Introduction
Angioedema is a commonly self‐limited but potentially life‐threatening disease that is characterized by an episodic swelling of submucosal or subcutaneous tissues as a result of increased vascular permeability. Submucosal involvement within the head and neck may lead to rapid airway compromise and necessitate the need for urgent intervention [1]. The reported lifetime prevalence of an acquired episode of angioedema is 7.4% [2].
Although a majority of these cases resolve either spontaneously or with medical therapy, persons with head and neck involvement have experienced rapid respiratory failure and death [3]. This emphasizes the importance of accurate, timely risk stratification to provide possible airway management.
Angioedema is most commonly mediated by either a histamine or bradykinin pathway [4]. The histaminergic profile is typically allergic, responding to medical management of antihistamines and steroids, whereas bradykinin‐mediated profiles, such as those caused by ACE inhibitors or patients with hereditary C1‐esterase inhibitor deficiency, may not respond to standard therapies, thus posing higher risks of airway compromise [4, 5].
Although there is extensive literature describing the progression and pathophysiology underlying angioedema‐related airway compromise, there are key gaps in knowledge which persist. The ability to quickly and accurately identify which patients are at the greatest risk for airway compromise is a continued urgent clinical need. Prior efforts have largely focused on the presence of specific high‐risk symptoms, such as dyspnea, stridor, or drooling.
Others have had an anatomical approach, isolating subsites of edema such as the tongue or supraglottis [6, 7, 8, 9]. However, there has not been a systematic approach to the stratification of these factors, which hinders the ability of clinicians to use multiple data points to synthesize decision trees. One of the key tools for evaluating the involvement of the upper airway in angioedema is flexible laryngoscopy, as it provides a direct visualization of edema beyond what is appreciable on clinical, transoral physical exam, which is limited in depth to the oropharynx [1, 10].
However, few studies have attempted to correlate findings on scope exams to clinical outcomes and management. Current clinical decision‐making tools use findings on physical and clinical exams, such as anterior tongue swelling, to predict need for intubation or other airway interventions [11]. The role of serial physical exams for assessment has also been discussed as a primary mode of surveillance management for possible intubation [12].
One unique component to angioedema is its dynamic, variable nature, which, to establish a protocol for, requires a structure that not only assesses static findings at presentation but includes clinical trajectories, such as when and why scoping occurs and changes to the edematous areas. There have been efforts to create staging systems for angioedema; however, they are limited. The Ishoo staging system was described in 1999, stratifying risk based on the visible anatomic involvement [13].
Other disciplines have created staging tools, such as the Derkay system, to guide management and prognosis for other pathologies, such as recurrent respiratory papillomatosis [14]. The Derkay System, a subsite‐based framework with anatomic scoring systems, has shown improved outcomes in other laryngeal diseases [15]. However, there is no validated angioedema‐specific scoring rubric that guides airway risk assessment based on anatomic and symptomatic presentation.
Additionally, many of the large cohort studies are focused exclusively on angiotensin‐converting enzyme inhibitor (ACE‐i) induced cases [6, 16, 17, 18]. Due to their isolated scope in etiology, these studies often lack comprehensive laryngoscopy exams, symptom profiles, or other detailed outcome tracking beyond intubation. Thus, the implications for ENT consultation patterns, requirement for repeat laryngoscopy exams, or variability in care pathways are incompletely characterized.
This study aims to fill these gaps in clinical assessment through the evaluation of patients with aerodigestive angioedema by correlating flexible laryngoscopy findings and presentation profiles with clinical outcomes. Specifically, we aim to identify concrete anatomical and symptomatic predictors of airway intervention, hospital admission, and length of stay (LOS) as we explore the influence of timing and laryngoscope exams on the decision‐making and care trajectories.
Methods
We performed a retrospective study of all patients presenting with angioedema at a single tertiary care academic center between January 2013 and 2023. ICD codes were used to identify 220 potentially eligible patients, and these were confirmed with manual chart review for presentation consistent with angioedema. Inclusion criteria were patients 18 years or older who presented with a primary diagnosis of airway angioedema involving the head and neck region.
