common miss: Herbal supplements (fish oil, ginkgo, vitamin E) that affect bleeding. SGLT2 inhibitors in patients with heart failure who don't identify as diabetic. Supplements stopped 1–2 days before surgery rather than the recommended 7–10 days.

   ⚠ Supplement interactions missed in ~31% of cases ✓ AI screens full medication list against 200+ anesthesia interactions
4. 4

   NPO Status Verification

   What to check: Last oral intake time for solids, liquids, and medications. Apply ASA fasting guidelines: 8 hours for solids, 6 hours for non-clear liquids (including milk), 4 hours for breast milk, 2 hours for clear liquids. Confirm patient understands which medications to take with sips of water the morning of surgery.

   Why it matters: Aspiration pneumonitis and aspiration pneumonia are preventable perioperative catastrophes. While true aspiration events are uncommon with appropriate fasting, the consequences are severe — mortality rates up to 70% in Mendelson's syndrome. Patient compliance with NPO instructions is highly variable; studies show 15–25% of patients consume something within the specified NPO window.

   Common miss: Patients who took oral medications with a full glass of water 2 hours before the case. Diabetic patients who drank juice to treat a hypoglycemic episode overnight. "Clear liquid" interpretation varies — coffee with cream is not a clear liquid.

   ⚠ NPO non-compliance present in ~15% of cases ✓ AI flags NPO status with timestamped last intake confirmation
5. 5

   Laboratory Review

   What to check: Relevant labs based on patient comorbidities and procedure. CBC for anemia and platelet count (transfusion threshold planning), BMP/CMP for electrolytes (hypokalemia and hyponatremia increase arrhythmia risk), creatinine and GFR for nephrotoxic drug dosing, coagulation studies for patients on anticoagulants or with bleeding history, HbA1c for diabetic patients, and type and screen for cases with significant blood loss potential.

   Why it matters: Unrecognized preoperative anemia increases transfusion rates, complications, and length of stay. Electrolyte abnormalities — particularly potassium below 3.0 mEq/L — significantly increase intraoperative arrhythmia risk and should prompt delay of elective cases for correction. Abnormal creatinine affects dosing decisions for multiple perioperative medications.

   Common miss: Labs ordered but not reviewed before the case starts. Labs that are 90+ days old for a patient whose clinical status has changed. Mild anemia (Hgb 9–10) in a patient scheduled for a procedure with moderate blood loss potential.

   ⚠ Actionable lab abnormalities missed in ~12% of reviews ✓ AI flags out-of-range values and staleness by comorbidity
6. 6

   Allergy Verification

   What to check: All documented allergies with reaction type (true anaphylaxis vs. intolerance vs. side effect). Specifically assess for: latex allergy, penicillin allergy (cross-reactivity with cephalosporins is low but matters for prophylaxis selection), NSAIDs, propofol in patients with egg or soy allergy (clinical significance debated but worth documenting), and iodine/contrast if imaging is expected intraoperatively.

   Why it matters: Perioperative anaphylaxis occurs in 1 in 10,000–20,000 procedures and carries a mortality rate of approximately 3–9%. Neuromuscular blocking agents, antibiotics, and latex are the most common triggers. The value of allergy verification isn't just reaction avoidance — it's having the full picture before induction so that if anaphylaxis occurs, you have a reasonable differential for the causative agent.

   Common miss: Latex allergy in patients with multiple prior surgeries who've been sensitized gradually. "Penicillin allergy" documented without reaction type, leading to avoidance of preferred prophylaxis antibiotics in patients who likely have no true allergy.

   ⚠ Reaction type undocumented in ~40% of allergy entries ✓ AI prompts reaction characterization and cross-reactivity flags
7. 7

   Pulmonary and Respiratory Assessment

   What to check: Active respiratory infections (contraindication to elective anesthesia), history of asthma or COPD with current control status, smoking history (pack-years, cessation timing), obstructive sleep apnea severity and CPAP compliance, recent pulmonary function tests for high-risk cases, and ability to cooperate with postoperative respiratory therapy.

