A National Overview of Surgical Misadventures in Head and Neck Surgery: “Oh no, you cut it”

Abstract

Objectives:

To examine the risk and prevalence of accidental intraoperative injury reported during head and neck surgeries and the associated outcomes.

Methods:

A retrospective cross-sectional analysis utilizing the Nationwide Readmissions Database, 2010 – 2014. Adults patients with a reported accidental intraoperative injury were compared to controls without such injuries.

Results:

A total of 173 cases and 105,659 controls were included. Most cases were reported in surgeries of the mouth/ tonsils (29.4%) and maxillofacial bones/mandible (22.5%). The remaining cases were reported in surgeries of the pharynx/larynx (17.5%), nose/paranasal sinuses (15.4%), salivary glands & ducts (6.2%), thyroid/parathyroid (5.2%), and ear (3.8%). The multivariate logistic regression model demonstrated that surgeries of the pharynx/larynx were associated with the highest risk of injuries compared to other site surgeries [OR: 2.51, 95%CI: (1.49, 4.25), p<0.001]. Concomitant neck dissection was also independently associated with the risk of injury [OR: 4.07, 95%CI: (2.05, 8.09), p<0.001]. Compared to controls, cases were not associated with an increased risk of mortality (p=0.63) or readmission (p=0.29), however, those cases had a significantly longer hospital-stay on average by 3.64±0.95 days/case (p<0.001) and a higher cost of treatment, on average by $13,478±119.42/case (p<0.001).

Conclusions:

This study reports on the prevalence and outcomes of accidental intraoperative injuries reported in head and neck surgeries. The prevalence is relatively low, and the annual trend appears stable, however, it is associated with a significant burden on the health system.

Introduction

Surgery-related accidents represent 39.6% of adverse events reported during hospitalization,¹ and it is estimated 15.5% of operative accidents are related to a technical complication or surgical failure.² We recently showed that intraoperative injuries of nearby structures reported in surgeries of ear, nose, mouth, and pharynx constitute 0.3% of all surgery-related injuries in the United States.³ Nonetheless, the accidental nature of such injuries renders even one event singularly disquieting. An intraoperative unintentional injury could result in significant harm and mortality,^3,4 and could potentially affect the surgeon and hospital reputation and develop into a malpractice litigation.^1,5

The complex anatomy of the head and neck region and the proximity of multiple vital structures make most surgical interventions carry an inherent risk of accidental injury of nearby structures, such as the facial nerve, recurrent laryngeal nerve, thoracic duct, salivary duct, sigmoid sinus, tegmen, jugular bulb, and carotid artery.^6–11 No previous study has addressed the overall prevalence of such injuries in otolaryngology practice at the national level. The purpose of this study is to provide a general perspective on accidental intraoperative injury of nearby structures reported in head and neck surgeries in the United States. We used national data to examine the risk factors, associated outcomes, and trend.

Methods

This study is a retrospective cross-sectional analysis using the Nationwide Readmissions Database (NRD) for the years 2010 – 2014. The NRD is part of the Healthcare Cost and Utilization Project (HCUP), sponsored by the Agency for Healthcare Research and Quality. ¹² The NRD is a unique and powerful database designed to support various types of analyses of national readmission rates for all payers and the uninsured. ¹² The database includes discharge data from 27 geographically dispersed States, accounting for 57.8% of the total U.S. resident population and 56.6% of all U.S. hospitalizations. ¹² Additionally, the database allows for a weighted analysis that has been utilized in this project to provide a better estimation of national outcomes. ¹² The NRD tracks patient across the sampled hospitals within a calendar year for any readmission. ¹² This database addresses a large gap in health care data-the lack of nationally representative information on hospital readmissions for all ages. ¹² The NRD consists of a publicly available de-identified data that does not meet the definition of human subject research.¹²

The study main objectives were to investigate the risk, trend, and outcomes of accidental intraoperative injuries in patients who underwent otolaryngologic, and head and neck surgeries.

The study population was defined as adult patients (≥18 years) who underwent head and neck surgeries as a primary procedure (Appendix A). International Classification of Disease, 9th Revision (ICD-9) is used in the NRD to code for the procedures and diagnoses reported during a hospital admission. ¹² The extracted study sample was then queried for patients with a reported accidental intraoperative injury using the ICD-9 codes: E87.00 (Accidental cut, puncture, perforation or hemorrhage during surgical operation), and 998.2 (Accidental puncture or laceration during a procedure, not elsewhere classified). Patients without those ICD-9 codes reported during their admission were considered controls. All records were checked for completeness of data. Records with missing values for the study parameters were excluded.