In patients who did not undergo laryngoscopic evaluation, the diagnosis of angioedema was established based on clinical assessment, including history and physical examination findings documented by the treating clinical team, with laryngoscopy performed selectively to further evaluate airway involvement when indicated. Seventeen patients were excluded because they lacked documentation of the specific site(s) of involvement or presenting symptoms or sites of involvement were beyond the region of the head and neck.
Demographic information of age, sex, race, BMI, and clinical data were manually abstracted from the EMR and uploaded into a secured RedCap platform. Presenting symptoms, comorbidities, suspected etiology, and site of initial evaluation (e.g., emergency department [ED] or inpatient setting) were all recorded. Laryngoscopic findings were extracted from procedure notes, and when available, still images were reviewed by an otolaryngology resident (K.W.) to confirm specific subsites of involvement.
Each anatomic subsite was coded using a binary scale to determine presence or absence of edema. For patients with multiple scope exams, data from the initial laryngoscopy was used for primary analysis.
To facilitate region‐based airway involvement analysis, we grouped individual anatomic subsites visualized on flexible laryngoscopy into five broader anatomic regions based on their functional and topographic relationships (Table 1). Subsites were coded numerically during data entry and then reassigned to regional categories as follows: the oral cavity included lips (1), floor of mouth (20), anterior tongue (2), and hard palate (3); the oropharynx included soft palate (4), base of tongue (5), and vallecula (6); the supraglottis encompassed the epiglottis (lingual and laryngeal surfaces; 7, 8), aryepiglottic folds (9, 10), false vocal folds (11, 12), and arytenoids (15, 16); the glottis included the true vocal folds (13, 14) and anterior/posterior commissures (17, 18); and the subglottis was defined by involvement of the subglottic trachea (19). This grouping enabled structured analysis of edema burden by functional airway region.
Subsite involvement was determined based on laryngoscopic visualization; a value of zero indicates no edema visualized on laryngoscopy, which may occur in cases where angioedema is present in regions not consistently assessed on flexible laryngoscopy (e.g., lips or anterior tongue) but identified on clinical examination.
The primary outcome was need for airway intervention, defined as intubation or surgical airway placement. Secondary outcomes included hospital admission and LOS. For admitted patients, LOS was measured in days from initial admission to discharge.
The need for ENT consultation and repeated laryngoscope examinations was also recorded.
Descriptive statistics were used to summarize demographics and presenting symptoms. Categorical variables were compared using chi‐square or Fisher's exact tests, and continuous variables were compared using Mann–Whitney U tests and independent t‐tests. Subgroup analyses analyzed associations between subsite involvement and airway intervention.
Odds ratios were calculated for predictors of intubation and admission. An ROC curve analysis was also performed to evaluate the predictive performance of significant variables for airway intervention where the area under the curve (AUC) was used to assess the model's ability to distinguish between those who did and did not need airway intervention. Significance was defined as p < 0.05.
All analyses were done using R (version 4.2.1).
This study was approved by the Institutional Review Board (IRB) at the University of Virginia (IRB‐HSR # 24864). All data were stored and analyzed in compliance with institutional HIPAA policies.
Results
Two hundred and three patients presented with angioedema over a 10‐year period at a single tertiary academic medical center. The mean age of all patients was 54.1 years (SD: 20.1), and the cohort was predominantly female (65.5%) and African American/Black (55.2%). One hundred and forty‐two patients had a documented BMI at the time of visit, and the mean was 30.8 (SD: 8.6).
Most patients (98%) presented to the ED, either by self‐presentation or EMS transport, while only four patients (2.0%) were evaluated as inpatients. At the time of presentation, 61 of the 203 patients (30.0%) had angioedema attributed to the use of ACE‐inhibitors, most commonly Lisinopril. Common comorbidities included hypertension (60.6%), diabetes mellitus (23.6%), and coronary artery disease (11.8%).
Eighty‐three (40.9%) patients required admission for management of their angioedema, for an average LOS of 2.78 days (SD: 3.72). In patients with diabetes mellitus, the mean hospital LOS was 4.5 days, compared to 2.1 days in those without (p = 0.059).
