Factors Associated with the Choice of Radiation Therapy Treatment Facility in Head and Neck Cancer

Abstract

Objective:

To analyze the clinicodemographic characteristics and treatment outcomes of patients receiving post-operative radiation therapy (PORT) at a different treatment facility rather than the initial surgical facility for head and neck cancer.

Methods:

Retrospective cohort analysis utilizing the National Cancer Data Base, 2004 – 2015, of patients with a diagnosis of oral cavity/oropharyngeal, hypopharyngeal, and laryngeal squamous cell carcinoma (SCC). Multivariate analysis was completed with multivariate regression and Cox proportional hazard model, and survival outcomes were examined using Kaplan-Meier analysis.

Results:

A total of 15,181 patients who had surgery for a head and neck cancer at an academic/research center were included in the study population. Of the study population, 4,890 (32.2%) patients completed PORT at a different treatment facility. Treatment at a different facility was more common among patients who were ≥65 years old, white, Medicare recipients, and those with a greater distance between residence and surgical treatment facility, and with lower income within area of residence (each p<0.05). Overall survival was worse in patients completing PORT at a different treatment facility versus at the institution where surgery was completed (61.9% vs. 66.4%; p=0.002).

Conclusions:

PORT at a different facility was more common in older individuals, Medicare recipients, those with greater distance to travel, and lower income individuals. Completing PORT outside the hospital where surgery was performed was associated with inferior survival outcomes among head and neck cancer patients.

Introduction

Head and neck cancer is the sixth most common form of cancer worldwide and is responsible for over 350,000 deaths yearly.^1,2 Radiation therapy, as a single-modality therapy or part of multimodality therapy, is an important component of head and neck cancer treatment. Radiation therapy in head and neck cancer is complex due to anatomical considerations and the function-altering nature of treatment.³ Additionally, the utilization of intensity-modulated radiation therapy (IMRT) necessitates more complex treatment planning compared to conventional radiation therapy techniques. These factors led to significant heterogeneity in treatment plans among even expert radiation oncologists.⁴ Due to the significant challenge of this patient population, the National Comprehensive Cancer Network (NCCN) recommends that all patients should have access to “specialists with expertise in the management of patients with head and neck cancer for optimal treatment”.⁵

Given the complexity of head and neck cancer treatment, survival outcomes might well be influenced by provider and treatment facility experience. A patient’s choice of radiation therapy treatment facility can be influenced by geographic location, transportation, housing availability, and social support.⁶ There is an increasing amount of literature examining the role of treatment facilities in head and neck cancer outcomes,^7–9 but data on patients who are not treated at the same hospital where surgery is performed is limited. Peters et al. found that noncompliant treatment plans and quality of radiation therapy treatment deviated the most in low volume centers.¹⁰ In the present national-based study, the objective was to analyze the clinicodemographic characteristics and outcomes of patients who had surgery at an academic/research hospital and had post-operative radiation therapy (PORT) at a different facility.

Methods

The study is a retrospective cohort analysis utilizing the National Cancer Data Base (NCDB), from 2004–2015. The database began in 1989 and is sponsored by the American Cancer Society and American College of Surgeons.¹¹ The NCDB comprises over 34 million de-identified records derived from 1,500 Commission on Cancer (CoC) accredited facilities, and more than 70% of yearly United States cancer diagnoses are captured.¹¹ This study did not meet criteria for human subject research.

The study population was defined as adults with squamous cell carcinoma (SCC) of the oral cavity, oropharynx, hypopharynx, and larynx treated with surgery at an academic/research center and who then completed adjuvant radiation therapy with or without chemotherapy. Academic/research centers were defined as those with National Cancer Institute comprehensive cancer center designation. We recorded whether radiation therapy was completed at the same hospital where surgery was performed or if the patients completed radiation therapy at a different facility. Patients were identified utilizing the International Classification of Diseases for Oncology, 3^rd edition (ICD-O-3: 807).

