Management of Foreign Bodies Obstructing the Airway in Children FREE

The aspiration and ingestion of foreign bodies (FBs) presents a potential lethal threat to infants and children. Approximately 2.5 million children across the United States are affected by FB aspirations each year, with up to 2000 deaths.¹ Most of these children are younger than 3 years and are at higher risk of inhaling or ingesting FBs owing to their tendency to place objects in their mouths, poor swallowing coordination, and immature dentition.² The development of federal regulations has reduced the frequency of FB aspirations and related mortalities.^3- 4 However, the number of mortalities associated with these cases remains unacceptably high.⁴ Pediatric deaths do occur annually from asphyxiation caused by children's products; unfortunately, approximately one-fourth of products associated with mortality passed criteria established by the Consumer Product Safety Commission.⁴ Furthermore, studies have shown that 50% of patients with an FB aspiration lacked causal history, and 20% of children were misdiagnosed and improperly given medications for at least 1 month before the correct diagnosis was made.² Delay of proper treatment inevitably increases the risk of complications and mortality in these patients.⁵ In 2001, it was estimated that among children younger than 14 years, 1 child died for every 110 children who were treated for a choking-related episode.⁶

To improve patient outcomes and to develop effective health care interventions, understanding diseases on a national scale is imperative. The Kids' Inpatient Database (KID) established by the Agency for Healthcare Research and Quality contains information on patient demographics and the use of hospital services by children in the United States.⁷ The KID documents procedures, outcomes, and charges of pediatric patients, making it suitable for the study of demographics and resource use among infrequently encountered disease processes.⁷

Cases involving FB aspirations require prompt and efficient evaluation and treatment. With the declining number of FB aspirations managed by health care providers, there are legitimate concerns regarding the training of physicians in these critical pediatric cases.⁵ Bronchoscopic removal of FBs is standard in the treatment of these children; however, the dearth of actual cases makes this skill difficult to learn and to assess in trainees.⁸ To resolve this deficiency, alternative educational paradigms for FB retrieval may need to be considered. Our goal is to provide a national perspective on the management of airway FBs (A-FBs) and esophageal FBs (E-FBs) that obstruct the airway in children.

Institutional review board approval was obtained. The International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes E911 (inhalation and ingestion of food causing obstruction of respiratory tract or suffocation) and E912 (inhalation and ingestion of other object causing obstruction of respiratory tract or suffocation) were used as inclusion criteria. E911 diagnoses were considered organic/food FBs, and E912 diagnoses were considered inorganic/nonfood FBs. There were no a priori exclusion criteria. To assess the rate of positive findings for A-FBs and E-FBs, we used the principal procedure to determine the cases in which FBs were removed as the numerator and the total number of respective interventions as the denominator. The ICD-9-CM procedure codes were used to assess the following: bronchoscopy (33.22, 33.23, and 33.24), bronchoscopy with FB removal (98.15), laryngoscopy or tracheoscopy (31.42), esophagoscopy (42.23 and 45.13), esophagoscopy with FB removal (98.02), FB removal from larynx, pharynx, or site not otherwise specified (98.13, 98.14, and 98.20), and tracheotomy (31.1).

The data source was the 2003 KID, which is part of the Healthcare Cost and Utilization Project sponsored by the Agency for Healthcare Research and Quality. The KID is designed to study the use of hospital services by examining data on diagnoses, outcomes, and charges. It samples 2 984 129 pediatric discharges from 3438 hospitals in 36 states. Pediatric discharges were defined as all discharges in which patients were 20 years or younger on admission. Community, nonrehabilitation hospitals, academic medical centers, and pediatric hospitals were included in this database. Hospital-specific information, as defined by the American Hospital Association, was obtained from the KID and included hospital location, teaching or academic status, type of hospital (nonchildren’s, children’s, or children's unit), and source of admission (emergency, outside hospital, or routine). Discharge weights were used to weight discharges in the KID core files to pediatric discharges from all US community, nonrehabilitation hospitals to obtain national estimates.⁷ Univariate analyses were performed on all admissions comparing pertinent variables with total charges. Separate analyses were also conducted to evaluate increased resource use for mortalities. The P value indicates the extent that the variable is an independent predictor of increased total charges.

