The Evaluation of Physical Exam Findings in Patients Assessed for Suspected Burn Inhalation Injury

Transcription

1 original article: 2014 aba paper The Evaluation of Physical Exam Findings in Patients Assessed for Suspected Burn Inhalation Injury Jessica A. Ching, MD,* Jehan L. Shah, BS, Cody J. Doran, BS, Henian Chen, MD, PhD, Wyatt G. Payne, MD,* David J. Smith, Jr, MD* The purpose of this investigation was to evaluate the utility of singed nasal hair (SN), carbonaceous sputum (CS), and facial burns (FB) as indicators of burn inhalation injury, when compared to the accepted standard of bronchoscopic diagnosis of inhalation injury. An institutional review board approved, retrospective review was conducted. All patients were suspected to have burn inhalation injury and subsequently underwent bronchoscopic evaluation. Data collected included: percent burn, burn injury mechanism, admission physical exam findings (SN, CS, FB), and bronchoscopy findings. Thirty-five males and twelve females met inclusion criteria (n = 47). Bronchoscopy was normal in 31 patients (66%). Data were analyzed as all patients and in subgroups according to burn and an enclosed space mechanism of injury. Physical exam findings (SN, CS, FB) were evaluated individually and in combination. Overall, the sensitivities, specificities, positive predictive values, and negative predictive values calculated were poor and inconsistent, and they did not improve within subgroup analysis or when physical findings were combined. Further statistical analysis suggested the physical findings, whether in isolation or in combination, have poor discrimination between patients that have and do not have inhalation injury (AUC < 0.7, P >.05) and poor agreement with the diagnosis made by bronchoscopy (κ < 0.4, P >.05). This remained true in the subgroup analysis as well. Our data demonstrated the findings of SN, CS, and FB are unreliable evidence for inhalation injury, even in the context of an enclosed space mechanism of injury. Thus, these physical findings are not absolute indicators for intubation and should be interpreted as one component of the history and physical. (J Burn Care Res 2015;36: ) Continued advancement in critical care and overall burn management has not alleviated the increased morbidity and mortality from burn inhalation injury among afflicted burn patients. 1,2 In burn inhalation injury, the chemical and thermal insult suffered by the upper and lower airways results in edema, epithelial sloughing, increased mucus secretion, inflammation, atelectasis, and ultimately airway obstruction. 1,3 5 With the potential of airway obstruction, vigilant airway protection is of the utmost importance, leading From the *Division of Plastic Surgery, University of South Florida Morsani College of Medicine, Tampa; University of South Florida Morsani College of Medicine, Tampa; University of South Florida, Tampa; Department of Epidemiology and Biostatistics, University of South Florida College of Public Health, Tampa; and Institute for Tissue Regeneration, Bay Pines VA Healthcare System, Florida. Address correspondence to Jessica A. Ching, MD, Department of Surgery-USF Health, 2 Tampa General Circle, Room 7015, Tampa, Florida Copyright 2014 by the American Burn Association X/2015 DOI: /BCR many to emergently intubate patients with suspected inhalation injury. This is balanced by the known appreciable increase in ventilator-related complications in burn patients and the possibility of unnecessary intubation and risk. 2 Although over 150,000 patients with burn injuries have presented for hospital evaluation in the last 10 years, only 7.7% of these patients have suffered burn inhalation injury. 2 Additionally, acute upper airway obstruction is estimated to occur in one fifth to one third of patients with inhalation injury. 5,6 This means only a small fraction of burn patients require acute intubation for airway protection due to inhalation injury and makes the accurate diagnosis of inhalation injury more challenging and more critical. Inhalation injury is ideally diagnosed by bronchoscopy With bronchoscopy, the airway is directly visualized, making it the accepted standard for the diagnosis of inhalation injury. 7 9,11 However, access to urgent bronchoscopy evaluation is not feasible for many first responders and healthcare practitioners. In 197

