Update on adherence to guidelines for time to initiation of postoperative radiation for head and neck squamous cell carcinoma

Background: A prior study reported that over half of patients with HNSCC initiated PORT after 6 weeks from surgery during 2006–2014. In 2022, the CoC released a quality metric for patients to initiate PORT within 6 weeks. This study provides an update on time to PORT in recent years. Methods: The NCDB and TriNetX Research Network were queried to identify patients with HNSCC who received PORT during 2015–2019 and 2015–2021, respectively. Treatment delay was defined as initiating PORT beyond 6 weeks after surgery. Results: In NCDB, PORT was delayed for 62% of patients. Predictors of delay included age >50, female sex, black race, nonprivate insurance/uninsured status, lower education, oral cavity site, negative surgical margins, increased postoperative length of stay, unplanned hospital readmissions, IMRT radiation modality, treatment at an academic hospital or in the Northeast, and surgery and radiation at different facilities. In TriNetX, 64% experienced treatment delay. Additional associations with prolonged time to treatment included never married/divorced/widowed marital status, major surgery (neck dissection/free flaps/laryngectomy), and gastrostomy/tracheostomy dependence. Conclusions: There continue to be challenges to timely initiation of PORT.


| INTRODUCTION
[6] In January of 2022, the American College of Surgeons (ACS) Commission on Cancer (CoC) released a national quality metric for patients with HNSCC to initiate PORT within 6 weeks. 7Given that 8 years passed since the last large-scale analysis of timeliness of care, this study provides an update on adherence to guidelines for time to initiation of PORT for patients with HNSCC.Additionally, we sought to define the current state of the problem, propose additional factors associated with delay, and need for further interventions.

| MATERIALS AND METHODS
The study data was obtained from the NCDB.The NCDB includes patient information from more than 1500 CoC accredited institutions across the United States. 8The study cohort, outcome measures, and study variables were based on methods previously described by Graboyes et al., 3 except for the time period of database review.Graboyes et al. analyzed  2006 through 2014; therefore, we reviewed 2015 through 2019.The cohort was composed of patients with HNSCC who initiated PORT, with treatment delay defined as beginning radiation beyond 6 weeks after the date of surgical intervention.Demographic characteristics collected included age, sex, race, ethnicity, insurance status, area of residence, educational attainment quartile, median household income, and distance to hospital.Educational attainment was determined by comparing patient zip codes at the time of diagnosis to American Community Survey data and is based on the percentage of adults in a given zip code who did not graduate high school.Clinical and surgical covariates included cancer site, American Joint Committee on Cancer (AJCC) clinical and pathological stages, surgical margin status, postoperative length of stay, 30-day hospital readmissions, radiation modality, concurrent chemoradiation, and Charlson-Deyo Comorbidity score.Patients diagnosed before 2018 were staged according to the AJCC 7th edition guidelines, and after 2018 using the 8th edition.Hospital characteristics analyzed included treatment facility type, number of facilities involved in treatment, whether surgery and radiation were conducted at the same facility, and region of care.Treatment facility type is based on the annual volume of newly diagnosed cancer cases and structure of facilities: community cancer program (100-500 annual diagnoses), comprehensive community cancer program (>500 annual diagnoses), academic/research program (>500 annual diagnoses plus resident trainees), and integrated network cancer program (organization owning a group of facilities designed to offer comprehensive and integrative cancer care services).
An additional analysis was conducted using the TriNetX Research Network.TriNetX is a global federated health research database that provided access to electronic medical records (diagnoses, procedures, medications, laboratory values) from health care organizations (HCOs) in the United States. 9The database was queried to identify patients with head and neck cancer diagnosed during 2015 through 2021 who initiated PORT.Similar exclusion criteria were applied as the NCDB analysis, that is, those who underwent surgery beyond 6 months after initial diagnosis or radiation beyond 6 months postoperatively, and those who underwent induction chemotherapy or received stereotactic radiation or brachytherapy.The diagnosis and procedure codes utilized to execute this search criteria are listed in Data S1, Supporting Information.Those who met inclusion and exclusion criteria were stratified according to whether they received PORT within or beyond 6 weeks of surgery.Demographic (age, sex, race, ethnicity, region, marital status), clinical (tumor site, concurrent chemoradiation, and presence of tracheostomy or gastrostomy tube), and surgical variables (neck dissection, free flaps, laryngectomy) were compared between groups.

