What prospective multicenter observational study should be conducted to standardize Hemoglobin-to-Mean-Drainage Ratio (HMDR) cut‑off values, compare diagnostic performance across studies, and evaluate HMDR as a single predictor in the target population?

Proposed Prospective Multicenter Observational Study Design for HMDR Standardization

A prospective multicenter observational study should enroll at least 500-1000 adult patients requiring airway assessment across multiple institutions, measuring HMDR in both neutral and extended head positions, correlating findings with modified Cormack-Lehane grades during direct laryngoscopy, and stratifying analysis by BMI categories to establish standardized cut-off values with adequate power for subgroup analysis.

Study Population and Sample Size

The target population should include consecutive adult patients (≥18 years) scheduled for elective surgery requiring endotracheal intubation. Based on the variability observed in existing studies, where HMDR cut-offs ranged from ≤1.20 in obese patients to ≤1.23 in non-obese patients 1, a sample size of 800-1000 patients would provide sufficient power (80-90%) to detect clinically meaningful differences in diagnostic performance across BMI strata and institutional settings.

Stratification should include:

• BMI categories: <25-30, >30 kg/m² (minimum 200 patients per stratum)
• Multiple centers (minimum 4-6 sites) to capture geographic and practice variation
• Balanced representation of sex and age groups

Standardized Measurement Protocol

Primary Predictor Variable

HMDR must be measured using a rigorous, standardized protocol:

• Hyomental distance measured in both neutral position and full head extension using a rigid ruler
• Measurements performed by trained personnel with documented inter-rater reliability (κ >0.80)
• HMDR calculated as: (HMD in extension) / (HMD in neutral position)
• All measurements recorded to nearest millimeter before intubation

This addresses the critical gap of measurement variability that has plagued prior studies, where cut-off values differed substantially between populations 1.

Reference Standard

Modified Cormack-Lehane (MCL) grading during direct laryngoscopy should serve as the gold standard outcome, with grades 2b, 3, and 4 defined as difficult laryngeal visualization (DVL). The laryngoscopist must be blinded to HMDR measurements to prevent incorporation bias.

Key Outcome Measures

Primary outcomes:

• Sensitivity and specificity of HMDR at various cut-points for predicting DVL
• Optimal cut-off values determined by Youden index for each BMI stratum
• Area under ROC curve (AUC) for HMDR as a single predictor

Secondary outcomes:

• Comparison of HMDR diagnostic performance versus modified Mallampati grade, thyromental distance, and neck circumference measurements
• Subgroup analyses by BMI category, age, and sex
• Inter-institutional variability in cut-off values and diagnostic accuracy

Statistical Analysis Plan

Addressing Current Evidence Gaps

The existing evidence shows HMDR performed well in a single-center study with AUCs of 0.72-0.91 depending on BMI status 1, but multicenter validation is absent. Your study must:

1. Establish universal cut-offs through pooled analysis:

• Calculate institution-specific ROC curves and optimal cut-points
• Use random-effects meta-analysis to derive pooled sensitivity/specificity estimates
• Assess heterogeneity using I² statistics to quantify between-center variability

2. Evaluate HMDR as standalone predictor:

• Perform multivariable logistic regression with HMDR as the sole predictor (unadjusted model)
• Compare discrimination (AUC) against models including traditional predictors
• Calculate net reclassification improvement to quantify added value

3. Validate across populations:

• Internal validation using bootstrap resampling (1000 iterations)
• Temporal validation by splitting cohort into derivation (first 60%) and validation (last 40%) sets
• External validation through prospective application at one holdout center not included in derivation

Critical Design Elements to Minimize Bias

Blinding protocol:

• Laryngoscopists blinded to all preoperative airway measurements
• Personnel measuring HMDR blinded to patient's Mallampati grade and other predictors
• Outcome assessors blinded to HMDR values when adjudicating difficult intubation

Standardization across sites:

• Centralized training for all personnel performing measurements (video-based modules with competency assessment)
• Quality control audits with random review of 10% of measurements by independent observers
• Standardized data collection forms with mandatory fields to prevent missing data

Addressing the Variability Problem

The fundamental issue is that HMDR cut-offs varied from 1.20 to 1.23 in the single available study 1, and no data exist comparing performance across different populations or settings. Your study must determine whether this variability represents:

1. True population differences requiring BMI-specific or demographic-specific cut-offs
2. Measurement error necessitating stricter standardization protocols
3. Institutional practice variation in laryngoscopy technique affecting the reference standard

Planned subgroup analyses stratified by BMI will directly address whether the 1.20 cut-off for obese patients (sensitivity 85.7%, specificity 91.7%) and 1.23 cut-off for non-obese patients (sensitivity 68.8%, specificity 72.4%) 1 represent genuine biological differences or study artifacts.

Data Collection and Management

Prospective data capture should include:

• Demographics: age, sex, BMI, comorbidities
• All competing airway predictors measured simultaneously: Mallampati grade, thyromental distance, neck circumference at hyoid and thyroid levels
• Intubation details: laryngoscopy grade, number of attempts, alternative devices used
• Operator experience level (to assess for interaction effects)

Quality assurance measures:

• Real-time data entry with range checks and logic constraints
• Weekly monitoring of enrollment rates and missing data patterns
• Interim analysis at 50% enrollment to assess feasibility and adjust protocols if needed

Sample Size Justification

Assuming DVL prevalence of 35-47% based on prior data 1, enrolling 1000 patients would yield approximately 400 events. This provides:

• Narrow confidence intervals around sensitivity/specificity estimates (±5%)
• Adequate power (>90%) to detect AUC differences of 0.05 between subgroups
• Sufficient events-per-variable (>10) for stable multivariable models including interaction terms

Implementation Considerations

Feasibility factors:

• Recruitment from high-volume surgical centers to achieve target enrollment within 18-24 months
• Minimal exclusion criteria (only emergency intubations and patients with anatomic airway abnormalities) to enhance generalizability
• Pragmatic design embedded within routine clinical care to maximize external validity

Common pitfalls to avoid:

• Selection bias from excluding difficult cases (include all comers)
• Verification bias (ensure all patients receive reference standard laryngoscopy regardless of HMDR value)
• Spectrum bias (include full range of BMI and airway difficulty)

The study should not attempt to evaluate HMDR for predicting intubation failure or complications, as the evidence base suggests it specifically predicts laryngeal visualization 1, not overall intubation success. Expanding outcomes prematurely would dilute focus and require substantially larger sample sizes.