The outcomes of interest in this study were: (i) the risk of accidental intraoperative injury of nearby structures, (ii) in-hospital mortality either during admission or during readmission within 30 days postoperatively, (iii) length of hospital stay (LOS) in days, the total length of hospital stay of the first admission and readmission if occurred, and (iv) cost of health services-the database contain hospital charges associated with each admission as well as a cost-to-charge ratio determined for each hospital that allows for conversion of charges values to cost values. ¹² All cost values were adjusted for the inflation rate to reflect 2017 U.S dollar value using Bureau of Labor Statistics inflation calculator. ¹³

Factors that were examined for their independent association with the risk of injury were: (i) Age: 18 – 40, >40 – <61, 61 – <75, ≥75 years, (ii) gender: Male, Female, (iii) modified Charlson comorbidity index score (CCIS), classified into: 0, 1, ≥2,¹⁴ (iv) body mass index (BMI): <25, ≥25, (iv) diagnosis of a head and neck cancer, (v) trauma as the reason for Admission, (vi) site of primary procedure, site is classified based on ICD-9 code (Appendix A), (vii) concomitant neck dissection: not performed, performed, (viii) intraoperative nerve monitoring in thyroid/parathyroid and salivary glands surgeries: not reported, reported, (ix) image-guided nose/paranasal sinuses surgery: not reported, reported, (x) robotic-assisted thyroid/parathyroid and mouth/tonsil surgery: not reported, reported, (xi) hospital volume defined as the number of head and neck surgeries performed in each hospital per year, classified by applying quartile classification to avoid bias into: low-volume hospitals (<25^th percentile: 1 – 32 surgeries/yr.), intermediate-volume hospitals (25^th – <75^th percentiles: 33 – 145 surgeries/yr.), high-volume hospitals (≥75^th percentile: ≥146 surgeries/yr.), and (xii) location and type of hospital as coded in the database into: metropolitan non-teaching, metropolitan teaching, non-metropolitan.¹²

Statistical analysis used weighted data to generate a more accurate national estimate. ¹² The records’ weights are available in the NRD and calculated based on the stratification variables that were used in the sampling methodology.¹²

Statistical analyses included cross-tabulation and Chi-square tests that were performed to assess the association of each factor of interest with the outcomes. Factors that demonstrated a significant association with the outcomes in the univariate analysis were included in the multivariate logistic regression models. Odds Ratio (OR) and 95% Confidence Interval (95% CI) were calculated from multivariate logistic regression models. Multivariate linear regression models were performed to compare means of LOS and cost of healthcare services while controlling for confounders. Linear regression was also used to examine the trend in intraoperative injury and associated LOS and cost. The level of significance was set as (α= 0.05). All data analyses were performed using SAS 9.4 for Windows (SAS Institute Inc., Cary, NC, USA).

Results

A total of 173 cases with a reported accidental intraoperative injury were identified and 105,659 controls who underwent comparable procedures were included (Table 1). The mean age of the study population was 47.70±0.16 years old. Patients who underwent otolaryngologic procedures in the United States were 46.3% male and 53.7% female patients. Trauma admissions represented 18.1% of all admissions. The overall perioperative mortality risk was reported in 0.2% of the sample. The distribution of injury prevalence (not risk) based on surgery site demonstrated that the main case-load was in surgeries that involved the mouth/tonsil (29.4%), maxillofacial bones/mandible (22.5%), pharynx/larynx (17.5%), and nose/paranasal sinuses (15.4%) (Figure 1).

Table 1.

