Introduction
Head and neck cancer accounts for approximately 4 per cent of cancers in the United States, with an estimated 900 000 cases annually worldwide.Reference Siegel, Miller, Fuchs and Jemal1, Reference Mella, Chabrillac, Dupret-Bories, Mirallie and Vergez2 Based on the heterogenous composition of primary sites, histology, cancer staging and patient characteristics, there are various treatment and prognostic outcomes. However, most head and neck cancers are generally treated with surgery, radiation and/or chemotherapy.Reference Mella, Chabrillac, Dupret-Bories, Mirallie and Vergez2
Surgery has long been a mainstay treatment; however, robotic surgery is a relatively new addition to the head and neck surgeon’s armamentarium. Robotic surgery for head and neck cancer was first described in 2005, and Food and Drug Administration (FDA) approval was granted in 2009.Reference Garg, Dwivedi, Sayed, Katna, Komorowski and Pathak3, Reference Chia, Gross and Richmon4 Specifically, robotic surgery can be utilised for T1–T2 malignancies and benign lesions of the oral cavity, pharynx and larynx.Reference Mella, Chabrillac, Dupret-Bories, Mirallie and Vergez2, Reference Chia, Gross and Richmon4, Reference Boudreaux, Rosenthal, Magnuson, Newman, Desmond and Clemons5 Surgical volume using robotic systems has increased since FDA approval, and new indications for robotic surgery in the head and neck are continuously being evaluated.Reference Hockstein, O’Malley and Weinstein6–Reference Garas and Arora9
The benefits of robotic surgery continue to be discussed and refined as information regarding surgeon experience, increased utilisation and treatment outcomes is gathered. While specific benefits may vary between cases, several general advantages have been identified, including improved functional outcomes such as shorter hospital stays, faster recovery, reduced pain, decreased blood loss, and lower complication rates and morbidity.Reference Garg, Dwivedi, Sayed, Katna, Komorowski and Pathak3, Reference Chia, Gross and Richmon4, Reference Weinstein, O’Malley, Desai and Quon10 Surgically, the use of a robotic system can improve access and visualisation, enable precise movements and, at certain primary sites, allow for the unique advantage of utilising two hands.Reference Garg, Dwivedi, Sayed, Katna, Komorowski and Pathak3, Reference Weinstein, O’Malley, Desai and Quon10, Reference Moore, Olsen and Kasperbauer11 Some disadvantages of robotic surgery include its high cost, limited accessibility to the necessary equipment and instruments, and the additional space and personnel required for its operation and maintenance.Reference Garg, Dwivedi, Sayed, Katna, Komorowski and Pathak3, Reference Chia, Gross and Richmon4, Reference Weinstein, O’Malley, Desai and Quon10 Additionally, there is a learning curve for surgeons, including the loss of tactile feedback compared to traditional approaches.Reference Garg, Dwivedi, Sayed, Katna, Komorowski and Pathak3, Reference Chia, Gross and Richmon4, Reference White12, Reference Albergotti, Gooding, Kubik, Geltzeiler, Kim and Duvvuri13 Operative times have varied, with both increases and decreases observed when comparing robotic surgery to traditional approaches.Reference Chia, Gross and Richmon4, Reference Weinstein, O’Malley, Desai and Quon10
Given the trend of improving outcomes following robotic surgery, it is of interest to determine what types of patients receive this type of treatment. Previous literature has demonstrated that patient demographics influence cancer diagnosis and treatment outcomes, with factors such as sex, age, race, ethnicity, treatment hospital, distance travelled to the hospital, insurance payer, socio-economic status, education level and comorbid conditions all playing a role.Reference Chen, Fedewa, Pavluck and Ward14–Reference Graboyes, Ellis, Li, Kaczmar, Sharma and Lentsch21 This study aims to identify the socio-economic and clinical factors associated with the increased use of robotic surgery and its potential to improve outcomes.