Patient demographics, clinical characteristics, treatments, and survival were analyzed. Factors that were examined for independent association with receipt of radiation treatment outside the hospital that performed the surgery were 1) age (<65, ≥65 years-old), 2) sex, 3) race, 4) Charlson Comorbidity Index¹², 5) cancer site (oral cavity/oropharynx, hypopharynx, larynx), 6) cancer stage, 7) type of insurance, 8) rurality of residence, 9) distance between residence and treating facility, median 10) income quartile of residence ZIP code¹³, 11) educational attainment of residence ZIP code¹³, and 12) hospital cancer volume of the surgical facility. Time between surgery and PORT was measured in days from surgery to the first day of radiation treatment.

Statistics

Chi square tests were used to assess associations between categorical variables. Factors with significant univariate associations were included in the multivariate logistic regression model of location of adjuvant therapy. Odds ratios (OR) and 95% confidence intervals (95% CI) were calculated from the multivariate regression model. Kaplan-Meier survival estimates were used to assess the association of radiation treatment location with overall survival (OS) outcomes. Cox proportional hazard model was used to calculate multivariate hazard ratios (HR) and 95% CI. All tests were considered statistically significant at p < 0.05. Statistical analysis was performed using SAS 9.4 (SAS Institute Inc., Cary, NC, USA).

Results

A total of 15,181 patients were included in the study population. Characteristics of the study population are listed in Table 1. 10,291 (67.8%) patients were treated with radiation therapy at the same hospital that performed surgery. The median of duration in days between surgery and radiation therapy was 47 days (interquartile range: 37–61 days) for patients treated at the same hospital where surgery was performed, while patients that were treated at a different hospital the median duration was 49 days (interquartile range: 39–64 days). The average radiation dose received at the treating hospital was (mean±standard error) 62.14±0.38 Gray (Gy), while average radiation dose received at the outside hospital was 60.13±0.53 Gy (p=0.005).

Table 1.

Descriptive statistics of the study population of patients with squamous cell carcinoma of the head and neck who had surgery of the primary site at an Academic/Research Program (includes NCI designated comprehensive cancer centers). The descriptive statistics are further classified based on whether the patient received recommended radiotherapy at the same facility or a different facility. National Cancer Database, 2004 – 2015.

	Study population, N=15,181 (%)a	Adjuvant Radiotherapy location (%)a		Pb
		Same hospital that performed surgery, n=10,291	Outside facility, n=4,890
Age (yr.)
<65	10718(70.6)	7411(72.01)	3307(67.63)
≥65	4463(29.4)	2880(27.99)	1583(32.37)	<0.001
Gender
Male	11316(74.54)	7716(74.98)	3600(73.62)
Female	3865(25.46)	2575(25.02)	1290(26.38)	0.07
Race
White	12961(85.38)	8557(83.15)	4404(90.06)
Black	1605(10.57)	1260(12.24)	345(7.06)
Other	615(4.05)	474(4.61)	141(2.88)	<0.001
Charlson Cormorbidity Index
0	11542(76.03)	7887(76.64)	3655(74.74)
1	2785(18.35)	1834(17.82)	951(19.45)
2	612(4.03)	404(3.93)	208(4.25)
≥3	242(1.59)	166(1.61)	76(1.55)	0.06
Site of cancer
Oral cavity/Oropharynx	11199(73.77)	7626(74.1)	3573(73.07)
Hypopharynx	740(4.87)	483(4.69)	257(5.26)
Larynx	3242(21.36)	2182(21.2)	1060(21.68)	0.23
Stage
I, II	2569(16.92)	1821(17.7)	748(15.3)
III, IV	12612(83.08)	8470(82.3)	4142(84.7)	<0.001
Chemotherapy
Not received	7345(48.38)	4812(46.76)	2533(51.8)
Received	7836(51.62)	5479(53.24)	2357(48.2)	<0.001
Type of insurance
No insurance	955(6.29)	653(6.35)	302(6.18)
Private	7574(49.89)	5276(51.27)	2298(46.99)
Medicaid	1951(12.85)	1359(13.21)	592(12.11)
Medicare	4701(30.97)	3003(29.18)	1698(34.72)	<0.001
Patient’s area of residence
Metro	12302(81.04)	8879(86.28)	3423(70)
Urban	2598(17.11)	1271(12.35)	1327(27.14)
Rural	281(1.85)	141(1.37)	140(2.86)	<0.001
Patient’s area of residence distance from treating facility (mile)
≤median: ≤20	7706(50.76)	6437(62.55)	1269(25.95)
>median - ≤90th percentile: >20 – ≤100	5855(38.57)	3169(30.79)	2686(54.93)
>90th percentile: >100	1620(10.67)	685(6.66)	935(19.12)	<0.001
Income quartile in patient’s area of residence
<1st quartile (low income area)	3140(20.68)	1988(19.32)	1152(23.56)
>1st quartile - <2nd quartile	3501(23.06)	2135(20.75)	1366(27.93)
>2nd quartile - <3rd quartile	3330(21.94)	2271(22.07)	1059(21.66)
>4th quartile (high income area)	5210(34.32)	3897(37.87)	1313(26.85)	<0.001
Quartile of no high school graduation in the patient’s area of residence
<1st quartile (low education area)	3380(22.26)	2274(22.1)	1106(22.62)
>1st quartile - <2nd quartile	3961(26.09)	2562(24.9)	1399(28.61)
>2nd quartile - <3rd quartile	4168(27.46)	2733(26.56)	1435(29.35)
>4th quartile (high education area)	3672(24.19)	2722(26.45)	950(19.43)	<0.001
Hospital volume (surgery/yr.)
Low: 1 – 10	6335(41.73)	4599(44.69)	1736(35.5)
Medium: 11 – 30	6013(39.61)	3848(37.39)	2165(44.27)
High: ≥ 31	2833(18.66)	1844(17.92)	989(20.22)	<0.001