It is imperative to underscore that using KID to assess resource utilization assumes that total charges are used as a surrogate for in-hospital resource utilization. This methodology has been previously demonstrated and was done similarly in this study.⁹ To assess resource utilization in this study, we used the variable of total charges, which includes hospital charges for each hospital stay but does not include the amounts actually paid or the physician charges.

Furthermore, the data from KID can be reported as either unweighted or weighted data. The distinction is that unweighted data represent the actual frequencies of each variable, while the weighted data can be used to predict national estimates. For this study, we used the weighted variables to present the national estimates and unweighted variables, where appropriate, to allow presentation of the actual frequencies of the reported variables. Data on otolaryngology residency programs were taken from the Accreditation Council of Graduate Medical Education records from the 2003-2004 academic year.¹⁰ Pertinent data were not available before that year. Data were acquired on the number of programs and residents on duty in the states that contributed to KID.

A total of 2771 patients (1642 unweighted) (59% male) were identified using ICD-9-CM diagnosis codes E911 and E912 for A-FBs and E-FBs. Foreign bodies were categorized as organic/food (E911, n = 1168 [42%]) and inorganic/nonfood (E912, n = 1603 [58%]). The data that follow are weighted data from the 2003 KID. The average age of patients with organic FBs was 2.5 (SE, 0.19) years, and for other FBs, it was 4.3 (SE, 0.21) years. The mean (SE) age of all patients was 3.5 (0.17) years, with 55% of the patients younger than 2 years. The average length of stay was 6.4 (SE, 0.6; median, 1.5) days, and 2.4 (SE, 0.1; median, 1.3) procedures were performed per patient. Forty-one percent of patients were admitted to nonchildren's hospitals, and 28% were admitted to children's hospitals. The mean (SD) total charge for all patients was $34 652 ($3453).

Univariate analyses revealed differences in mean total charges within subgroups (Table 1). Asians had higher mean total charges (P = .05) than other races. Seventy-one percent of patients were admitted to teaching hospitals and had higher mean total charges than patients at community hospitals (P < .001). The emergency department was the primary source of admissions (52.0%). Outside hospital admissions (13.5%) yielded the highest charges compared with other admission sources (P = .002). Children's hospitals had higher charges than other hospitals (P = .03). Patients treated at urban hospitals had higher mean charges than those treated at rural hospitals (P < .001). There was notable geographic variation of mean total charges (P < .001), with patients in the western part of the country (30%) accruing the highest average charges at $49 122 (SD, $8505).

Five states accounted for 51% of all A-FB and E-FB admissions: California (20%), New York (9%), Texas (9%), Florida (8%), and Missouri (5%). The highest average charges per admission were in Missouri ($78 332; SD, $29 317), Hawaii ($74 884; SD, $47 749), and Nevada ($69 105; SD, $28 500). Figure 1 shows the distribution of patients and charges by state.

In the admissions studied, 2502 otolaryngology procedures were performed (Figure 2). Of these, 822 were bronchoscopy procedures and 488 were bronchoscopy procedures with FB removal, yielding a positive bronchoscopy rate of 37% for retrieval of an FB. Laryngoscopy with or without tracheoscopy totaled 377 procedures; this total was obtained from the ICD-9-CM procedure codes. There were 370 esophagoscopy procedures and 325 esophagoscopy procedures with FB removal, yielding a positive esophagoscopy rate of 46%. In all A-FB and E-FB cases with documented procedures, 52.4% of the procedures (n=1310) were bronchoscopy procedures, 27.8% (n=695) were esophagoscopy procedures, 15.1% (n=377) were laryngoscopy and/or tracheoscopy procedures, and 1.7% (n=42) were tracheotomy procedures.

There were 94 mortalities, which is 3.4% (n = 2771) of the study population. In the weighted analysis, 42.77 of deaths (3.7%) were in patients diagnosed with food FBs, and 51.62 deaths (3.2%) were in children with other FBs. The proportion of deaths in the 2 groups were similar (χ²P value, .65). The mean (SE) age was 4.6 (0.9) years, and 59% of the patients were male. The average length of stay was 11.7 (SE, 2.7) days, and, on average, 6.2 (SE, 0.7) procedures were performed per patient. The mean (SD) total charges for these patients were $116 792 ($30 270). A large proportion of mortalities were admitted to teaching hospitals, with mean (SD) total charges of $113 099 ($33 478). Univariate analysis of the mortalities revealed that only the geographic location had a significant variation (P = .01) in mean total charges (Table 2).

