eTable 1. ICD-9 diagnosis and procedure codes for included cases
eTable 2. ICD-9 diagnosis codes for medical and surgical complications
eTable 3. ICD-9 diagnosis codes for airway and swallowing outcomes
Customize your JAMA Network experience by selecting one or more topics from the list below.
Motz K, Chang H, Quon H, Richmon J, Eisele DW, Gourin CG. Association of Transoral Robotic Surgery With Short-term and Long-term Outcomes and Costs of Care in Oropharyngeal Cancer Surgery. JAMA Otolaryngol Head Neck Surg. 2017;143(6):580–588. doi:10.1001/jamaoto.2016.4634
What is the impact of transoral robotic surgery (TORS) on short- and long-term outcomes and cost of care in surgically treated patients with oropharyngeal cancer (OPC)?
In surgically treated patients with OPC, TORS was associated with lower rates of tracheostomy and gastrostomy tube use. For patients requiring postoperative radiation the use of chemoradiation was less likely with TORS. In addition, TORS was associated with decreased overall treatment-related costs of care.
The use of TORS in OPC has the potential to lower the use of adjuvant chemoradiation and reduce treatment-related morbidity and costs of care
The treatment of oropharyngeal cancer has undergone a paradigm shift in the past 2 decades, with an increase in the use of nonoperative treatment owing to poor functional outcomes associated with traditional surgical approaches. Transoral robotic surgery (TORS) allows surgical resection of oropharyngeal cancer (OPC) with less morbidity through a minimally invasive approach.
To investigate the relationship among TORS and short- and long-term outcomes and costs in surgically treated patients with OPC.
Design, Setting, and Participants
Retrospective cross-sectional analysis of 3573 patients who underwent an ablative procedure for OPC in 2010 to 2012 using the MarketScan Commercial Claim and Encounters database.
Main Outcomes and Measures
The association between TORS and short- and long-term outcomes, length of hospitalization, and treatment-related costs was analyzed using descriptive statistics and multivariate regression modeling.
Transoral robotic surgery was performed in 304 surgical cases (8.5%); 94.7% of patients were 40 to 64 years old, and 70.7% were male. The use of TORS increased from 4.1% of surgical cases in 2010 to 13.2% of surgical cases in 2012. Patients who underwent TORS had a lower rate of tracheotomy during treatment (3.9% vs 11.4%), and posttreatment gastrostomy tube use (21.9% vs 34.2%), compared with patients undergoing non-TORS procedures. On multivariate analysis, TORS was not associated with significant differences in postoperative complications or length of hospitalization. There was no significant difference in the odds of receiving postoperative radiation therapy between patients who underwent TORS and those who did not; however, among patients receiving radiation therapy, chemoradiation was significantly less likely following TORS (odds ratio [OR], 0.52; 95% CI, 0.29-0.90). TORS was associated with significantly decreased odds of posttreatment gastrostomy (OR, 0.54; 95% CI. 0.30-0.95) and tracheostomy during treatment (OR, 0.17; 95% CI, 0.06-0.55) at 1 year, and was associated with significantly decreased overall treatment-related costs of care (mean incremental cost, −$22 724).
Conclusions and Relevance
The use of TORS for surgical resection of OPC is increasing in the United States and is associated with significantly lower use of adjuvant chemoradiation, late gastrostomy and tracheostomy dependence, and lower overall treatment-related costs of care. These data have implications for discussions of value in OPC care at a time of health care reform.
Oropharyngeal carcinoma (OPC) is one of the most common head and neck malignant neoplasms, and its incidence has increased over the past 30 years.1,2 The increase in OPC cases has been attributed to human papillomavirus (HPV) infection, and HPV-associated OPC has established itself as a distinct disease entity with a specific clinicopathologic profile.3 Patients with HPV-associated OPC are more commonly younger male patients and demonstrate improved survival compared with patients with non–HPV-related OPC.3 The changing demographic of OPC has prompted a reassessment of current treatment strategies and has led to new treatment paradigms aimed at decreasing treatment-related morbidity.