^aPercentage values may not add up to 100% due to rounding.

^bChi-square test.

Table 2 lists the results of multivariate analysis of variables associated with receipt of radiation therapy at a non-academic hospital. Patients that had radiation therapy at a different hospital from the hospital that performed surgery tended to be aged 65 and older (OR 1.17, 95% CI 1.05–1.31; p=0.006), white (OR 1.0), Medicare recipients (OR 1.12, 95% CI 1.00–1.26; p=0.044), and lived a long distance (>90^th percentile) from the treating facility (OR 1.52, 95% CI 1.35–1.70; p<0.001). Patients who had chemotherapy as part of their planned adjuvant therapy were less likely to seek treatment at a different facility (OR 0.80, 95%CI 0.74–0.86, p<0.001). Advanced stage of disease was a predictor of having radiation therapy at a different facility (p=0.048).

Table 2.

Demographic and clinical characteristics of patients with head and neck squamous cell carcinoma who received adjuvant radiotherapy outside the hospital that performed the surgery. National Cancer Database, 2004 – 2015 (N=15,181)

	Radiotherapy at outside hospital (%)	aORa	95%CI	Pb
Age (yr.)
<65	30.85	Reference
≥65	35.47	1.17	1.05, 1.31	0.006
Race
White	33.98	Reference
Black	21.50	0.63	0.55, 0.73	<0.001
Other	22.93	0.84	0.68, 1.02	0.08
Stage
I, II	29.12	Reference
III, IV	32.84	1.11	1.00, 1.24	0.048
Chemotherapy
Not received	34.49	Reference
Received	30.08	0.80	0.74, 0.86	<0.001
Type of insurance
No insurance	31.62	1.05	0.89, 1.23	0.57
Private	30.34	Reference
Medicaid	30.34	1.05	0.93, 1.18	0.46
Medicare	36.12	1.12	1.00, 1.26	0.044
Patient’s area of residence
Metro	27.82	Reference
Urban	51.08	1.04	0.94, 1.15	0.48
Rural	49.82	0.87	0.67, 1.11	0.26
Patient’s area of residence distance from treating facility (mile)
≤median: ≤20	16.47	0.25	0.23, 0.28	<0.001
>median - ≤90th percentile: >20 – ≤100	45.88	Reference
>90th percentile: >100	57.72	1.52	1.35, 1.70	<0.001
Income quartile in patient’s area of residence
<1st quartile (low income area)	36.69	Reference
>1st quartile - <2nd quartile	39.02	0.86	0.77, 0.97	0.011
>2nd quartile - <3rd quartile	31.80	0.71	0.62, 0.81	<0.001
>4th quartile (high income area)	25.20	0.61	0.53, 0.71	<0.001
Quartile of no high school graduation in the patient’s area of residence
<1st quartile (low education area)	32.72	Reference
>1st quartile - <2nd quartile	35.32	1.05	0.94, 1.17	0.40
>2nd quartile - <3rd quartile	34.43	1.18	1.05, 1.34	0.008
>4th quartile (high education area)	25.87	1.05	0.90, 1.21	0.54
Hospital volume (surgery/yr.)
Low: 1 – 10	27.40	Reference
Medium: 11 – 30	36.01	1.05	0.96, 1.14	0.28
High: ≥ 31	34.91	0.93	0.84, 1.04	0.21