A total of 1147 teaching hospitals were identified in 35 states. Based on resident census,¹⁰ approximately 873 otolaryngology residents were on duty in these states during 2003, and, on average, each resident would have ideally managed 1.3 pediatric FB admissions assuming an equal distribution of cases. Regional variations were noted, with residents managing between 0.2 to 3.9 cases per year, based on their location of training.

Airway FBs and E-FBs are infrequently seen in children, accounting for 0.06% of all admissions documented in the KID. Despite that, more than 3% of this population died during hospitalization. This report presents a novel analysis of demographic and national data on the management and resource utilization of A-FBs and E-FBs.

Previous studies have indicated that organic matter, including seeds and nuts, is the most common FB removed in children with A-FBs.^2,5- 6 Our data show an increase in inorganic FBs recovered from children, at almost 60% of all FB admissions; however, we cannot ascertain from the data set whether these FBs were obstructing the airway from the airway or the esophagus. In industrialized nations such as the United States, the frequent use of small parts in the toy industry suggests an increase in inorganic FBs ingested and inhaled by children. Furthermore, primary care physicians and the lay public have a reasonable understanding of what types of food are safe for children; this essentially results in a decrease in the rate of organic FBs.¹¹ There are federal standards that define the parameters of safety for such items as rattles and pacifiers; however, many small objects still pass inspection and present a lethal threat to infants.³ Prior studies have shown that children younger than 3 years are at higher risk for A-FBs and E-FBs.² Our data corroborate this finding, as 80% of children with recovered organic FBs were younger than 3 years. However, only about 60% of children with inorganic FBs were younger than this age. In our study, the average age of children with recovered inorganic FBs was 4.3 years, which may suggest a need to increase the suspicion of FB inhalation or ingestion in older children who present with respiratory distress.

The analysis of resource utilization presented several interesting findings on the national level. Admissions that were from outside hospitals (ie, transfers to another hospital) had higher resource utilization, which is potentially explained by the complexity of these cases on arrival to the referred hospital. Race was found to be a determinant of increased resource utilization; however, the implications of this finding are difficult to interpret and to place in a clinical context for an emergency such as an A-FB or an E-FB. Increased resource utilization was also observed in children's hospitals and teaching hospitals. This increase most likely is explained by the highly specialized care provided by these hospitals and the obvious selection bias in which more difficult cases are eventually referred to these institutions, thereby increasing resource utilization. There is also geographic variation of resource utilization, which was significantly higher in the West and Midwest. The highest charges were observed in Missouri and Hawaii, while the lowest charges were in South Dakota and Oregon. This information is difficult to explain, although further, more detailed studies may be able to elucidate these findings.

In many surgical procedures that are performed in an exploratory manner, there is an accepted rate of not finding a lesion. However, no such data exist for performing a bronchoscopy and esophagoscopy for A-FBs or E-FBs. There is no established standard measure of the rate of positive FB findings for bronchoscopy and esophagoscopy. Such a measure would allow comparison of care from one center to another. Our data show a 33% rate of positive FB finding for bronchoscopy and 46% for esophagoscopy. These figures permit institutions to compare and analyze their algorithmic approaches to FB management and result in improved patient outcome. However, because of the inherent weaknesses of a database study, these numbers should be cautiously interpreted and perhaps compared with other studies that may hopefully be able to obtain such a metric.

The percentage of mortalities in our study was found to be higher than that in other studies.⁵ More than 3% of the patient population died, meriting further analysis of discriminating factors in this cohort. These patients experienced longer lengths of stay and more procedures, which is intuitive, as their cases were likely more complex. It is plausible to conclude that these cases were the most severe and acute cases. A higher proportion of patients in this population were referred to teaching hospitals, suggesting a possible selection bias of critical cases to these institutions for specialized care. As anticipated, resource utilization of mortalities was higher than that of the total population of A-FBs and E-FBs.