Descriptive statistics of the study population

	Sample population, (%) (N=105,832)	Accidental intraoperative injury (%)
		Not reported (n=105,659)	Reported (n=173)	Pa
Age (yr.)
18 – 40	37.2	37.2	30.8
>40-<61	36.2	36.2	33.6
61 – <75	18.8	18.8	21.8
≥75	7.8	7.8	13.8	0.06
Gender
Male	46.3	46.3	53.0
Female	53.7	53.7	47.0	0.14
CCIS
0	71.2	71.2	64.8
1	21.6	21.6	24.6
≥2	7.3	7.3	10.6	0.22
Body mass index
<25	90.4	90.4	90.4
≥25	9.6	9.6	9.6	0.99
Head and neck cancer
No	84.0	84.0	75.2
Yes	16.0	16.0	24.8	0.015
Trauma admission
No	81.9	81.9	80.4
Yes	18.1	18.1	19.6	0.72
Site of surgery
Thyroid/Parathyroid	36.8	36.8	5.2
Ear/Skull base	3.6	3.6	3.8
Nose/Paranasal sinuses	9.2	9.2	15.4
Mouth/Tonsil	19.4	19.4	29.4
Salivary glands & ducts	6.1	6.1	6.2
Pharynx/Larynx	4.7	4.7	17.5
Maxillofacial bones/Mandible	20.3	20.3	22.5	<0.001
Neck dissection
Not performed	95.6	95.6	79.6
Performed	4.4	4.4	20.4	<0.001
Intraoperative nerve monitoring (Thyroid/Parathyroid)
Not reported	99.5	99.5	100.0
Reported	0.5	0.5	0.0	NAb
Intraoperative nerve monitoring (Salivary glands & ducts)
Not reported	97.6	97.6	100.0
Reported	2.4	2.4	0.0	NAb
Image-guided surgery (Nose/Paranasal sinuses)
Not reported	99.6	99.6	100.0
Reported	0.4	0.4	0.0	NAb
Robotic-assisted surgery (Thyroid/Parathyroid)
Not reported	100.0	100.0	100.0
Reported	0.0	0.0	0.0	NAb
Robotic-assisted surgery (Mouth/Tonsil)
Not reported	99.8	99.8	100.0
Reported	0.2	0.2	0.0	NAb
30 days Postoperative death
No	99.8	99.8	99.3
Yes	0.2	0.2	0.7	0.22
Hospital volume (surgeries/yr.)
Low: 1 – 32	25.8	25.8	25.0
Intermediate: 33 – 145	50.3	50.3	52.7
High: ≥146	23.9	23.9	22.3	0.89
Hospital type
Metropolitan non-teaching	29.8	29.8	28.6
Metropolitan teaching	65.1	65.1	69.5
Non-metropolitan hospital	5.1	5.1	1.8	0.30

Abbreviations: CCIS, charlson comorbidity index score; NA, not applicable.

^aChi-square test.

^bNot applicable because no event is reported in the injured group.

Figure 1:

Distribution of accidental intraoperative injury by the site of surgery.

The univariate analyses demonstrated the following factors to be associated with accidental intraoperative injury (Table 1) (Figure 2): diagnosis of head and neck cancer, site of surgery, and neck dissection. However, only the site of surgery and concomitant neck dissection operations retained the significant association with the risk of injury in the multivariate analysis (p<0.05 each) (Table 2). The multivariate logistic regression model identified three level of risk of accidental injury based on the site of surgery: (i) high-risk surgery sites: pharynx/larynx (risk: 0.71%); (ii) medium-risk surgery sites: mouth/tonsil (risk: 0.29%), ear (risk: 0.20%), nose/paranasal sinuses (risk: 0.32%), salivary glands & ducts (risk: 0.19%), and maxillofacial bones/mandible (risk: 0.21%); and (iii) low-risk surgery site: thyroid/ parathyroid (risk: 0.03%). Notably, hospital volume and hospital teaching status did not associate with the risk of accidental intraoperative injuries (Table 1).

Figure 2:

Unadjusted risk and 95% confidence intervals of accidental intraoperative injury by the site of surgery.

Table 2.

Risk of accidental intraoperative injury in patients who underwent head and neck surgeries.

Factor		% Accidental intraoperative injury	ORa	95% CI	P
Head and neck cancer
	No	0.17	Reference
	Yes	0.30	1.02	0.56, 1.88	0.95
Site surgery
	Mouth/Tonsil	0.29	Reference
	Thyroid/Parathyroid	0.03	0.12	0.06, 0.24	<0.001
	Ear	0.20	0.78	0.36, 1.67	0.52
	Nose/Paranasal sinuses	0.32	1.39	0.76, 2.54	0.29
	Salivary glands & ducts	0.19	0.52	0.25, 1.08	0.08
	Pharynx/Larynx	0.71	2.51	1.49, 4.25	<0.001
	Maxillofacial bones/Mandible	0.21	0.89	0.50, 1.58	0.68
Neck dissection
	Not performed	0.16	Reference
	Performed	0.88	4.07	2.05, 8.09	<0.001

Abbreviations: OR, adjusted odds ratio; CI, confidence interval; CCIS, charlson comorbidity index score.