Materials and methods
Data were analysed from the National Cancer Database (NCDB), which is a cancer registry from the American College of Surgeons (ACS). While not a population-based registry, the NCDB includes approximately 70 per cent of cancer diagnoses in the United States.Reference Bilimoria, Stewart, Winchester and Ko22 The NCDB uses de-identified data. This study was determined to be exempt from approval under Creighton University Institutional Review Board (IRB) review submission number 2003810.
Patients with head and neck cancer surgically treated between the years 2010 and 2020 were identified. Head and neck cancers of the larynx, pharynx, oral cavity and salivary glands primary sites were included, under NCDB primary site codes C000–C006, C008–C009, C019–C024, C028–C031, C039–C041, C048–C052, C058–C062, C068–C069, C079–C081, C088–C091, C098–C104, C108–C113, C118–C119, C129–C132, C138–C140, C142, C148, C320–C323 and C328–C329. All histological types found within these primary sites were included. Patients who were not treated with surgery were excluded from the analysis. Patients were then divided into groups based on the surgical approach of the primary site. The robotic group included patients who had approaches of robotic-assisted or robotic converted to open. The non-robotic group included patients who underwent minimally invasive, minimally invasive converted to open, and open or unspecified approaches.
The following demographic variables were examined in the study: year of diagnosis, sex, age, race, ethnicity, facility type, population of residence, great circle distance, primary insurance payer, median household income, high school degree attainment, Charlson–Deyo Score (CDS) and cancer stage. Age was categorised into twenty-year intervals. Race was categorised as White, Black, American Indian, Aleutian or Eskimo, Asian, other and unknown to ensure inclusivity of various racial groups. Ethnicity was classified into three groups: Spanish or Hispanic origin, not of Spanish or Hispanic origin and unknown. The facility type was categorised into four groups: Community Cancer Programs, Comprehensive Community Cancer Programs, Academic/Research Programs and Integrated Network Programs, reflecting different levels of healthcare facility resources and capabilities. The population of residence was classified into three areas: metro, urban and rural, which can impact healthcare access and treatment options. Great circle distance, a measure of the straight-line distance between the patient’s residence and the treatment hospital, was grouped into 4 categories: 10, 25, 50 and 100 miles, to assess the effect of distance on treatment choices. Primary insurance payer was grouped into five categories: uninsured, private insurance, Medicaid, Medicare and other government insurance, as well as an unknown category, to evaluate the impact of insurance type on access to robotic surgery. Median household income was divided into 4 quartiles based on the median income of a patient’s zip code, with quartile 1 representing an income of less than $ 46 277 and quartile 4 representing an income greater than $ 74 062. High school degree attainment was determined by the educational level within a patient’s zip code, grouped into 4 categories: areas with more than 15.2 per cent, 9.1–15.2 per cent, 5.0–9.0 per cent and less than 5.0 per cent of people lacking a high school degree, to examine the potential influence of education on health outcomes. Cancer stage, as defined by the NCDB, was grouped into Stage 0, Stage I, Stage II, Stage III and Stage IV, reflecting the progression of the disease. The CDS, which provides a measure of a patient’s comorbid conditions, was used to assess the severity of other health conditions. Charlson–Deyo Score was categorised as 0, 1, 2, or 3 or greater, reflecting the number and severity of co-existing medical conditions.Reference Charlson, Pompei, Ales and MacKenzie23
Descriptive statistics were initially employed to examine the variables, followed by comparisons based on surgical approach, using Pearson’s chi-square test and cross-tabulations. To assess the relationship between variables and surgical approach while adjusting for potential confounders, multivariate binary logistic regression was performed. Cases with missing data were excluded from the analysis. All statistical analyses were conducted using SPSS version 29 (IBM, Armonk, New York, USA), with a significance threshold set at p < 0.05.
Results and analysis
This study identified 212 449 patients who underwent robotic surgery for their head and neck cancer between the years 2010 and 2020. Table 1 displays the frequencies of surgical approaches. The robotic approach was the least common (5.9 per cent), whereas the open or unspecified approach comprised the majority of cases (79.1 per cent) and minimally invasive represented (14.4 per cent) of cases.