| STATISTICAL ANALYSIS
For the NCDB analysis, bivariate associations were first assessed using chi-square tests to compare between those with initiation of PORT less than or equal to 6 weeks versus beyond 6 weeks.The factors with significant bivariate associations were then considered for the multivariable logistic regression modeling process.All factors were included in the full model, which was subsequently reduced using a manual backwards selection procedure to only include those factors which remained significant at the 0.05 level.Analysis was conducted utilizing SAS statistical software version 9.4 (SAS Institute Inc., Cary, NC).For the TriNetX subanalysis, odds ratios and 95% confidence intervals were calculated to determine the odds of receiving substandard care for those who belonged to a particular demographic or had certain clinical or surgical variables coded in their medical record.Statistical analyses were performed within the TriNetX platform, which is based on R, JAVA, and Python.

| ETHICAL APPROVAL
Both TriNetX and the NCDB only use aggregated counts of de-identified information, with no protected health information or personal data made available.Therefore, this study was exempted by the Penn State Institutional Review Board review (STUDY00018629 for TriNetX and STUDY00018678 for NCDB).

| NCDB analysis
In the NCDB, 47 331 patients were identified as potentially eligible for analysis.Excluded were 4770 patients who underwent brachytherapy, stereotactic radiosurgery, or had an unknown radiation modality, 570 patients who received palliative treatment, and 249 patients who underwent neoadjuvant therapy.There were an additional 1244 patients who underwent surgery beyond 6 months of diagnosis and 334 who began radiation beyond 6 months of surgery who were also excluded.The final cohort for analysis included 40 164 patients, with demographics described in Table 1.

| TriNetX analysis
In TriNetX, there were 3651 patients with head and neck cancer who were treated with surgery and PORT during 2015 through 2021.Of this cohort, 486 patients were excluded for the following reasons: received stereotactic radiation or brachytherapy (n = 14), received induction chemotherapy (n = 35), and underwent surgery beyond 6 months after initial diagnosis (n = 115) or PORT beyond 6 months of surgery (n = 322).This left 3165 patients who met criteria for analysis.Demographics are described in Tables 7 and S6.