^aThe model includes: cancer status, surgery site, and neck dissection.

Among the 12 injuries reported in thyroid/parathyroid surgeries, none of them was robotic-assisted surgery or utilized intraoperative nerve monitoring (Table 1). Also, among the 27 injuries reported in nose/paranasal sinuses surgeries, none of them used image-guided technology. Of the 44 injuries reported in surgeries that involved the mouth/tonsil, none of them was robotic-assisted surgery (Table 1). The absence of injury cases among patients in whom intraoperative nerve monitoring was used or had robotic-assisted or image-guided surgeries prevented performing statistical analysis (the database did not have cases to compare to controls).

The mortality risk in patients who experienced accidental intraoperative injury during otolaryngologic surgeries was not significantly different from controls (0.7% vs. 0.2%, p=0.63), similarly, the readmission risk was also not significantly different (6.3% vs. 3.3%, p=0.29) (Table 3). However, cases had a significantly longer hospital stay on average by 3.64±0.95 days/case (p<0.001) and a higher cost of health services on average by $13,478±119.42/case (p<0.001) (Table 3).

Table 3.

Risk of selected outcomes of cases with a reported accidental intraoperative injury of nearby structure compared to controls underwent similar procedures.

Outcome	Accidental intraoperative injury	%	ORa	95%CI	P
30 days Postoperative deatha	Not reported	0.2	Reference
	Reported	0.7	1.64	0.22, 12.49	0.63
30 days readmissionb	Not reported	3.3	Reference
	Reported	6.3	1.50	0.71, 3.19	0.29
		Mean (SEM)	Pc
Length of stay (days)	Not reported	3.91 (0.07)	Reference
	Reported	7.55 (0.95)	<0.001
Cost ($)	Not reported	16,319.00 (41.10)	Reference
	Reported	29,797.00 (115.05)	<0.001

Abbreviations: OR, adjusted odds ratio; CI, confidence interval; SEM, standard error of the mean.

^aLogistic regression model includes: accidental intraoperative injury, age, gender, charlson comorbidity index score, head and neck cancer, trauma status, site surgery, hospital volume, and hospital teaching type.

^bLogistic regression model includes: accidental intraoperative injury, age, gender, charlson comorbidity index score, head and neck cancer, trauma status, and site of surgery.

^cLinear regression model includes: accidental intraoperative injury, age, gender, charlson comorbidity index score, head and neck cancer, trauma status, site surgery, hospital volume, hospital teaching type, and (body mass index in the cost model only)

Extrapolating the incidence of accidental intraoperative injury to an annual rate per 10,000 otolaryngologic procedures performed in the United States demonstrated a downward trend from 22.1 injuries/10,000 procedure in 2010 to 15.3 injuries/10,000 procedures in 2014, however, the trend was not significant (p=0.73) (Figure 3). The burden of this incidence rate in terms of hospital stay translated into an additional 80.3 days/10,000 procedures in 2010 and decreased to 55.8 days/10,000 procedures in 2014, while the burden in terms of cost of health services was $0.3 million/10,000 procedures in 2010 and decreased to $0.21 million/10,000 procedures in 2014 (Figure 3). Another perspective on the burden of those injuries, we calculated that on average there are 48 accidental intraoperative injuries annually in the United States, and those 48 cases alone on average caused an additional annual hospital-stay of 176.2 days and an additional cost of $652,874.32.

Figure 3:

Incidence rate of accidental intraoperative injury and the associated additional length of hospital-stay and cost per 10,000 procedures.

Discussion

This study provides an overall and general perspective of accidental intraoperative injury reported during otolaryngologic procedures in the United States. The study utilized The Nationwide Readmissions Database for the period of 2010 – 2014. The highest prevalence of those injuries was reported in procedures that involved the mouth, tonsil, mandible, maxillofacial bone, pharynx, larynx, nose, and paranasal sinuses. While the prevalence represents the case-load, the multivariate logistic regression model comparing cases to controls identified procedures with a risk of such injuries. Based on the model, surgeries of the pharynx and larynx are associated with the highest risk of intraoperative injury. Although a cancer diagnosis was not associated with the risk of injury in this study, our study showed a neck dissection was independently associated with the risk of intraoperative injury.