Frequency of surgical approaches
Variable frequency and cross-tabulation of patients receiving non-robotic and robotic surgery; Results of Chi-square analysis with a significance level < 0.05; CDS = Charlson–Deyo Score
Table 2 lists frequency values, results of cross-tabulations and chi-square analysis for all variables. The majority of patients in the cohort were male (65.9 per cent), aged between 61 and 80 years (49.3 per cent), identified as White (87.0 per cent) and reported as not Hispanic or Spanish (93.1 per cent). The incidence of head and neck cancer was stable over time. The most common treatment facility type was an Academic/Research Program (59.0 per cent), with the majority of patients residing in metro areas (82.9 per cent) and living within a 10-mile great circle distance from the treatment centre (35.0 per cent). Medicare (44.9 per cent) and private insurance (40.2 per cent) were the most prevalent insurance types. Additionally, the majority of patients had a CDS of 0 (73.7 per cent). The most frequently diagnosed stage was Stage IV (37.8 per cent), followed by Stage I (21.6 per cent), with Stage 0 being the least common (2.9 per cent).
Table 3 presents the results of the multivariate analysis. Overall, there was an increase in the likelihood of receiving robotic surgery over time, with patients diagnosed in 2020 being 116 per cent more likely to undergo robotic surgery than those diagnosed in 2010. Female patients were 54 per cent less likely to receive robotic surgery than male patients. Age-related differences showed that patients aged 0–20, 21–40 and 80+ were significantly less likely to receive robotic surgery compared to the 61–80 age group, with reductions of 82 per cent, 56 per cent and 71 per cent, respectively. However, patients aged 41–60 were 43 per cent more likely to undergo robotic surgery than those aged 61–80. Patients from racial minority groups were less likely to undergo robotic surgery compared to White patients. Specifically, Black patients were 14 per cent less likely, American Indian, Aleutian or Eskimo patients were 38 per cent less likely and Asian patients were 70 per cent less likely to receive robotic surgery. No significant differences were found for ethnicity. Regarding treatment centres, patients at Comprehensive Community Cancer Programs were 82 per cent more likely, Academic/Research Programs 418 per cent more likely and Integrated Network Cancer Programs 183 per cent more likely to receive robotic surgery compared to those treated at Community Cancer Programs. Location-based factors also influenced robotic surgery rates. Rural patients were 36 per cent less likely and urban patients 20 per cent less likely to undergo robotic surgery compared to those living in metro areas. Interestingly, patients living further from treatment centres were more likely to receive robotic surgery, with the likelihood increasing as distance from the treatment centre increased. Insurance type played a role in the likelihood of undergoing robotic surgery, with patients with private insurance, Medicaid, Medicare and other government insurance being more likely to undergo robotic surgery compared to uninsured patients. The likelihood of receiving robotic surgery was 165 per cent, 70 per cent, 84 per cent and 190 per cent higher for these groups, respectively. Income level also influenced the likelihood of robotic surgery, with patients in the highest median income quartile being 37 per cent more likely to undergo robotic surgery compared to those in the lowest income quartile. No significant difference was observed between the middle two income quartiles. Educational attainment in the patient’s area of residence was associated with a greater robotic surgery likelihood. Compared to areas with a higher percentage of individuals with no high school degree (> 15.3 per cent), patients in areas with 9.1–15.2 per cent, 5.0–9.0 per cent and less than 5.0 per cent without a high school degree were 12 per cent, 21 per cent and 15 per cent more likely, respectively, to undergo robotic surgery. Charlson–Deyo Score also influenced the likelihood of receiving robotic surgery. Patients with a CDS of 1, 2 or 3 or greater were 6 per cent, 19 per cent and 12 per cent less likely, respectively, to receive robotic surgery compared to patients with a CDS of 0. Finally, patients with Stage I cancer were the most likely to undergo robotic surgery, while those with Stage II, III, and IV cancer were 67 per cent, 63 per cent and 60 per cent as likely, respectively, to opt for robotic surgery.