   Why it matters: Pulmonary complications are the second most common cause of perioperative morbidity and mortality. Active upper respiratory infections in pediatric patients increase laryngospasm risk 10-fold. In adults, active bronchospasm contraindicates elective cases. COPD with significant obstruction (FEV1 <50% predicted) affects ventilator management strategy and postoperative disposition planning.

   Common miss: "Just a little cough for the past week" that represents active bronchitis. Undertreated asthma — patients often minimize symptom frequency to avoid case delay.

   ⚠ Active respiratory symptoms underreported in ~19% of cases ✓ AI uses branching questions to characterize cough/wheeze onset and severity
8. 8

   Bleeding Risk Assessment

   What to check: Personal or family history of abnormal bleeding, current anticoagulation or antiplatelet therapy and last dose, history of prior surgical bleeding complications, liver disease (synthetic function and coagulation factor production), renal disease (uremic platelet dysfunction), and for female patients — current menstrual cycle timing relative to case date.

   Why it matters: Unrecognized bleeding diathesis is a leading cause of unexpected intraoperative hemorrhage. Beyond the obvious anticoagulation hold protocols, hereditary bleeding disorders (Von Willebrand disease affects ~1% of the population), acquired coagulopathies from liver disease, and medication effects from antiplatelet agents all require pre-op identification and planning.

   Common miss: Von Willebrand disease in women who describe "heavy periods" as normal. Aspirin taken for primary prevention that the patient doesn't consider a "real medication." Low-dose aspirin within 7 days of procedures with significant bleeding risk.

   ⚠ Aspirin use missed in ~24% of pre-op medication reviews ✓ AI specifically queries aspirin, NSAIDs, supplements, and bleeding history
9. 9

   Anesthesia History Review

   What to check: Prior anesthetic experiences and any adverse events, family history of malignant hyperthermia or pseudocholinesterase deficiency, history of postoperative nausea and vomiting (PONV), previous awareness under anesthesia, and prior difficult intubation documentation.

   Why it matters: Malignant hyperthermia is rare (1:5,000–1:100,000) but uniformly fatal without prompt recognition and treatment. A family history of "fever during anesthesia" or "unexplained death during surgery" must trigger volatile agent avoidance and dantrolene availability. PONV affects 30% of patients without prophylaxis — prior history is the strongest independent predictor and should drive antiemetic selection before the case starts.

   Common miss: Family history of MH described as "grandparent died during a surgery years ago" — not connected by patients to their own anesthetic risk. Prior PONV documented only in the nursing postoperative note, not the anesthesia summary the next provider sees.

   ⚠ Family MH history undercaptured in ~28% of pre-op screens ✓ AI asks direct MH family history questions with clinical context
10. 10

    Metabolic and Endocrine Status

    What to check: Diabetes mellitus — type, control (recent HbA1c), insulin regimen and morning-of dose guidance, target intraoperative glucose range. Thyroid disease — hypothyroidism or hyperthyroidism requiring perioperative management. Adrenal insufficiency — steroid use history and stress dosing protocol. Obesity — weight, BMI, and implications for drug dosing, positioning, and postoperative disposition.

    Why it matters: Perioperative hyperglycemia is an independent predictor of surgical site infection, impaired wound healing, and increased mortality. Patients on insulin often receive conflicting instructions from surgery, primary care, and anesthesia. Adrenal suppression from chronic steroids requires stress dosing to prevent perioperative adrenal crisis — a potentially fatal event that can occur hours after surgery.

    Common miss: Prednisone 5mg/day for rheumatoid arthritis — patient doesn't consider it "steroids." Type 2 diabetic on SGLT2 inhibitor for cardiac indication who stopped medication 1 day (not 3 days) before surgery.