In a previous study by our group that utilized a different national database, the Nationwide Inpatient Sample (NIS), we identified 213 iatrogenic injuries reported in surgeries that involved the ear, nose, mouth, and pharynx out of 61,663 total injuries reported in all surgical specialties in the United States for the years 2003 – 2010.³

In a retrospective study by Hohman et al., they reported on the risk of iatrogenic facial nerve injuries of patients referred to the Facial Nerve Center of Massachusetts Eye and Ear Infirmary over the period of 2002 to 2012.⁸ The study identified 102 facial nerve injury out of 1,810 patient records.⁸ Albeit the broader definition of accidental injury in this study, Hohman et al. reported the majority of iatrogenic nerve injury were observed in oral and maxillofacial surgeries (40%), resection of head and neck lesions (25%), and otologic procedures (17%).⁸

A statistical analysis could not be performed to evaluate the impact of using intraoperative nerve monitoring, image-guided technology, and robotic-assisted technology due to lack of injury cases among patients in whom those modalities were used. A subpopulation analysis was also not feasible for this sample because the data-using agreement prohibit analyses of very low sample size that could potentially lead to patients identification. The data warrant further investigation of the impact of those technologies on the risk of such injuries as we observed that the number of patients with a reported accidental intraoperative injury in whom no such technologies were used is considerably high (total: 83 patients). Sample size has been a recognized factor that hindered examining the impact of emerging medical technologies. In a study by Dralle et al, they calculated in order to achieve enough statistical power to detect the impact of intraoperative nerve monitoring in patients undergoing thyroid surgery for a thyroid malignancy the study would need to include 39,907 cases and 39,907 controls.¹⁵ They also calculated that the required sample size would be considerably higher if intraoperative nerve monitoring impact to be evaluated in patients with a multiple nodular goiter (9,439,686 patients in each study group). ¹⁵

This current study did not identify a significant risk difference between teaching and non-teaching hospitals, in contrast to a previous study that evaluated all surgeries performed in the United States in which management in a teaching hospital was an independent significant risk factor of intraoperative injury.³ This might indicate that the nature of those accidents is not related to lack of experience but rather the intrinsic danger of such injuries with otolaryngologic procedures.

This study also demonstrated iatrogenic intraoperative injury to be a source of a burden on the health system. Such burden has also been observed in other surgical specialties.³ The incidence rate appeared to be decreasing over the study period, nonetheless, the trend is not steadily downward rendering it not statistically significant. Likely a significant downward trend would be more pronounced as more annual data become available. The downward trend is likely indicative of the progress made with surgical techniques, technology utilization, development of training programs with appropriate supervision, and other measures taken at the level of health systems such as the mortality and morbidity meetings.¹⁶

The study has multiple limitations. The data is collected in a cross-sectional fashion which prevents establishing causality. The data also is administrative in nature and it lacks clinical details. The ICD-9 codes used to identify the injuries lack specificity regarding the structure that was injured and the severity of the injury, also there is no information regarding what type of corrective measures employed to mitigate the injury. The ICD-9 codes of intraoperative injury assumes that the injury was recognized intraoperatively and not speculated, thus it captures cases of clear injury and miss injuries that could result from traction and compression during surgical manipulation. The database also does not capture the larger medicolegal costs associated with this patient group which is likely significant. Overall, the database and the weighted analysis applied is more suited to give an overall and general picture of intraoperative injury at the national level with accurate national estimation of the associated burden. The database also includes a wide array of variables that were evaluated for their potential confounding effect and subsequently included in the multivariate analysis. National databases, like the one used in this study, are transitioning to the ICD-10 system, as this data becomes available it might help improve the quality of studies performed using such resources.

Conclusions

Accidental intraoperative injuries while not associated with a higher mortality or readmission risk, are associated with a significant burden on the health system in terms of an extended hospital-stay and a higher cost. It appears that the incidence of intraoperative injuries reported during otolaryngologic procedures is stable in the United States. The risk factors identified in this study could help guide future study and focus attention on the population that remains at risk of such foreboding injuries. Greater awareness of these risk factors, and the exact nature and local anatomy involved in such injuries embedded in residency training and health system protocols may allow avoidance of these occurrences.