2 198 Ching et al January/February 2015 such cases, clinical exam findings are of paramount importance. Thus, the diagnosis of inhalation injury is often based on the clinical exam findings of singed nasal hair (SN), carbonaceous sputum (CS), and facial burns (FB), especially when combined with an enclosed space mechanism of burn injury. 1,6,12 If inhalation injury is suspected, this means the subsequent decision on whether or not to intubate the patient is biased by these same clinical findings. With thousands of burns presenting for evaluation each annum and only a small portion of these having inhalation injury, it is vital that the clinical criteria for intubation are rooted in reliable indications. Although findings of SN, CS, and FB in combination with an enclosed space mechanism of burn injury are commonly referenced as reliable evidence of inhalation injury and the need for intubation, there are no studies which compare these exam findings to the accepted standard of bronchoscopic diagnosis of inhalation injury. Therefore, the purpose of this investigation was to evaluate the utility of SN, CS, and FB as indicators of burn inhalation injury, when compared to the direct bronchoscopic diagnosis of inhalation injury. METHODS Study Location and Design A single American Burn Association (ABA) Verified Burn Center was the site of the study. An institutional review board approved, retrospective chart review of admissions from November 2011 to April 2013 was conducted. Inclusion criteria were: a suspicion of burn inhalation injury was present which prompted a bronchoscopic evaluation during the admission. Patients who were not evaluated by bronchoscopy were excluded, as the presence or absence of inhalation injury findings on bronchoscopic evaluation was used as the accepted standard measurement. Patients less than 18 years old were also excluded. The primary analysis examined the presence on admission of SN, CS, and FB, along with presence or absence of inhalation injury on bronchoscopy during the hospitalization. The secondary analysis examined the presence of SN, CS, and FB along with presence or absence of inhalation injury on bronchoscopy among subgroups of the patient population, according to burn and burn injury mechanism. The specific subgroups chosen for statistical analysis were required to have a minimum of 20 patients each. Data Collection Patients were initially identified by codes from the Ninth Revision of the International Classification of Diseases (ICD-9) associated with burn inhalation injury or skin burns of all size, in conjunction with the Current Procedural Terminology codes for bronchoscopy. Charts were reviewed according to the inclusion and exclusion criteria. Data were then collected retrospectively from the qualifying patient records. Data collected included: age, gender, past medical history, personnel who performed the intubation, percent burn, burn injury mechanism, admission clinical exam findings (SN, CS, FB), and bronchoscopy findings (the presence or absence of inhalation injury). Definitions Physical exam findings were defined as those present on admission evaluation by the Burn Team. These findings were analyzed in isolation and in combination for a total of seven categories: singed nasal hair (SN), carbonaceous sputum (CS), facial burns (FB), singed nasal hair and carbonaceous sputum (SN + CS), singed nasal hair and facial burns (SN + FB), carbonaceous sputum and facial burns (CS + FB), and singed nasal hair with carbonaceous sputum and facial burns (SN + CS + FB). was defined by the assessment of the Burn Physician who responded to the arrival of the patient in the Emergency Department. Inhalation injury was identified as present or absent per the first bronchoscopy evaluation conducted after admission. In patients who arrived intubated, the airway distal to the endotracheal tube was assessed by bronchoscopy, and in those who did not arrive intubated, the entire upper and lower airway was assessed by bronchoscopy. Inhalation injury was classified as present where mucosal erythema, blistering, edema, erosions, or necrosis was found in the airways, or where particulate matter was present in the tracheobronchial tree, as documented in the bronchoscopy report. 7 Conversely, inhalation injury was classified as absent where none of these findings were documented in the bronchoscopy report. Statistical Methods We examined data of inhalation injury according to physical exam findings and bronchoscopy. Bronchoscopic diagnosis was used as the accepted standard for the diagnosis of inhalation injury. Thus, a true positive was defined as presence of the physical exam finding(s) when the bronchoscopic evaluation also suggested inhalation injury, while a true negative was defined as the absence of the physical exam finding(s) when the bronchoscopic evaluation did not suggest inhalation injury. For each physical exam finding category, we assessed the sensitivity (the proportion