Historically, surgery followed by radiation therapy was the primary approach for treatment of head and neck malignant neoplasms. The success of organ preservation protocols in laryngeal cancer has led to an increase in the use of nonoperative treatment in OPC as a result of better speech and swallowing outcomes.4-7 However, there is increasing recognition that organ preservation strategies are associated with significant acute and late treatment-related toxic effects, including gastrostomy dependence, long-term dysphagia, aspiration, and non–cancer-related mortality.7-11 As a result, treatment protocols that deescalate the total radiation dose and that use alternative targeted chemotherapeutic agents are being investigated.12
The use of transoral robotic surgery (TORS) in the treatment of OPC has been associated with decreased use of adjuvant therapy, a shorter recovery period, quicker return of swallowing function, and lower gastrostomy utilization rates, with survival outcomes comparable with nonsurgical treatment in select groups.13-17 Most studies reporting TORS outcomes have been limited to single institutions. We sought to investigate the nationwide use of TORS in patients with OPC using a commercial insurance database and to examine associations between TORS and short- and long-term complications, length of hospitalization, functional outcomes, and costs.
A cross-sectional analysis of patients with a diagnosis of OPC was performed using data from the MarketScan Commercial Claims and Encounters Database and the MarketScan Laboratory Database (Truven Health Analytics). This is a large, US-based, employment-based database containing individual-level inpatient and outpatient insurance billing claims for employees and their dependents from approximately 45 large employers covered by more than 100 commercial payers in the United States. MarketScan allows longitudinal tracking of patients across different sites of care over multiple years and contains information regarding inpatient and outpatient treatment, demographic data, primary and secondary diagnoses, primary and secondary procedures, and costs.
The International Classification of Diseases, Ninth Revision (ICD-9) codes were used to identify adult patients (≥18 years) who underwent treatment for a malignant oropharyngeal neoplasm in the years 2010 to 2012 (eTable 1 in the Supplement). The study cohort was defined by a claim with an OPC diagnosis on 1 inpatient claim, or in 2 different outpatient visit claims within 6 months at least 30 days apart, with the first date of the cancer claim used as the diagnosis date. Patients were followed from the index date until death, termination of health insurance, or end of database availability. Follow-up data were available through 2013.
Initial treatment was categorized as surgery only or surgery with postoperative radiation therapy (including postoperative chemoradiation), based on treatment claims provided within the first 180 days following diagnosis. For the purpose of this study, we limited our analysis to procedures that were amenable to a TORS approach: excision and/or destruction of lesion, tonsillectomy, and partial glossectomy, with or without neck dissection, and excluded patients who underwent total glossectomy, pharyngectomy, partial or total laryngectomy, and mandibulectomy, and free flap reconstruction). Patients undergoing outpatient surgery were excluded from analysis because these were unlikely to represent ablative procedures. TORS was identified using ICD-9 codes for robotic-assisted surgery.
Comorbidity was graded using the Romano adaptation of the Charlson comorbidity index,18-20 excluding ICD-9 codes for the index cancer diagnosis from the solid tumor category. Acute medical complications were derived from codes for acute cardiac events, acute pulmonary edema or failure, acute renal failure, acute hepatic failure, acute cerebrovascular events, sepsis, pneumonia, and urinary tract infection assigned at the time of hospital discharge or resulting in admission during the initial treatment period up to 30 days after the end of initial treatment. Surgical complications were derived from codes for complications directly resulting from surgical procedures assigned at the time of hospital discharge or resulting in admission up to 30 days after surgery for patients undergoing surgery as part of initial treatment (eTable 2 in the Supplement).
Dysphagia, weight loss, esophageal stricture, tracheostomy dependence, gastrostomy dependence, and speech-language pathology (SLP) care were defined using ICD-9 and Healthcare Common Procedure Coding System codes in claims (eTable 3 in the Supplement). Tracheostomy and gastrostomy dependence were adjusted for removal of tracheostomy or gastrostomy codes when codes for removal were present. Because removal may not be reimbursed and therefore may not be documented, we considered a gastrostomy or tracheostomy to be absent unless codes for placement or use (ie, supplies) were present. Pretreatment dysphagia, gastrostomy, tracheostomy, and weight loss were defined by the occurrence of codes in claims for these conditions on or after the diagnosis date but before the first date of initial treatment. Posttreatment dysphagia, gastrostomy, tracheostomy, and weight loss were defined by the occurrence of codes in claims for these conditions more than 30 days after the end of initial treatment.
