Mateo Londoño Barrientos, Carlos Alberto López Zapata, David Alejandro Mejía Toro, Carlos Andrés Delgado López
In this cohort, hard signs were highly specific but identified only about half of patients with major injury, which is consistent with the rationale for imaging-based selective management rather than evidence that any particular strategy is safe. A composite of clinical signs and haemodynamic status captured more patients but left a non-trivial residual risk and low specificity. These single-centre, internally validated findings are exploratory and require external validation before informing practice.
PURPOSE: Hard signs guide operative decision-making in penetrating neck trauma (PNT), yet their accuracy and predictive value remain incompletely defined. We quantified the diagnostic accuracy of hard signs for major injury and for operative or endovascular management, and explored multivariable risk estimation.
METHODS: Single-centre retrospective cohort of 310 consecutive patients with PNT (June 2013-July 2025). Hard and soft signs were assigned by the attending surgeon at initial evaluation and abstracted retrospectively from the clinical record. The primary outcome was major injury (confirmed vascular or aerodigestive injury). Operative or endovascular management was analysed as a secondary outcome describing institutional practice, not as an independent diagnostic property, because the same signs informed both the index test and the treatment decision. We estimated sensitivity, specificity, predictive values and likelihood ratios with exact binomial confidence intervals, and fitted exploratory multivariable models (STARD/TRIPOD).
RESULTS: Major injury occurred in 92/310 (29.7%) and operative or endovascular management in 96/310 (31.0%). Hard signs were specific but insensitive for major injury (sensitivity 55.4%, 95% CI 44.7-65.8; specificity 88.1%, 83.0-92.1) and for management (sensitivity 59.4%, 48.9-69.3; specificity 90.7%, 85.9-94.2). Outcome frequency rose across no signs, soft signs only and hard signs (major injury 5/97, 36/136 and 51/77; management 9/97, 30/136 and 57/77; trend p < 0.001). The composite "any sign or haemodynamic instability" reached sensitivity 96.7% (90.8-99.3) for major injury at a specificity of 34.4%, yet 5 of 78 composite-negative patients (6.4%, 2.1-14.3) still underwent operative or endovascular management. Restricting the reference standard to imaging- or operatively-verified patients lowered the negative predictive value of hard signs from 82.4% to 80.0%; under the extreme assumption that every clinically-verified patient harboured an occult injury it fell to 68.7%. Exploratory models discriminated moderately (optimism-corrected AUC 0.83 and 0.80) and better than the composite rule, but on decision-curve analysis offered no advantage over hard signs alone at clinically relevant thresholds.
CONCLUSION: In this cohort, hard signs were highly specific but identified only about half of patients with major injury, which is consistent with the rationale for imaging-based selective management rather than evidence that any particular strategy is safe. A composite of clinical signs and haemodynamic status captured more patients but left a non-trivial residual risk and low specificity. These single-centre, internally validated findings are exploratory and require external validation before informing practice.