3 Volume 36, Number 1 Ching et al 199 that the physical exam findings correctly identified inhalation injury out of the total who had inhalation injury by bronchoscopy) and specificity (the proportion that the absence of the physical exam findings correctly identified a no inhalation injury out of all those who had no inhalation injury by bronchoscopic evaluation). For adjunctive analysis of each physical exam category, the positive predictive value (the proportion of true positives to total positives of the physical exam findings) and negative predictive value (the proportion of true negatives to total negatives of the physical exam findings) were calculated as well. We also used receiver operating characteristic (ROC) analysis methods, based on the sensitivity and specificity, to evaluate how well the presence or absence of the physical exam finding(s) discriminated between a patient that had inhalation injury according to bronchoscopy and a patient that did not have inhalation injury on bronchoscopy. Analysis of the ROC provides a precise and valid measure of diagnostic accuracy uninfluenced by prior probabilities. The area under the ROC curve (AUC) was then compared to standard statistical guidelines, where AUC 0.5 is no discrimination, 0.5 < AUC 0.7 is poor discrimination, 0.7 < AUC 0.8 is acceptable discrimination, 0.8 < AUC 0.9 is excellent discrimination, and AUC > 0.9 is outstanding discrimination. 13 Thus, the minimum threshold for acceptable discrimination is an AUC of 0.7. Statistical significance was indicated when the associated P value was less than.05. Agreement between the assessed physical findings and bronchoscopy was also evaluated using Cohen s kappa (κ), which is based on the number of true positives and true negatives of the physical exam findings. This assessed the degree to which the presence or absence of the physical exam finding(s) agreed with the presence or absence of inhalation injury according to bronchoscopy. Excellent agreement was indicated when κ 0.75, while 0.4 < κ < 0.75 indicated fair to good agreement, and κ 0.4 indicated poor agreement. Statistical significance was indicated when the associated P value was less than.05. All analyses were performed with the use of SAS, version 9.3 (SAS Institute, Inc., Cary, NC). A twosided P value of less than.05 was considered to indicate statistical significance. RESULTS Patients A total of 47 patients met criteria for inclusion in the study. This included 35 male and 12 female patients. Patient age ranged from 18 years old to 78 years old (mean = 51.7 years, median = 54 years, SD = 14.2 years). ranged from 1% to 97% (mean = 29.6%, median = 23%, SD = 27.9%). All patients either arrived intubated or were intubated at some point during the hospital admission. The majority of intubations were performed prior to evaluation by our Burn Team; such intubations were executed by first responders or an outside hospital provider (66%) and our emergency department physicians (11%). According to burn injury mechanism, the largest subgroup was the enclosed space mechanism of injury (n = 20), followed by an open space injury (n = 10), an explosion (n = 10), and a flash flame injury from home oxygen use while smoking (n = 7). Bronchoscopy Comparison Bronchoscopy was normal in 31 patients (66%), and it was consistent with inhalation injury in 16 patients (34%). Bronchoscopic evaluation for inhalation injury was performed within 24 hours of admission in 44 patients. The remaining three patients were closely monitored for suspected inhalation injury and underwent bronchoscopy on hospital day 3, 5, or 6 due to changes in respiratory status or persistent concerns for inhalation injury. Of these three patients, only the patient who underwent bronchoscopy on hospital day 6 had findings consistent with inhalation injury, while the other two patients had normal airway findings. Group analysis was initially performed for all patients collectively (n = 47). Then further subgroup analysis of the physical exam findings was performed to distinguish if the subgroup classification impacted statistical analysis, positively or negatively. The subgroups chosen for more specific analysis were: those patients injured in an enclosed space (n = 20), those patients with burn greater than 20% (n = 26), and those patients with burn less than 20% (n = 21). These specific subgroups were chosen for statistical analysis as each contained a minimum of 20 patients. Additional subgroups according to or other injury mechanisms could not be reliably analyzed as they contained less than 20 patients each. Overall, SN or FB possessed the greatest sensitivity of 0.82 in burns greater than 20%, while FB alone also demonstrated a sensitivity of 0.75 in all patients and those injured in an enclosed space. The combination of CS + FB possessed the highest specificities of 0.74 for all patients and 0.8 for burns greater than 20%, while the combination of SN + CS + FB also had a specificity of 0.71 for all patients and 0.8 for burns greater than 20%.