The MarketScan Commercial Claims and Encounters Database includes only commercially insured individuals, and thus patients who are at least 65 years old, uninsured, or receiving Medicare or Medicaid are not included. Race, education, income, hospital characteristics and identifiers, American Joint Commission on Cancer tumor stage, tumor grade, histological subtype, and survival after discharge are not available from the MarketScan database. Metropolitan statistical area (MSA) is provided in MarketScan, which is a geographical region that contains a core population of 50 000 or more, consisting of 1 or more counties that have a high degree of social and economic integration.21 The MSA-level US Census Bureau median household income for the year of diagnosis was determined as an approximate measure of socioeconomic status via linkage to the US Census Bureau and was divided into quintiles.22 The MSA was also used to derive a surrogate for treatment volume. The average annual number of OPC surgical procedures performed per year of surgical activity was obtained by calculating the mean number of procedures performed each year for in each MSA, for the years in which at least 1 OPC surgery was performed within that MSA. The MSA volume was stratified by quintiles, which resulted in cutoff values for annual case volume of 0 to 1, 2 to 3, 4 to 6, 7 to 11, and 12 or more cases per year, which were used to classify MSAs as very low, low, intermediate, high, and very high volume.
Postoperative complications, postoperative outcomes, length of hospitalization, and overall cost of treatment were examined as dependent variables. Secondary independent variables included were age, sex, race, region, payer source (commercial insurance; health maintenance organization; preferred provider organization; point-of-service; or other, including consumer-driven health plans and high-deductible health plans), comorbidity, alcohol and tobacco use, MSA-level median income quintile, MSA-case volume quintile, pretreatment dysphagia, gastrostomy or tracheostomy, postoperative surgical complications, acute medical complications, and adjuvant treatment. Costs were evaluated using all paid amounts from all standard analytic files, including inpatient, outpatient, physician and/or supplier, hospice, home health, and durable medical equipment. Costs were categorized as inpatient, outpatient, or other, and were combined into overall costs. Costs were adjusted for inflation with results converted to 2015 US dollars.23
Data were analyzed using Stata statistical software (version 12; StataCorp). Associations between variables were analyzed using cross-tabulations, multivariate logistic regression analysis, and multinomial logistic regression analysis. National projections of case volumes in the commercially insured population were extrapolated using a proprietary methodology developed by MarketScan, using sampling weights derived from similar subpopulations in the Medical Expenditure Panel Survey and corrected for changes in sampling over time.24 Variables with missing data for more than 10% of the population were coded with a dummy variable to represent the missing data in regression analysis. The primary clinical end points were evaluated using multiple logistic regression analysis. Generalized linear regression modeling with a log link was used to analyze costs and length of stay because these variables were not normally distributed. This protocol was deemed exempt by the Johns Hopkins Medical Institutions institutional review board.
There were a total of 25 385 patients with a diagnosis of OPC who underwent treatment with surgery alone (13%), radiation alone (18%), surgery with postoperative radiation therapy (9%), surgery with postoperative chemoradiation (18%), and chemoradiation (42%) in 2010 to 2012. After excluding outpatient surgical cases and nonoperative treatment, there were 4940 patients with OPC treated with primary surgery in 2010 to 2012, representing 20% of all OPC cases treated during that time period. There were 3573 patients included in this study who underwent excision and/or destruction, tonsillectomy, or partial glossectomy, with or without neck dissection. Most patients were male (77%) and 40 to 64 years old (92%) (Table 1). TORS was used in 304 surgical cases (8.5%) and increased from 4.1% of surgical cases in 2010 to 13.2% of cases in 2012 (mean difference, 9.1%; 95% CI, 1.5%-16.8%]). TORS was more likely to be performed in the North Central part of the United States, and in patients from high median income MSAs. Patients treated with TORS were more likely to receive SLP care during treatment and were more likely to be smokers.