| DISCUSSION
Substantial strides in cancer care resulted from the advent of novel treatments, adoption of the multidisciplinary approach to patient-centered care, and implementation of evidencebased practice guidelines.However, variability in adherence to guidelines results in suboptimal outcomes across various malignancies. 102][13][14][15][16][17] Graboyes et al. reported that time to PORT deviated from this guideline for approximately half of this patient cohort during 2006-2014. 3Our updated analysis demonstrated a persistence of this trend, with nearly two-thirds of patients initiating PORT beyond the 6-week timepoint.
In our study, a variety of demographic, clinical, surgical, and hospital factors predicted treatment nonadherence, with many variables sharing commonality with the prior NCDB analysis. 3Persistent associations with treatment delay included black race, public insurance, lower educational attainment, oral cavity tumor site, increased postoperative length of stay or unplanned readmissions, treatment at an academic center, and fragmented surgery and PORT at different locations.4][5][6] Additionally, temporal trends are suggestive of worsening adherence over time.Variables with the strongest associations with PORT delay included increased length of hospitalization, unplanned readmission, oral cavity cancer site, and insurance status.When designing interventions, targeting these areas may provide the greatest opportunity for improvement.
There are reports of a few isolated but promising quality improvement endeavors resulting in reduced PORT delay.One institution reported that the percentage of free flap patients receiving timely PORT increased from 10.5% to 50% due to three interventions: standardized care-pathway order sets, prompt referrals to radiation oncology, and most importantly, assigning a patient navigator to coordinate care and serve as the liaison between patients, physicians, and schedulers. 18Another project reduced preventable delays from 24% to 9% and improved overall timeliness of care from 62% to 73% by initiating dental evaluation at the time of the new patient visit with scheduled extraction during surgery, placing radiation oncology referrals at the time of identified indication for PORT, setting automated reminders to review surgical pathology if there was a clinical concern for PORT, and creating a standardized head and neck oncology clinic visit checklist to recap key aspects of the care timeline. 19The most successful intervention to date, Navigation for Disparities and Untimely Radiation thErapy (NDURE), involves a series of three appointments with a patient navigator throughout the treatment course to coordinate care, provide patient education, and identify and resolve potential modifiable barriers to PORT.This resulted in 86% of patients receiving PORT within 6 weeks; however, it should be noted that there were only 15 patients in total in the study. 20Additionally, nomograms and machine learning algorithms are available to identify patients at highest risk for PORT delay. 21,22olonged length of stay and unplanned hospital readmissions are often due to surgical complications.Therefore, efforts to minimize complications via emphasis on measures such as adequate preoperative nutrition, medical optimization, and thromboprophylaxis may be useful.Importantly, multidisciplinary communication is critical in advance of surgery.Patients should either see their radiation oncologist preoperatively or have a postoperative appointment scheduled in advance.This may help avoid last minute care coordination and delays if a patient desires surgical and radiation care at different facilities for example, a variable that was significantly associated with delay in the current study.It is essential that institutions managing HNSCC patients evaluate and implement similar interventions to improve adherence to national guidelines.
A distinct advantage of TriNetX is analysis of additional factors that were not studied previously.Specifically, patients who underwent a neck dissection, certain free flaps, or laryngectomy were more likely to experience PORT delay.This also applied to those with a tracheostomy or gastrostomy tube.This suggests patients with more extensive surgeries or in need of additional ancillary services are at increased risk for PORT delay.Additionally, patients in TriNetX who were never married, divorced, or widowed were significantly more likely to receive PORT beyond 6 weeks.A possible explanation for this could be that partners were likely to provide tangible services and informal caregiving, such as transportation and care coordination that resulted in increased adherence to guidelines.The improved timeliness of treatment in the married cohort may also be extrapolated to any unmarried patients with caring partners who will provide various levels of support after surgery.Additionally, other family members, friends, neighbors, or patient navigators may provide a similar positive influence.Initiatives aimed at providing these conveniences may be worth pursuing in patients lacking support.Now that time to PORT is officially a CoC national quality metric, 7 head and neck oncologists and other members of the multidisciplinary care team will prioritize similar efforts to improve adherence.Our results provide a performance baseline, with future research necessary to continue trending outcomes moving forward.While our study provides a recent update regarding adherence to treatment guidelines and is the first to identify certain factors as predictive of delay, it is not without limitations.First, it is limited by the quality of data entry in the NCDB and TriNetX, which is a recognized weakness in performing a retrospective review and using a large de-identified database.However, the similar rate of treatment delay in two databases increased our confidence in our findings, and the large sample size of patients strengthened the generalizability.Furthermore, TriNetX was only able to capture those patients who received both surgery and radiation within the same HCO, which likely excluded a significant number of patients, but these patients with fragmented care were able to be included in the analysis of the NCDB.It is possible that there could be some patient overlap between databases, however the second database was useful in that it allowed for the presentation of more granular data.A limitation affecting our analysis of both databases is that we were unable to determine whether patients were subject to circumstances that justifiably delayed adjuvant radiation, which may have inflated our reported rate of delay.Components that are recognized as leading to prolonged time to PORT, such as delays in dental evaluation, necessary treatment preradiation, and transportation issues, were also not able to be captured through these databases.
In conclusion, timely initiation of PORT continues to be a challenge for nearly two-thirds of patients with HNSCC.With the introduction of a head and neck specific CoC quality measure evaluating this benchmark, we anticipate renewed institutional focus on vulnerable patients and care processes associated with delay.A lack of family support, more complex surgical care, and demographic factors associated with poorer access to care continue to be risk factors for delayed adjuvant radiation.

FIGURE 2 .
FIGURE 2.Temporal trends in the failure to initiate PORT within 6 weeks of surgery according to year of diagnosis

TABLE 1
NCDB analysis of demographic characteristics

n = 40 164) Initiation of PORT ≤6 weeks (n = 15 263) Initiation of PORT >6 weeks (n = 24 901) OR [95% CI]
Head Neck.Author manuscript; available in PMC 2024 January 19.Abbreviations: CI, confidence interval; OR, odds ratio; PORT, postoperative radiation therapy; Ref, reference category.a Certain columns may not sum to the total due to missing observations.Head Neck.Author manuscript; available in PMC 2024 January 19.

TABLE 2
NCDB and TriNetX time to PORT breakdown

TABLE 3
NCDB analysis of clinical and surgical characteristics Head Neck.Author manuscript; available in PMC 2024 January 19.

TABLE 4
NCDB analysis of hospital characteristics

Patient variable Total patients (n = 40 164) Initiation of PORT ≤6 weeks (n = 15 263) Initiation of PORT >6 weeks (n = 24 901) OR [95% CI]
Abbreviations: CI, confidence interval; CoC, commission on cancer; OR, odds ratio; PORT, postoperative radiation therapy; Ref, reference category.a Certain columns may not sum to the total due to missing observations.Head Neck.Author manuscript; available in PMC 2024 January 19.

TABLE 5
Risk factors for prolonged length of stay

TABLE 6
Risk factors for hospital readmission

TABLE 7
TriNetX analysis of demographic characteristics

TABLE 8
TriNetX analysis of clinical and surgical characteristics
Head Neck.Author manuscript; available in PMC 2024 January 19.Head Neck.Author manuscript; available in PMC 2024 January 19.