4 200 Ching et al January/February 2015 The highest positive predictive value of 0.57 occurred in the categories of CS + FB and CS + FB + SN in burns greater than 20% ; all other positive predictive values were less than For the all patients group, CS exhibited a negative predictive value greater than 0.7, while FB had a negative predictive value greater than 0.7 for burns greater than 20%. In those patients with burns less than 20%, CS, SN + CS, CS + FB, and the combination of SN + CS + FB, all had a negative predictive value greater than 0.7. For the enclosed space subgroup, no positive predictive values or negative predictive values were greater than 0.7, and 86% of those calculated for the subgroup were less than or equal to 0.5. The majority of sensitivities, specificities, positive predictive values, and negative predictive values calculated were less than 0.7 and did not increase appreciably when multiple physical exam findings were combined. Of the all the sensitivities, specificities, positive predictive values, and negative predictive values calculated, 59% were 0.5 or less and 86.7% were less than 0.7. No physical exam finding category exhibited sensitivities, specificities, positive predictive values, and negative predictive values greater than 0.7 consistently across all groups analyzed. A summation of the sensitivities, specificities, positive predictive values, and negative predictive values for all patients (Table 1) and for those with an enclosed space mechanism of injury (Table 2) are included for reference. ROC analysis was performed to assess the ability of the presence or absence of the physical exam finding(s), individually or in combination to discriminate between a patient that had inhalation injury according to bronchoscopy and a patient that did not have inhalation injury on bronchoscopy. The ROC analysis was completed for all physical finding categories in the context of all patients, those patients injured in an enclosed space, those patients with burn greater than 20%, and those patients with burn less than 20%. All physical exam finding categories suggest no discrimination or poor discrimination in determining the presence or absence of inhalation injury on bronchoscopy, with all patients combined or in any subgroup analysis (AUC range, ). The AUC values for all patients and subgroups, when physical findings were isolated and combined, did not reach statistical significance (P >.05). Agreement analysis utilizing Cohen s kappa (κ) yielded values below 0.4 for all patients and all subgroups (enclosed space mechanism, burn greater than 20%, and burn less than 20%), which denotes poor agreement between exam findings and the presence or absence of inhalation injury on bronchoscopy. Additionally, all kappa values were nonsignificant (P >.05) for all categories of exam findings with all patients combined as well as for each subgroup. Summations of the ROC (Table 3) and kappa (Table 4) statistical analyses for all patients and all subgroups are included for reference. DISCUSSION The purpose of our retrospective study was to evaluate the utility of SN, CS, and FB as indicators of burn inhalation injury, with bronchoscopic diagnosis of inhalation injury as the standard of comparison. Our data demonstrated the inconsistency of SN, CS, and FB, whether occurring individually or in combination, across all burn patients in our sample, burn injuries greater than 20%, burn injuries less than 20%, and an enclosed space mechanism of burn injury. In comparison with the objective diagnosis of inhalation injury by bronchoscopy, the present statistical analysis yielded no evidence for the use of these physical findings to discriminate between patients that have and do not have inhalation injury, nor did our analysis show adequate agreement between these physical findings and the diagnosis made by bronchoscopy. The lack of reliability of SN, CS, and FB in indicating inhalation injury imparts significant clinical application for those evaluating burn patients without access to urgent fiberoptic bronchoscopy. Table 1. Analysis of the predictive ability of physical characteristics compared to fiberoptic bronchoscopy for the diagnosis of smoke inhalation injury, utilizing sensitivities, specificities, positive predictive values, and negative predictive values for all patients (n = 47) SN CS FB SN+CS SN+FB CS+FB SN+CS+FB Sensitivity Specificity Positive predictive value Negative predictive value