Multiple logistic regression analysis of independent variables associated with the odds of TORS is shown in Table 2. After controlling for the effects of all variables, patients who underwent TORS were more likely to live in the North Central United States, have a history of tobacco use, and receive pretreatment SLP care, and they were less likely to have a neck dissection. In models of postoperative complications, there was no significant association between TORS and acute postoperative morbidity or mortality, or readmission. Patients undergoing neck dissection were significantly more likely to undergo radiation therapy (odds ratio [OR], 1.85; 95% CI, 1.27-2.69) or chemoradiation (OR, 2.39; 95% CI, 1.69-3.38). There was no significant association between TORS and postoperative radiation use: however, among patients treated with radiation, chemoradiation was significantly less likely in patients treated with TORS (OR, 0.52; 95% CI, 0.29-0.90).
Multivariate analysis of variables associated with dysphagia, weight loss, gastrostomy, and tracheostomy use demonstrated that pretreatment dysphagia was a significant predictor of short- and long-term dysphagia and gastrostomy use, and a predictor of posttreatment tracheostomy (Table 3). Tongue base primary site was a predicator of posttreatment tracheostomy tube use. TORS was not associated with significant differences in posttreatment dysphagia or weight loss, but was significantly associated with a lower odds of tracheostomy during treatment and posttreatment gastrostomy tube use. Postoperative radiation therapy and chemoradiation were significantly associated with posttreatment dysphagia, weight loss, gastrostomy, and tracheostomy use.
Multivariate generalized linear regression analyses of independent variables predictive of length of hospital stay and total 1-year costs of care are shown in Table 4, with mean values representing the change in the value of the intercept mean. Pretreatment dysphagia, weight loss, tongue base primary site disease, increased comorbidity, and nonroutine discharge were significantly associated with greater length of hospitalization. Pretreatment gastrostomy was associated with decreased length of hospitalization, whereas TORS was not associated with differences in length of hospitalization. Pretreatment gastrostomy, residence in the top US Census MSA median income quintile, and gastrostomy or tracheostomy during treatment were significantly associated with increased treatment-related costs, whereas TORS and neck dissection was associated with decreased treatment-related costs. The highest treatment-related costs were associated with adjuvant treatment, with significantly higher costs of care for patients receiving postoperative chemoradiation.
Our findings demonstrate that in recent years, the use of TORS for OPC surgery is increasing in the United States and was associated with decreased dysphagia during treatment, decreased odds of long-term gastrostomy, and tracheostomy tube use. In addition, TORS was associated with lower odds of postoperative chemotherapy and lower treatment-related costs. These data have implications for the contemporary management of OPC in an era in which value-based medicine, focusing on both outcomes and costs, is increasingly emphasized.
Similar to findings in previous studies, we found that patients who underwent TORS were less likely to receive postoperative chemoradiation than patients treated with non-TORS approaches. This observation may reflect patient selection because the effectiveness of TORS is questionable if adequate surgical margins cannot be achieved and/or extracapsular spread is present in cervical nodes preoperatively, both being indications for adjuvant or primary chemoradiotherapy. However, in appropriately selected patients, TORS offers the potential for treatment deintensification, making it a desirable therapeutic tool in the armamentarium of the head and neck surgical oncologist. This is particularly important given the epidemic of younger patients with HPV-associated OPC who will likely live long enough to experience the sequela of treatment-related toxic effects. Treatment deintensification with TORS is currently the focus of the Eastern Cooperative Oncology Group E3311 transoral surgery clinical trial, a multiinstitutional trial studying the use of TORS for deintensification through sparing radiation doses in low- and intermediate-risk patients with OPC.25
We found that patients undergoing TORS were less likely to have late gastrostomy or require a tracheostomy during treatment . This observation may reflect the decreased morbidity associated with TORS approaches that allow for earlier return to oral intake, but may also reflect patient selection and tumor stage because patients with more limited primary site disease and good preoperative function are ideal candidates for TORS. The lower use of chemoradiation observed in the population of patients treated with TORS may also influence the observed rates of long-term gastrostomy dependence by avoiding the increased acute mucosal toxic effects associated with chemoradiation.26 The increased use of SLP care we observed in patients undergoing TORS may also influence this finding. The prophylactic use of SLP management during radiation therapy for head and neck cancer has been shown to maintain muscle function and preserve the capacity for a near-normal diet following radiation.27-29 Regardless of cause, the reduction of gastrostomy use is a significant benefit of TORS because routine gastrostomy use has been shown to adversely affect long-term swallowing function.29