    ⚠ Steroid use missed in ~16% of reviews when prescribed for non-endocrine indications ✓ AI flags all corticosteroid use regardless of indication
11. 11

    Informed Consent Verification

    What to check: Anesthesia consent completed and documented, patient understanding of anesthetic options (general vs. regional vs. MAC) and associated risks, discussion of PONV risk and management plan, discussion of awareness risk, and for regional techniques — confirmation of specific consent including nerve block risks.

    Why it matters: Informed consent for anesthesia is both a legal requirement and an ethical obligation — but in practice it often happens under time pressure in the preoperative holding area, with patients who are anxious and may not retain information. A pre-op screening process that identifies the appropriate anesthetic plan before the day of surgery allows for meaningful consent conversations, not rushed signatures.

    Common miss: Consent obtained but anesthetic plan changed based on day-of clinical findings without re-consent. Regional anesthesia consent obtained without specific discussion of the nerve to be blocked and associated risks.

    ⚠ Plan changes after consent underdocumented in ~20% of cases ✓ AI generates pre-op summary that becomes consent conversation framework

Where manual pre-op review breaks down

Each item above sounds straightforward in isolation. The problem is cumulative complexity. A patient presenting for elective laparoscopic cholecystectomy might have: type 2 diabetes on a SGLT2 inhibitor, a penicillin allergy documented without reaction type, a BMI of 38 with untreated OSA, a family history of "heart problems," and a fish oil supplement they didn't think to mention. Each of those is individually manageable. Together, they represent five separate threads to pull before anyone can safely proceed.

Manual review can catch all of this — if the chart is complete, the intake is thorough, and the clinician has time. In practice, none of those conditions holds consistently. Charts are incomplete. Intake forms are free-text. And 15 minutes of pre-op review is optimistic in a high-volume setting.

AI pre-op screening addresses this by turning the checklist into a structured protocol that runs automatically. The patient completes a digital intake that covers every item on this list using branching logic — the system doesn't stop at "do you have diabetes?" It follows up with insulin type, dose, last SGLT2 inhibitor ingestion, and morning glucose. The output isn't a completed questionnaire. It's a structured clinical summary with each checklist item either cleared or flagged, ready for the anesthesiologist to review in two minutes instead of twenty.

How AI catches what manual review misses

The miss rates noted for each checklist item above aren't failures of clinical skill — they're failures of information architecture. The information existed. It just wasn't in the right place at the right time in the right format.

AI catches these misses through three mechanisms:

1. Structured, branching intake. A clinical intake questionnaire built on the 11-item framework above doesn't ask about bleeding history once — it asks about personal history, family history, prior surgical bleeding, and then specifically asks about aspirin, NSAIDs, and supplements in separate questions. The miss rate for aspirin drops from 24% to near zero when it's asked about explicitly.
2. Cross-referencing between items. A patient who lists metformin as a medication, is diabetic, and also reports taking a "heart medication" gets automatically flagged for SGLT2 inhibitor verification — because the combination of diabetes and cardiac indication is the exact risk pattern. A human reviewer would need to make that connection consciously. The AI makes it structurally.
3. Consistent application regardless of volume or time pressure. The tenth patient of the day gets the same 11-item screening as the first. No items get deprioritized because the schedule is running late. For group practices focused on standardizing outcomes across multiple providers, this consistency is as valuable as any individual accuracy improvement.

The clinical judgment about what to do with the flagged findings still belongs to the anesthesiologist. AI identifies that the patient has a family history consistent with MH risk and hasn't had genetic testing. What happens next — whether to proceed with total intravenous anesthesia, pursue MH testing, or consult genetics — is a clinical decision that requires human expertise and context. The AI made sure you knew. The decision is yours.

That's the right division of labor. Structured data collection and pattern-matching is exactly what AI does reliably. Clinical reasoning from incomplete and ambiguous information is exactly what experienced anesthesiologists do. Use both for what they're built for.