Abbreviations: aOR, adjusted odds ratio; CI, confidence interval.

^aThe model includes all the factors listed in the table.

^bMultivariate logistic regression model.

Treatment facility utilization trends over time are shown in Figure 1. Although the trend change was statistically significant over time, the change was not clinically meaningful. Figure 2 demonstrates that patients treated at a different facility had poorer OS compared to patients treated with radiation therapy at the hospital where surgery was performed (5 year OS 61.9% vs. 66.4%, HR 1.10, 95% CI 1.04–1.16; p=0.002). The multivariate model to assess survival analysis controlled for age, gender, race, Charlson Cormorbidity Index, site of cancer, stage, insurance status, area of residence, income, education, and hospital volume.

Figure 1:

Time trend illustrating where patients received radiation therapy following surgery for head and neck cancer

Figure 2:

Kaplan-Meier survival estimates stratified by where patients received adjuvant radiation therapy following surgery for head and neck cancer

Discussion

Choice of radiation therapy treatment facility depends on referral networks, personal relationships, recommendations, geographic location, transportation, housing availability, social support, and many other factors.⁶ In this nationally-representative study, we examined the role of treatment facilities for adjuvant radiation therapy after surgery for SCC of the head and neck. Patients who were treated a facility other than where they had surgery included older age, white race, Medicare insurance, greater distance between residence and surgical treatment facility, and lower quartile of income within an area of residence. We found that OS is worse (HR 1.10, 95% CI 1.04–1.16; p=0.002) when PORT is administered at a different facility.

Timeliness of PORT is an important aspect in consideration of treatment facility. Delay of timely treatment can lead to microscopic tumor repopulation and progressive decrease in local control of disease.¹³ The NCCN recommends initiation of PORT within 6 weeks of surgery.⁵ An analysis of 41,291 patients in the NCDB demonstrated that there was a progressive decrease in survival when PORT was delayed beyond 7 weeks.¹³ In our study, time to adjuvant radiation therapy initiation was similar among both cohorts (median 47 and 49 days between same treatment facility versus different facility, respectively) and likely not associated with the decreased survival seen in patients treated at a different facility.