Similar to findings in previous reports, we found that TORS was associated with lower treatment-related costs of care in the first year of OPC care.30 The cost-benefit of TORS has been proposed to result from its association with shorter length of hospitalization, earlier return to swallowing, reduced rates of gastrostomy use, and reduced use of adjuvant therapy.30,31 We did not find a significant difference in length of hospitalization, which may reflect an inability to distinguish procedures, such as tonsillectomy or excision and/or destruction of lesion, performed without TORS for diagnostic purposes, which may be associated with short length of hospitalization, from those done with curative intent. In addition, there is a considerable learning curve with TORS, and early experience may be associated with greater length of hospitalization. The reduced costs associated with TORS may reflect unmeasured patient characteristics associated with patient selection but also likely reflect improved functional outcomes resulting in less resource use and a lower incidence of postoperative chemoradiation because costs associated with chemotherapy have been shown to be an important driver of costs of care.32 These data have implications at a time when value is increasingly the focus of health care reform efforts, measured by outcomes achieved per dollar spent. Treatment approaches that are associated with preserved or improved outcomes and lower costs will increasingly be desirable in an effort to curb health care spending, and are also the right thing to do if value can be demonstrated.
There are several limitations to the use of claims data in risk documentation and risk adjustment that may influence our findings. The MarketScan database provides no information on patient race/ethnicity and captures only commercially insured patients younger than 65 years. While this is the primary demographic of HPV-associated OPC, limited demographic data limit conclusions that can be drawn. Disease-specific information, including stage of disease, grade, subtype, HPV status, or survival, is not available. As a result, there may be differences in the type of patient or disease treated with TORS that are not adequately captured. There may be differences in tumor stage and characteristics in patients selected for TORS that cannot be captured using administrative data. While there is a claims code for the use of TORS, there are no corresponding codes to identify other methods of transoral resection, such as transoral laser microsurgery, and thus the lack of a difference in length of hospitalization between patients treated with or without TORS could potentially be related to the use of non-TORS transoral surgical approaches as well as limited procedures performed for diagnostic rather than therapeutic purposes in patients not treated with TORS. While comorbidity scores were used for risk adjustment, the ability to adequately control for case mix is limited when discharge diagnoses from administrative databases are used. Finally, poor documentation of conditions associated with functional outcomes may lead to underestimation of treatment-related toxic effects, and the severity of toxic effects may not be reflected in claims that rely on administrative coding.
Despite these limitations, these data suggest that the use of TORS in the surgical management of OPC is associated with significantly lower odds of adjuvant chemoradiation, late gastrostomy, and tracheostomy dependence, and lower overall treatment-related costs of care. These data have implications for improving the value of OPC care at a time of health care reform.
The use of TORS for surgical resection of OPC is increasing in the United States and is associated with significantly lower use of adjuvant chemoradiation, late gastrostomy and tracheostomy dependence, and lower overall treatment-related costs of care. These data provide further support for the role of TORS in OPC care.
Corresponding Author: Christine G. Gourin, MD, MPH, Department of Otolaryngology–Head and Neck Surgery, Johns Hopkins Outpatient Center, 601 N Caroline St, Ste 6260, Baltimore, MD 21287 (email@example.com).
Accepted for Publication: December 13, 2016.
Published Online: March 30, 2017. doi:10.1001/jamaoto.2016.4634
Author Contributions: Drs Motz and Gourin had full access to all of 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: Motz, Gourin.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Motz.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Motz, Chang, Gourin.
Obtained funding: Gourin.
Administrative, technical, or material support: Chang, Gourin.
Study supervision: Gourin.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest, and none were reported.
Funding/Support: Support for Dr Motz’s effort on this publication is through a National Institutes of Health (NIH) T32 training grant.
Role of the Funder/Sponsor: The NIH had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Meeting Presentation: This research was presented as an oral presentation at the American Head and Neck Society Ninth International Conference on Head and Neck Cancer; July 17, 2016; Seattle, Washington.