Of the 203 patients, 69 underwent laryngoscopic evaluation, including 54 of the 83 patients who were admitted (65.1%). Nine (13%) of these patients initially received laryngoscopy performed by the ED team, with the remainder being initially scoped by the ENT team. Among the patients initially scoped by the ED, 44% required re‐scoping by the ENT team, compared to 34% of patients when initially scoped by ENT, and 67% received an ENT consult (Table 2).
Additionally, among patients scoped initially by the ED versus ENT team, there was no significant difference in the rates of hospital admission (67% vs. 78%, respectively, p = 0.430) or airway intervention (22% vs. 14%, p = 0.611).
Among the 69 patients who underwent flexible laryngoscopy, 31 (44.9%) had still images available in the medical record, all of which were accompanied by written descriptions of exam findings. The remaining examinations were documented descriptively without stored images. Individual site involvement was recorded across 21 possible subsites.
The most commonly affected laryngoscopic subsites were the lips (43.5%), anterior tongue (40.6%), and base of tongue (31.9%).
Subsites were grouped into broader anatomic regions as described in the Methods. Oropharyngeal involvement was significantly associated with both increased likelihood of airway intervention (32.0%, p = 0.0037) and prolonged hospital stay (p = 0.0336) (Table 3). In contrast, oral cavity, supraglottis, glottis, and subglottis involvement were not independently associated with these outcomes.
The mean number of involved subsites per patient was 2.6 (range: 0–10), and patients with 4–7 sites involved had a significantly higher rate of airway intervention (50.0%) compared to those with 0–3 sites (7.4%, p = 0.0036). Among these 69 patients who received laryngoscopic evaluation, patients who had oropharyngeal involvement (base of tongue, vallecula, soft palate) of angioedema experienced statistically significant longer hospital stays (p = 0.0336) and higher rates of airway intervention (32.0%, p = 0.0037) than those without oropharyngeal involvement.
Airway intervention occurred in 17 patients (8.4%), including 15 patients who underwent endotracheal intubation and 2 patients who required a surgical airway. Of these patients, stridor (OR 12.5, p = 0.0088), odynophagia (OR 7.6, p = 0.0083), dysphonia (OR 4.6, p = 0.0033), and tongue swelling (OR 6.0, p = 0.0039) were significantly associated with airway intervention (Figure 1). Among the 17 patients who required airway intervention, 10 underwent flexible laryngoscopic evaluation prior to intervention.
Based on this data, demonstrating predictability for airway intervention, tongue swelling on clinical exam, stridor, and odynophagia were selected as three binary predictors for a reduced multivariable logistic regression model in order to identify a symptom‐based criteria for identifying patients at increased risk for the need of airway intervention (Figure 2). The model produced an area under the receiver operating characteristic curve (AUC) of 0.79, as shown in Figure 2, which means that the model, based on only those 3 variables, correctly assigns a high risk score to patients requiring intubation or the creation of a surgical airway. Thus, these 3 variables alone have a strong ability to differentiate patients at high risk of requiring airway intervention.
This model was derived from readily observable clinical exam findings and symptoms, which allows it to serve as a potential foundation for a practical triage scoring criterium.
Discussion
This study determines key clinical and laryngoscopic predictors of airway compromise in patients presenting with angioedema. Stridor, tongue swelling, and odynophagia emerged as the most significant indicators of the need for airway intervention, which is consistent with prior literature discussing the severity of edema involving the larynx and posterior oropharynx [13, 16, 17]. These findings emphasize the role of early, targeted airway evaluation and management protocols for high‐risk patients.
Though symptoms provide a key part of the initial screen, the use of laryngoscopic evaluation enhances diagnostic specificity [18, 19]. Our results demonstrate that the use of laryngoscopy to determine the precise anatomic localization of edema provides a critical insight beyond the transoral examination. Specifically, oropharyngeal involvement of angioedema was significantly associated with longer hospital stays and a higher likelihood of airway intervention.