5 Volume 36, Number 1 Ching et al 201 Table 2. Analysis of the predictive ability of physical characteristics compared to fiberoptic bronchoscopy for the diagnosis of smoke inhalation injury, utilizing sensitivities, specificities, positive predictive values, and negative predictive values for patients with a history of a burn in an enclosed space (n = 20) SN CS FB SN+CS SN+FB CS+FB SN+CS+FB Sensitivity Specificity Positive predictive value Negative predictive value The distrust of clinical exam findings or mechanism of burn injury to indicate inhalation injury is longstanding, while quantification of the perceived inconsistency is lacking. Frequently referenced landmark studies include Moylan et al and Moylan and Chan. 14,15 In 1972, Moylan et al 15 found FB, CS, hoarseness, and wheezing to have unreliable value in comparison to an abnormal 133 Xenon lung scan; however, only five of these patients underwent bronchoscopy for definitive diagnosis of inhalation injury. It was echoed by Moylan and Chan soon after with added reservation regarding the association of enclosed space accidents with inhalation injury. 14 As such the inconsistency of FB, CS, and an enclosed space mechanism of injury in indicating inhalation injury has been questioned for some time. The unreliability of the physical findings assessed in the present study may be partially due to the twofold mechanism of inhalation injury: chemical and thermal injury. Chemical burns can occur from the inhalation of toxic gases throughout the tracheobronchial tree and particulate matter deposition in the lower airways. 1 CS may evince smoke exposure and inhaled particulate matter, but it may also be the Table 3. The receiver operating characteristic discrimination analysis for all patients (n = 47), the enclosed space subgroup (n = 20), the burn < 20% subgroup (n = 21), and the burn 20% subgroup (n = 26) All Patients* Enclosed Space* < 20%* 20%* SN CS FB SN + CS SN + FB CS + FB SN + CS + FB Values less than 0.7 indicate poor discrimination, and values less than 0.5 indicate no discrimination. *P >.05. result of carbon deposits in the upper airways rather than the lower airways. 6 CS also does not necessarily indicate high-temperature smoke exposure. 6,10 Thermal injury to the airway results when air in excess of 150 degrees Celsius is inhaled. 1 Normally, the hot air is cooled by the pharynx and very rarely continues into the lower airways at these high temperatures. 10,16 SN and FB are indicators of thermal damage prior to air passage into the pharynx and, thus, do not equate to upper airway thermal injury or acute airway compromise. The inconsistency of SN, CS, and FB indicated by the present data should invoke caution to use of these findings alone in diagnosing inhalation injury and determining the need for intubation. These clinical findings, by the current analysis, are nearly equivalent to flipping a coin. It is possible that with additional burn and patient variables not included in the present study, these clinical findings could prove more accurate. It should be noted there are a multitude of physical exam findings not studied in our data, including hoarseness, stridor, tachypnea, increased work of breathing, and shortness of breath, which warrant further study to assess their utility in diagnosing burn inhalation injury and its associated airway compromise. Unfortunately, the majority of patients included in the study arrived intubated, making it difficult to accurately assess many of the aforementioned additional physical findings. As a retrospective chart review, data collection inherently relied on the accuracy of health provider documentation. This potential for error could be minimized and the study design strengthened with a subsequent prospective study. Also, only patients with suspected burn inhalation injury underwent bronchoscopic evaluation. Although future research with bronchoscopic evaluation of all patients, both with and without suspected burn inhalation injury, would provide more robust data regarding physical exam findings, the procedural risks and healthcare costs associated with bronchoscopy in the absence of suspected burn inhalation injury should be heavily