Multivariate analysis including odds ratios, confidence intervals, and p-value with a significance level < 0.05; CDS = Charlson–Deyo Score
Discussion
With the advancement of medical technology, the utilisation of robotic surgery has significantly increased across various surgical subspecialties, including within the field of head and neck surgery. According to Sheetz et al., robotic surgery saw an 8.4-fold increase from 2012 to 2018Reference Sheetz, Claflin and Dimick24. This study demonstrated that while robotic surgery remains less commonly used than traditional open surgery, there has been an increasing trend in patients’ willingness to undergo robotic procedures in recent years. If these trends persist, the acceptance of robotic surgery for head and neck cases, as well as the total number of procedures performed, is likely to continue rising.
Although some variability was observed, patients between the ages of 41 and 80 were generally the most likely to receive robotic surgery. Interestingly, this age group also represents the most common demographic for developing head and neck cancer. It is possible that these patients had more commonly presenting conditions that made them suitable candidates for robotic surgery, though there is limited literature suggesting that age is a significant deterrent to robotic procedures.Reference Aloisi, Tseng, Sandadi, Callery, Feinberg and Kuhn25, Reference Ceccarelli, Andolfi, Biancafarina, Rocca, Amato and Milone26
However, the uptrend in robotic surgery use is not without notable demographic disparities. For example, this study demonstrated that female patients were less likely to receive robotic surgery as compared to men. This finding is remarkable, given that men generally have a higher proportion of later-stage disease, thus making them less likely to qualify for robotic surgery.Reference Vaidya, Partha and Karmakar27–Reference Dittberner, Friedl, Wittig, Buentzel, Kaftan and Boeger29 One explanation for this could be that female patients are less likely to undergo robotic surgery by choice, supported by McDermott et al. who suggested that female patients are less likely to perceive robotic surgery positively.Reference McDermott, Choudhury, Lewin-Runacres, Aemn and Moss30
In addition to gender disparities, this study also revealed racial disparities in the utilisation of robotic surgery, with non-White patients being less likely to receive this treatment. This finding aligns with previous research by Frenkel et al., which demonstrated that Black and Hispanic patients were less likely to undergo transoral robotic surgery with neck dissection compared to their Caucasian counterparts.Reference Frenkel, Yang, Zhang, Regenbogen, Telem and Samara31 Similar patterns have been observed across other surgical subspecialties, such as in pelvic cancer robotic surgeries, where non-White patients were less likely to receive robotic treatment.Reference Mao, Genkinger, Rundle, Wright, Aryal and Liebeskind32 A potential contributing factor to these racial disparities is that non-White patients are more frequently diagnosed at later stages of disease, often due to a combination of socio-economic barriers and racial inequalities in healthcare access.Reference Shin and Truong15, Reference Liu, Egleston, Blackman and Ragin33 This delayed diagnosis may result in ineligibility for robotic surgery. Overall, this study reveals the racial disparities in access to emerging surgical technologies and highlights the need for addressing inequities in healthcare access and treatment options.
Beyond demographic factors, socio-economic variables play a pivotal role in access to robotic surgery. Patients in the highest income quartile, as well as those with higher levels of education, were more likely to pursue robotic surgery. Additionally, patients residing in areas with presumably higher educational attainment were more likely to undergo robotic procedures. This trend correlated with insurance type, where individuals with private insurance were more likely to receive robotic surgery compared to those with Medicare or Medicaid. Given that robotic surgery is often more costly for both hospitals and patients, its increased feasibility among individuals with higher socio-economic status demonstrates the influence of economic factors on the availability and accessibility of advanced medical treatments.Reference Ng, Sanaiha, Bakhtiyar, Ebrahimian, Branche and Benharash34
Institutional and geographic factors also significantly influence the utilisation of robotic surgery. This study demonstrated that patients treated at Academic/Research facilities and those residing in more densely populated areas were more likely to undergo robotic surgery. This finding is not surprising, as these institutions typically possess the resources required to afford robotic systems and often lead in the early adoption of new technologies. As noted by previous research, rural centres generally acquire advanced medical technologies later than their urban counterparts.Reference Haggstrom, Lee, Dickinson, Kianersi, Roberts and Teal35 Similarly, Frenkel et al. observed that the majority of patients undergoing transoral robotic surgery were from urban areas.Reference Frenkel, Yang, Zhang, Regenbogen, Telem and Samara31 This study found that patients who travelled longer distances to their treatment centres were more likely to receive robotic surgery, suggesting that some individuals may specifically seek out these centres for access to robotic procedures. In contrast, patients not receiving robotic surgery may have been more inclined to visit the nearest hospital, rather than going out of their way to access robotic surgery, although the NCDB does not provide specific reasoning for patient choices.