The need for systematic laryngoscopic evaluation, such as the categorization used in this study, adapted from the Derkay staging system, is reinforced by the correlation between the number of involved subsites and intubation risk [14]. To our knowledge, this is the first study to quantify the specific site of laryngeal involvement with this level of granularity in terms of airway outcomes in patients with angioedema.
Additionally, our data showed that initial laryngoscopic examination in the ED spared 33% of patients scoped in the ED from requiring consultation and re‐scoping by the ENT team. For the remainder, specialist evaluation still played an important role, as over half of the patients required repeat scope exams. Rather than diminishing the role of ENT, these findings highlight that ED laryngoscopy can streamline care for a subset of patients with clear and reassuring findings, while ensuring that patients with more concerning or evolving symptoms continue to benefit from specialist expertise.
We also found a trend of longer hospital stays in patients with diabetes mellitus. Likely, this reflects the effects of steroid‐induced hyperglycemia, complicating the ability to achieve adequate glucose control prior to discharge. As steroids are a key part of management in angioedema, this is an expected but under‐recognized complication within angioedema management.
Prior studies have shown that corticosteroid use in diabetic patients can delay recovery and prolong hospital stay [20, 21]. In the future, clinicians should monitor glucose closely in these patients, considering this interaction when selecting and dosing steroids.
Nearly one third of our cases of angioedema were due to ACE inhibitor‐associated angioedema, which aligns with national data in Black populations [22]. Given the setting of our study in Central Virginia, where Black residents comprise roughly 20% of the population, the overrepresentation in our sample (55.2%) demonstrates the disparity of this side effect in these patients. Beyond prevalence, this highlights the potential disparities in medication prescribing, access to preventative care, or emergency care use patterns [23].
Future studies should investigate this at the community level, assessing interventions and alternative anti‐hypertensive regimens to reduce this disparity.
Lastly, our predictive model achieved an AUC of 0.79, which demonstrates good discrimination in the need for intubation while using only three clinical features: tongue swelling, odynophagia, and stridor. This has a significant impact for use at bedside triage, while also demonstrating its role in early, symptom driven risk stratification. While these results are promising, there is a need for further external validation prior to implementation.
Additionally, while laryngoscopy was essential for this study, we acknowledge the limitations in access and training that may restrict its use in some community or hospital settings.
Limitations to our study include the retrospective design, limited documentation of laryngoscope findings, and intubations performed prior to full airway evaluation in the hospital. Laryngoscopic data were derived from a combination of procedure notes and still images which may introduce variability in the precision of subsite classification, particularly in cases with less detailed documentation. Additionally, some patients required airway intervention prior to complete laryngoscopic evaluation, limiting the availability of full anatomic assessment in the most severe presentations.
However, our findings do provide actionable and promising insights that highlight pathways for improved risk assessment and early management decisions in patients presenting with angioedema.
Conclusion
This study identifies both symptom and anatomic based predictors of airway compromise in patients with angioedema. The results offer an approach for early risk stratification through the combination of physical exam and formal laryngoscopic evaluation. Clinical features of tongue swelling, odynophagia, and stridor were strongly associated with the need for airway intervention and established the basis for our predictive model, which had a strong discriminatory performance (AUC = 0.79).
Specific laryngoscopic findings of oropharyngeal involvement and the total number of edematous subsites had an increased risk of intubation rates and extended length of hospital stay, reinforcing the value of systematic airway visualization in these patients.
Our results support and expand on previous frameworks of symptom severity correlates and demonstrate the necessity of a combination approach of symptom and scope‐based anatomical localization to add specificity. This data provides a strong foundation for a future, evidence‐based scoring rubric that integrates both presenting symptoms and laryngoscopic subsites to guide airway management decisions during triage assessment.
Though external validation is needed, these findings have important clinical implications for emergency and otolaryngological care teams. A structured, symptom‐driven protocol with the addition of flexible laryngoscopy can help identify patients who may necessitate airway intervention, decrease unnecessary admissions or scopes, and improve care efficiency. Future prospective studies are needed to validate such a model, assessing its utility across diverse clinical settings and populations, and explore its ability to be integrated into clinical decision making pathways.
Funding
The authors have nothing to report.
Conflicts of Interest
The authors declare no conflicts of interest.