Training of residents and fellows occurs primarily through the experience of actively treating patients. It has been contended that the incidence of A-FBs and E-FBs is decreasing, thus presenting a challenge to the proper training of resident otolaryngology physicians. In 2003, there were 873 otolaryngology residents working in academic hospitals across the 36 states evaluated in the KID. The average number of A-FBs and E-FBs managed per resident was 1.3. Despite the fact that most cases were referred to teaching hospitals, our data suggest that residents need further training to properly address A-FBs and E-FBs. This concern has been discussed in the literature regarding stapedectomy procedures in which learning curves were established to assess proficiency.¹² Similar estimations may be warranted in A-FB and E-FB management, and it may be that residents need to receive adequate training in the management of A-FBs and E-FBs via alternative educational paradigms. Furthermore, the criteria for competency in otolaryngology residency and pediatric otolaryngology fellowship programs may need to be evaluated to account for the infrequency of A-FBs and E-FBs.

With the ability to emulate lifelike situations, the use of surgical simulation has developed as an educational tool. This method of training provides substantial benefits, including proficiency, without patient risk and enhanced operational competency.^13- 14 Animal-based laboratory simulations have been designed for bronchoscopic FB removal¹⁴; however, in 1988, only 48% of the 105 otolaryngology residency programs had such simulators (more recent data are lacking).⁵ Studies have shown that bronchoscopy simulation is an effective means of skill development, although there may be some limitations of skill translation to the operating room.^8,13- 14 Consideration of bronchoscopy simulation as a standard method of training should cautiously be evaluated for residency training programs.

Inherent limitations to our study and the use of the KID should be noted. Namely, ICD-9-CM diagnosis codes were used as inclusion criteria for the data compiled in the KID, and accuracy beyond what was reported in this database cannot be confirmed because the data entry depends on accurate coding at a hospital level. Possible biases may have existed in coding information, and some patient information was missing or not reported by hospitals. Specifically, our choice of inclusion criteria, codes E911 and E912, may have introduced bias because there are non–E codes that may have been used (933, FB in the pharynx and larynx; 934, FB in the trachea, bronchi, and lungs; and 935, FB in the mouth, esophagus, and stomach). We concluded that using the E codes would be a more accurate estimation of the severity of the FB because the descriptor for codes E911 and E912 state that the FB was obstructing the airway (however, it does not specify whether the FB was an A-FB or an E-FB). Other sources of bias include the stipulation that, in the data use agreement, when the frequency of a variable is less than 10 sampled admissions, we are not to discuss this variable because it is not statistically valid. It could be that not further examining these variables might have led to potential bias because some potentially important variables were mitigated.

In conclusion, pediatric A-FBs and E-FBs that obstruct the airway occur infrequently. Most cases are referred to teaching institutions. In patients admitted with a diagnosis of airway obstruction from an A-FB or an E-FB, the rates of positive findings at surgery were 33% and 46%, respectively. A surprisingly high mortality rate of 3.4% was noted. Alternative education measures should be considered to train physicians in the management of this infrequent, potentially lethal condition.

Correspondence: Rahul K. Shah, MD, Division of Otolaryngology, Children's National Medical Center, 111 Michigan Ave NW, Washington, DC 20010 (rshah@cnmc.org).

Submitted for Publication: May 8, 2009; final revision received September 17, 2009; accepted October 29, 2009.

Author Contributions: Drs Shah, Patel, and Lander had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Shah, Patel, and Choi. Acquisition of data: Shah, Patel, and Lander. Analysis and interpretation of data: Shah and Patel. Drafting of the manuscript: Shah and Patel. Critical revision of the manuscript for important intellectual content: Shah, Patel, Lander, and Choi. Statistical analysis: Lander. Administrative, technical, and material support: Shah. Study supervision: Shah and Choi.

Financial Disclosure: None reported.

Previous Presentation: This study was presented in part at the American Society of Pediatric Otolaryngology Annual Meeting; May 25, 2009; Seattle, Washington.

Additional Contributions: We acknowledge Carol Stocks, Agency for Healthcare Quality and Research, for the review of the manuscript and the integrity of the sampling.