Finally, this study noted that those with a high CDS, a way to predict the long-term survival of a patient, were less likely to receive robotic surgery. This may be simply due to some comorbidities being a contraindication to robotic surgery, or surgeons favouring a more standard approach with more complex patients.Reference Niewinski and Golusiński36 Stage I was found to be the most common for robotic surgery, likely since patients in this stage tend to have fewer comorbidities and better overall prognosis, making them more suitable candidates for minimally invasive procedures like robotic surgery. This finding is consistent with a multicentric cohort by de Almeida et al., the largest of its kind, which showed that the vast majority of patients with head and neck squamous cell carcinoma (SCC) had early-stage disease (83.5 per cent).Reference De Almeida, Li, Magnuson, Smith, Moore and Lawson37
There are several limitations that must be considered when interpreting the results. First, the NCDB is not population based and includes only around 70 per cent of cancer diagnoses.Reference Palma38 Given the limited data, our statistical analysis may be incomplete. Additionally, this study allows for the statistical interpretation of odds, but the reasoning behind these findings is unable to be elucidated from the NCDB and is open to interpretation. Large databases, such as the NCDB, often suffer from incomplete data entry and inconsistencies in how data is reported, which can introduce challenges in analysis and raise the potential for misclassification. This issue is particularly relevant in this cohort of patients with head and neck malignancies, where critical variables were missing for a significant portion of cases. These omissions could have introduced differences between groups that potentially impacted the analysis and conclusions drawn.
Furthermore, the use of non-surgical therapy in some instances may serve as a surrogate for other adverse health factors, such as significant medical comorbidities, inadequate social support, advanced disease severity or poor performance status. While we accounted for many of these confounding variables in the multivariate analysis, residual confounding cannot be entirely ruled out. Additionally, the NCDB, while an invaluable resource for quality improvement and research on rare tumours, has intrinsic limitations. It primarily draws data from accredited cancer centres, introducing a selection bias that limits its generalisability to the broader population. This is in contrast to the Surveillance, Epidemiology, and End Results (SEER) database, which captures all cases within defined geographic regions and is more representative of the general population.Reference Mohanty and Bilimoria39
Despite these challenges, the NCDB remains a powerful tool for examining large patient cohorts and offers valuable insights into survival outcomes and the factors influencing treatment utilisation. Its unique strengths make it an essential resource for understanding trends and outcomes in oncology, even as researchers must remain mindful of its inherent limitations.
Future studies are needed to elucidate the effect of the head and neck primary site on robotic surgery and the intersection with the above findings. Subgroup analyses of populations with large proportions of patients treated with robotic surgery, i.e. oropharyngeal and laryngeal primaries, may show larger socio-economic and gender discrepancies than those seen in this present study. Likewise, the addition of tumour factors such as human papillomavirus (HPV) status poses an opportunity for future study.
Conclusion
Patients more likely to undergo robotic surgery include those diagnosed in recent years, male patients, individuals aged 41–80 years, White patients, those treated at Academic/Research facilities, residents of metro areas, patients who travel further for treatment, those with insurance, individuals from higher socio-economic backgrounds, residents of more educated areas and those with a lower CDS. These results emphasise possible socio-economic and clinical disparities in the utilisation of robotic surgery, underscoring the obstacles that may preclude some patients from accessing this advanced treatment. They also show the role of social determinants of health in influencing medical decisions and outcomes. A more thorough understanding of these factors is imperative to advocate for patients to have more equitable access to the most effective treatment options.
Data availability statement
All data used is accessible through the National Cancer Database.
Competing interests
The authors declare none