We did not find a meaningful difference between facility selection with respect to hospital volume. These results suggest it is not hospital volume that effects treatment outcomes, but improved care coordination may play a significant role if PORT is delivered at the same facility where surgery is performed. When PORT is delivered at a different facility, there is an inherent risk of fragmented care and less timely coordination due to additional communication, referral pathways, and navigational issues. As healthcare moves to person-centered care, coordination and multidisplinary teams are an increasingly integral component in the treatment of head and neck cancer. Not only does treatment require radiation oncologists to have a understanding of complex patterns of disease spread and expertise in contouring head and neck targets, but coordination may play an equally important role that is not easily quantifiable. This includes a team of support staff (i.e. physicists, dieticians, speech pathologist, social workers, occupational therapist, physical therapist) that must become quickly immersed in a patient’s disease status.^14–16 To date, there has been little information on treatment facility selection for PORT in SCC of the head and neck. The present study demonstrates that OS is worse when performed at a different facility. A recently-published analysis of NCDB patients with aerodigestive tract, salivary gland, or head and neck melanoma showed improved overall survival when treatment was completed at an academic medical center compared to community cancer programs.¹⁷ That report did not address the possibility that multidisciplinary treatment took place at more than once facility. Our results of treatment facility selection are similar to an analysis by Lee et al. of the NCDB which was published during prepration of our manuscript, who found that treatment facility for PORT was an independent predictor of OS.¹⁴ However, there are important differences. First, their study population consisted of only stage III-IV patients (n=6844), and our larger and more comprehensive cohort included all stages (n=15,181). Secondly, our findings uniquely demonstrated that certain demographic and socioeconomic factors, such as educational status and distance to travel to treatment facility, may play important roles in PORT facility selction that were not addressed by Lee et al. With multivariate analysis, we showed that patients who had a greater distance between residence and surgical treatment facility were more likely to receive PORT at a different treatment facility. Third, we have shown that among types of insurance status (no insurance, private, Medicaid, Medicare), Medicare patients were most likely to complete PORT at a different treatment facility. As the government strives to improve insurance coverage for patients, reporting health outcomes as it relates to insurance status is critical to improve coverage and care. Finally, we have shown that the treatment trend of PORT following surgery for head and neck cancer was relatively stable over the time period of our study. Our findings contrast with those of Phillips et al., who found in a retrospective analysis of 336 patients who had surgery at one center that there was no difference in OS among patients whose adjuvant therapy was given at high-volume versus low-volume centers.¹⁸ It is impossible to analyze based on registry data the myriad qualitative factors that differ between radiation oncology practices to understand the difference in reported outcomes. Future prospective trials are needed to further address this important question.

Over recent years, there has been an increasing focus on the relationship between socioeconomic status and head and neck cancer.^19–21,22We did not have patient-level socioeconomic data, so we used an ecological approximation based on residence zip code. We found lower area income was significantly associated with receiving treatment at a different facility. Additionally, the distance between a patient’s residence and initial treatment facility was significantly (OR 1.52, 95% CI 1.35–1.70; p<0.001) associated with receiving treatment at a different facility. Patients required to travel great distances for cancer treatment have reported fatigue from traveling, potentially worsened over time by treatment itself.²³ Travel distance and absence can be more difficult for a radiation therapy episode than for surgical care due to the 6 to 7 elapsed weeks.²⁴ If these patients are socioeconomically depressed and traveling from a great distance, they are less likely to have the financial means or social support to facilitate living away from home to receive treatment at an academic medical center. Travel accommodations and financial burdens are clearly pertinent issues that will need to be further addressed in the future as healthcare may head to a more centralized process.

There are inherent limitations to this database analysis. The NCDB classification system can lead to individual patient and data coding errors. The patients in the present study are only from CoC-accredited hospitals, which can lead to selection bias.¹¹ Causality cannot be inferred from this database study, and there are likely to be unmeasured confounding factors including socioeconomic status and comorbidities, which could directly affect survival. Interaction effects were not explored because the NCDB does not include the information that is sufficiently detailed to address interaction between the outcome and patient resources. The NCDB does not specify the type of radiation delivered, and the increasing use of the more complex IMRT may influence outcomes. Additionally, the study’s findings should not guide radiation therapy facility choice for non-surgically treated patients as this group was excluded from the study population. Even with these limitations, we present several important findings in treatment facility utilization of patients with SCC of the head and neck.

In conclusion, the present study provides a national perspective on the effect of treatment facility for PORT following surgery for SCC of the head and neck. Patients who received PORT at a different facility tended to have worse OS. Completing PORT at a different facility was more common among patients ≥65 years-old, white race, and Medicare insured patients. Patients with longer travel distances and lower income patients are more likely to be treated at a different facility, which demonstrates the additional disparities seen in patients with head and neck cancer.