6 202 Ching et al January/February 2015 Table 4. The Cohen s kappa agreement analysis for all patients (n = 47), the enclosed space subgroup (n = 20), the burn < 20% subgroup (n = 21), and the burn 20% subgroup (n = 26) All Patients* Enclosed Space* < 20%* 20%* SN CS FB SN + CS SN + FB CS + FB SN + CS + FB Values less than 0.4 indicate poor agreement. *P >.05. considered. While bronchoscopy is considered the accepted standard for diagnosing burn inhalation injury, development of a more definitive test would increase the accuracy of future evaluations. Other possibilities for further research consist of a larger sample size, among multiple burn centers, with the inclusion of pediatric patients in order to increase the generalizability of the present results. It is clear the evaluation of inhalation injury should not rely solely on SN, CS, and FB to determine the presence of inhalation injury or airway compromise. Rather, these findings should be interpreted in the context of a full history and physical before proceeding with intubation for suspected inhalation injury. Providers should also consider if there are overriding reasons to intubate that make the presence of inhalation injury obsolete in the intubation decision, such as unresponsive hypoxia, obtundation, or hemodynamic instability. In settings where bronchoscopy is available, it is reasonable for SN, CS, and FB to prompt evaluation of the airway with bronchoscopy; however, if bronchoscopy is not available, then urgent transfer to an ABA Verified Burn Center is recommended in accordance with transfer guidelines. 10 However, in the presence of SN, CS, and/ or FB without evidence of airway compromise, if bronchoscopy is not feasible, close observation for 24 to 96 hours may be equally reasonable. 17 Above all, in patients with suspected burn inhalation injury, intubation is not absolutely required based on findings of SN, CS, and/or FB, even in the context of an enclosed space burn injury. CONCLUSION Contrary to the classic tenet that SN, CS, and FB consistently indicate the presence of inhalation injury, especially in an enclosed space burn injury, our data suggest these findings have poor discrimination ability and poor agreement with the bronchoscopic diagnosis of inhalation injury. Thus, these findings should not be interpreted in isolation but as one component of the history and physical to avoid unnecessary intubation and risk. References 1. Palmieri TL. Inhalation injury: research progress and needs. J Burn Care Res 2007;28: American Burn Association. National burn repository report. Chicago, IL:American Burn Association Abdi S, Evans MJ, Cox RA, Lubbesmeyer H, Herndon DN, Traber DL. Inhalation injury to tracheal epithelium in an ovine model of cotton smoke exposure. Early phase (30 minutes). Am Rev Respir Dis 1990;142(6 Pt 1): Cox RA, Burke AS, Soejima K, et al. Airway obstruction in sheep with burn and smoke inhalation injuries. Am J Respir Cell Mol Biol 2003;29(3 Pt 1): Haponik EF, Meyers DA, Munster AM, et al. Acute upper airway injury in burn patients. Serial changes of flowvolume curves and nasopharyngoscopy. Am Rev Respir Dis 1987;135: Mlcak RP, Suman OE, Herndon DN. Respiratory management of inhalation injury. Burns 2007;33: Hassan Z, Wong JK, Bush J, Bayat A, Dunn KW. Assessing the severity of inhalation injuries in adults. Burns 2010;36: Hunt JL, Agee RN, Pruitt BA Jr. Fiberoptic bronchoscopy in acute inhalation injury. J Trauma 1975;15: Masanes MJ, Legendre C, Lioret N, Maillard D, Saizy R, Lebeau B. Fiberoptic bronchoscopy for the early diagnosis of subglottal inhalation injury: comparative value in the assessment of prognosis. J Trauma 1994;36: Gallagher JJ, Herndon DN. Controversy in inhalation injury and burn resuscitation. Emerg Med Crit Care Rev 2007: Wanner A, Cutchavaree A. Early recognition of upper airway obstruction following smoke inhalation. Am Rev Respir Dis 1973;108: ABA Evidence-based Guidelines Group, Saffle J. Inhalation injuries: diagnosis. J Burn Care Rehabil 2001(Suppl):S Hosmer DW, Lemeshow S. Applied logistic regression. Wiley series in probability and statistics: texts and references section. 2nd ed. New York: Wiley; 2000: Moylan JA, Chan CK. Inhalation injury an increasing problem. Ann Surg 1978;188: Moylan JA Jr, Wilmore DW, Mouton DE, Pruitt BA Jr. Early diagnosis of inhalation injury using 133 xenon lung scan. Ann Surg 1972;176: Moritz AR, Henriques FC, McLean R. The effects of inhaled heat on the air passages and lungs: an experimental investigation. Am J Pathol 1945;21: Madnani DD, Steele NP, de Vries E. Factors that predict the need for intubation in patients with smoke inhalation injury. Ear Nose Throat J 2006;85: